; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mcpu=skx %s -o - | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl %s -o - | FileCheck %s
define <4 x double> @test_double_to_4(double %s) {
; CHECK-LABEL: test_double_to_4:
; CHECK-LABEL: test_masked_double_to_4_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %ymm1 {%k1}
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_double_to_4_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x double> undef, double %s, i32 0
; CHECK-LABEL: test_masked_double_to_4_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %ymm1 {%k1}
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_double_to_4_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x double> undef, double %s, i32 0
; CHECK-LABEL: test_masked_double_to_4_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %ymm1 {%k1}
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_double_to_4_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x double> undef, double %s, i32 0
; CHECK-LABEL: test_masked_double_to_4_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %ymm1 {%k1}
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_double_to_4_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x double> undef, double %s, i32 0
; CHECK-LABEL: test_masked_double_to_8_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_double_to_8_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x double> undef, double %s, i32 0
; CHECK-LABEL: test_masked_double_to_8_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $103, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_double_to_8_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $103, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x double> undef, double %s, i32 0
; CHECK-LABEL: test_masked_double_to_8_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-56, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_double_to_8_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-56, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x double> undef, double %s, i32 0
; CHECK-LABEL: test_masked_double_to_8_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $78, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_double_to_8_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $78, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x double> undef, double %s, i32 0
; CHECK-LABEL: test_masked_float_to_4_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %xmm1 {%k1}
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_float_to_4_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %xmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x float> undef, float %s, i32 0
; CHECK-LABEL: test_masked_float_to_4_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %xmm1 {%k1}
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_float_to_4_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %xmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x float> undef, float %s, i32 0
; CHECK-LABEL: test_masked_float_to_4_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %xmm1 {%k1}
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_float_to_4_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %xmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x float> undef, float %s, i32 0
; CHECK-LABEL: test_masked_float_to_4_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %xmm1 {%k1}
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_float_to_4_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %xmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x float> undef, float %s, i32 0
; CHECK-LABEL: test_masked_float_to_8_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $72, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %ymm1 {%k1}
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_float_to_8_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $72, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x float> undef, float %s, i32 0
; CHECK-LABEL: test_masked_float_to_8_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-64, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %ymm1 {%k1}
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_float_to_8_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-64, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x float> undef, float %s, i32 0
; CHECK-LABEL: test_masked_float_to_8_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %ymm1 {%k1}
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_float_to_8_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x float> undef, float %s, i32 0
; CHECK-LABEL: test_masked_float_to_8_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $64, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %ymm1 {%k1}
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_float_to_8_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $64, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x float> undef, float %s, i32 0
; CHECK-LABEL: test_masked_float_to_16_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-5916, %ax # imm = 0xE8E4
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_float_to_16_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-5916, %ax # imm = 0xE8E4
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x float> undef, float %s, i32 0
; CHECK-LABEL: test_masked_float_to_16_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-1130, %ax # imm = 0xFB96
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_float_to_16_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-1130, %ax # imm = 0xFB96
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x float> undef, float %s, i32 0
; CHECK-LABEL: test_masked_float_to_16_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-12439, %ax # imm = 0xCF69
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_float_to_16_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-12439, %ax # imm = 0xCF69
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x float> undef, float %s, i32 0
; CHECK-LABEL: test_masked_float_to_16_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-6413, %ax # imm = 0xE6F3
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_float_to_16_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-6413, %ax # imm = 0xE6F3
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = insertelement <2 x float> undef, float %s, i32 0
; CHECK-LABEL: test_masked_double_to_4_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %ymm0 {%k1}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_z_double_to_4_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_double_to_4_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %ymm0 {%k1}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_z_double_to_4_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_double_to_4_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %ymm0 {%k1}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_z_double_to_4_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_double_to_4_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %ymm0 {%k1}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_z_double_to_4_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_double_to_8_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $120, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_z_double_to_8_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $120, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_double_to_8_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $26, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_z_double_to_8_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $26, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_double_to_8_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $111, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_z_double_to_8_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $111, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_double_to_8_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-100, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_z_double_to_8_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-100, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load double, double* %p
; CHECK-LABEL: test_masked_float_to_4_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %xmm0 {%k1}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_z_float_to_4_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %xmm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_float_to_4_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %xmm0 {%k1}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_z_float_to_4_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %xmm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_float_to_4_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %xmm0 {%k1}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_z_float_to_4_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %xmm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_float_to_4_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %xmm0 {%k1}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_z_float_to_4_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %xmm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_float_to_8_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $67, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %ymm0 {%k1}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_z_float_to_8_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $67, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_float_to_8_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-51, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %ymm0 {%k1}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_z_float_to_8_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-51, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_float_to_8_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-116, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %ymm0 {%k1}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_z_float_to_8_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-116, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_float_to_8_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %ymm0 {%k1}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_z_float_to_8_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_float_to_16_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-18370, %ax # imm = 0xB83E
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_z_float_to_16_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-18370, %ax # imm = 0xB83E
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_float_to_16_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $26137, %ax # imm = 0x6619
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_z_float_to_16_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $26137, %ax # imm = 0x6619
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_float_to_16_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-11480, %ax # imm = 0xD328
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_z_float_to_16_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-11480, %ax # imm = 0xD328
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_float_to_16_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-21749, %ax # imm = 0xAB0B
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
%s = load float, float* %p
; CHECK-LABEL: test_masked_z_float_to_16_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-21749, %ax # imm = 0xAB0B
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%s = load float, float* %p
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mcpu=skx %s -o - | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl,+avx512bw %s -o - | FileCheck %s
define <16 x i8> @test_i8_to_16(i8 %s) {
; CHECK-LABEL: test_i8_to_16:
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mcpu=skx %s -o - | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl,+avx512dq %s -o - | FileCheck %s
define <8 x float> @test_2xfloat_to_8xfloat(<8 x float> %vec) {
; CHECK-LABEL: test_2xfloat_to_8xfloat:
; CHECK-LABEL: test_masked_2xfloat_to_8xfloat_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm1 {%k1} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xfloat_to_8xfloat_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xfloat_to_8xfloat_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm1 {%k1} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xfloat_to_8xfloat_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xfloat_to_8xfloat_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-35, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm1 {%k1} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xfloat_to_8xfloat_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-35, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xfloat_to_8xfloat_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $62, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm1 {%k1} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xfloat_to_8xfloat_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $62, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xfloat_to_16xfloat_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $21312, %ax # imm = 0x5340
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm1 {%k1} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xfloat_to_16xfloat_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $21312, %ax # imm = 0x5340
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xfloat_to_16xfloat_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-8490, %ax # imm = 0xDED6
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm1 {%k1} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xfloat_to_16xfloat_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-8490, %ax # imm = 0xDED6
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xfloat_to_16xfloat_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $12522, %ax # imm = 0x30EA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm1 {%k1} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xfloat_to_16xfloat_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $12522, %ax # imm = 0x30EA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xfloat_to_16xfloat_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-28344, %ax # imm = 0x9148
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm1 {%k1} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xfloat_to_16xfloat_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-28344, %ax # imm = 0x9148
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xdouble_to_4xdouble_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} ymm0 {%k1} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_z_2xdouble_to_4xdouble_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_2xdouble_to_4xdouble_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} ymm0 {%k1} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_z_2xdouble_to_4xdouble_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_2xdouble_to_4xdouble_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} ymm0 {%k1} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_z_2xdouble_to_4xdouble_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_2xdouble_to_4xdouble_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} ymm0 {%k1} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_z_2xdouble_to_4xdouble_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_2xdouble_to_8xdouble_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $21, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} zmm0 {%k1} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_z_2xdouble_to_8xdouble_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $21, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_2xdouble_to_8xdouble_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $82, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} zmm0 {%k1} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_z_2xdouble_to_8xdouble_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $82, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_2xdouble_to_8xdouble_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} zmm0 {%k1} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_z_2xdouble_to_8xdouble_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_2xdouble_to_8xdouble_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-19, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} zmm0 {%k1} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_z_2xdouble_to_8xdouble_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-19, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x2 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_4xdouble_to_8xdouble_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $28, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_z_4xdouble_to_8xdouble_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $28, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_4xdouble_to_8xdouble_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-115, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_z_4xdouble_to_8xdouble_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-115, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_4xdouble_to_8xdouble_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-76, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_z_4xdouble_to_8xdouble_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-76, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_4xdouble_to_8xdouble_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-116, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_z_4xdouble_to_8xdouble_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-116, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf64x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; CHECK-NEXT: movb $-49, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm0 {%k1} = xmm1[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; CHECK-NEXT: movb $-49, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; CHECK-NEXT: movb $-118, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm0 {%k1} = xmm1[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; CHECK-NEXT: movb $-118, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; CHECK-NEXT: movb $-11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm0 {%k1} = xmm1[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; CHECK-NEXT: movb $-11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; CHECK-NEXT: movb $-102, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm0 {%k1} = xmm1[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; CHECK-NEXT: movb $-102, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; CHECK-NEXT: movw $-27027, %ax # imm = 0x966D
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm0 {%k1} = xmm1[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; CHECK-NEXT: movw $-27027, %ax # imm = 0x966D
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; CHECK-NEXT: movw $29162, %ax # imm = 0x71EA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm0 {%k1} = xmm1[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; CHECK-NEXT: movw $29162, %ax # imm = 0x71EA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; CHECK-NEXT: movw $-26458, %ax # imm = 0x98A6
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm0 {%k1} = xmm1[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; CHECK-NEXT: movw $-26458, %ax # imm = 0x98A6
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; CHECK-NEXT: movw $25225, %ax # imm = 0x6289
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm0 {%k1} = xmm1[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; CHECK-NEXT: movw $25225, %ax # imm = 0x6289
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x2 {{.*#+}} zmm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x float>, <2 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_to_8xfloat_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-109, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} ymm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_to_8xfloat_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-109, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_to_8xfloat_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $74, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} ymm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_to_8xfloat_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $74, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_to_8xfloat_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $49, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} ymm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_to_8xfloat_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $49, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_to_8xfloat_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $48, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} ymm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_to_8xfloat_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $48, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_to_16xfloat_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-25378, %ax # imm = 0x9CDE
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_to_16xfloat_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-25378, %ax # imm = 0x9CDE
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_to_16xfloat_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-22502, %ax # imm = 0xA81A
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_to_16xfloat_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-22502, %ax # imm = 0xA81A
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_to_16xfloat_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $31229, %ax # imm = 0x79FD
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_to_16xfloat_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $31229, %ax # imm = 0x79FD
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_to_16xfloat_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $5887, %ax # imm = 0x16FF
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_to_16xfloat_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $5887, %ax # imm = 0x16FF
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_to_16xfloat_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-15887, %ax # imm = 0xC1F1
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x8 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_to_16xfloat_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-15887, %ax # imm = 0xC1F1
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x8 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_to_16xfloat_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-8077, %ax # imm = 0xE073
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x8 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_to_16xfloat_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-8077, %ax # imm = 0xE073
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x8 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_to_16xfloat_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-5023, %ax # imm = 0xEC61
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x8 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_to_16xfloat_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-5023, %ax # imm = 0xEC61
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x8 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_to_16xfloat_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-10326, %ax # imm = 0xD7AA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x8 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_to_16xfloat_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-10326, %ax # imm = 0xD7AA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcastf32x8 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mcpu=skx %s -o - | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl,+avx512dq %s -o - | FileCheck %s
; FIXME: fixing PR34394 should fix the i32x2 memory cases resulting in a simple vbroadcasti32x2 instruction.
; CHECK-LABEL: test_masked_2xi32_to_4xi32_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 %xmm0, %xmm1 {%k1}
; CHECK-NEXT: vmovdqa %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xi32_to_4xi32_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 %xmm0, %xmm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x i32> %vec, <4 x i32> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xi32_to_4xi32_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 %xmm0, %xmm1 {%k1}
; CHECK-NEXT: vmovdqa %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xi32_to_4xi32_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 %xmm0, %xmm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x i32> %vec, <4 x i32> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xi32_to_4xi32_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 %xmm0, %xmm1 {%k1}
; CHECK-NEXT: vmovdqa %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xi32_to_4xi32_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 %xmm0, %xmm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x i32> %vec, <4 x i32> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xi32_to_4xi32_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 %xmm0, %xmm1 {%k1}
; CHECK-NEXT: vmovdqa %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xi32_to_4xi32_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 %xmm0, %xmm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x i32> %vec, <4 x i32> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xi32_to_8xi32_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $92, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm1 {%k1} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovdqa %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xi32_to_8xi32_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $92, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i32> %vec, <8 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xi32_to_8xi32_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-15, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm1 {%k1} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovdqa %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xi32_to_8xi32_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-15, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i32> %vec, <8 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xi32_to_8xi32_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-95, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm1 {%k1} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovdqa %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xi32_to_8xi32_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-95, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i32> %vec, <8 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xi32_to_8xi32_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm1 {%k1} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovdqa %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xi32_to_8xi32_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i32> %vec, <8 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xi32_to_16xi32_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-18638, %ax # imm = 0xB732
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} zmm1 {%k1} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovdqa64 %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xi32_to_16xi32_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-18638, %ax # imm = 0xB732
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} zmm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x i32> %vec, <16 x i32> undef, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xi32_to_16xi32_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $25429, %ax # imm = 0x6355
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} zmm1 {%k1} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovdqa64 %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xi32_to_16xi32_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $25429, %ax # imm = 0x6355
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} zmm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x i32> %vec, <16 x i32> undef, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xi32_to_16xi32_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $27159, %ax # imm = 0x6A17
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} zmm1 {%k1} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovdqa64 %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xi32_to_16xi32_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $27159, %ax # imm = 0x6A17
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} zmm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x i32> %vec, <16 x i32> undef, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK-LABEL: test_masked_2xi32_to_16xi32_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-22884, %ax # imm = 0xA69C
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} zmm1 {%k1} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: vmovdqa64 %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xi32_to_16xi32_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-22884, %ax # imm = 0xA69C
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} zmm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x i32> %vec, <16 x i32> undef, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
; CHECK: # BB#0:
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 {%k1} = xmm1[0,2,0,2]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = mem[0],zero,mem[1],zero
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 {%k1} {z} = xmm0[0,2,0,2]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 {%k1} = xmm1[0,2,0,2]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = mem[0],zero,mem[1],zero
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 {%k1} {z} = xmm0[0,2,0,2]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 {%k1} = xmm1[0,2,0,2]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = mem[0],zero,mem[1],zero
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 {%k1} {z} = xmm0[0,2,0,2]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 {%k1} = xmm1[0,2,0,2]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = mem[0],zero,mem[1],zero
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 {%k1} {z} = xmm0[0,2,0,2]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; CHECK-NEXT: movb $-94, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm0 {%k1} = xmm1[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; CHECK-NEXT: movb $-94, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; CHECK-NEXT: movb $97, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm0 {%k1} = xmm1[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; CHECK-NEXT: movb $97, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; CHECK-NEXT: movb $-33, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm0 {%k1} = xmm1[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; CHECK-NEXT: movb $-33, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; CHECK-NEXT: movb $-111, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm0 {%k1} = xmm1[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; CHECK-NEXT: movb $-111, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x2 {{.*#+}} ymm0 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm2 = [0,2,0,2,0,2,0,2,0,2,0,2,0,2,0,2]
; CHECK-NEXT: movw $27158, %ax # imm = 0x6A16
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermd %zmm1, %zmm2, %zmm0 {%k1}
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm1 = [0,2,0,2,0,2,0,2,0,2,0,2,0,2,0,2]
; CHECK-NEXT: movw $27158, %ax # imm = 0x6A16
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermd %zmm0, %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm2 = [0,2,0,2,0,2,0,2,0,2,0,2,0,2,0,2]
; CHECK-NEXT: movw $26363, %ax # imm = 0x66FB
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermd %zmm1, %zmm2, %zmm0 {%k1}
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm1 = [0,2,0,2,0,2,0,2,0,2,0,2,0,2,0,2]
; CHECK-NEXT: movw $26363, %ax # imm = 0x66FB
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermd %zmm0, %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm2 = [0,2,0,2,0,2,0,2,0,2,0,2,0,2,0,2]
; CHECK-NEXT: movw $-19542, %ax # imm = 0xB3AA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermd %zmm1, %zmm2, %zmm0 {%k1}
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm1 = [0,2,0,2,0,2,0,2,0,2,0,2,0,2,0,2]
; CHECK-NEXT: movw $-19542, %ax # imm = 0xB3AA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermd %zmm0, %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm1 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm2 = [0,2,0,2,0,2,0,2,0,2,0,2,0,2,0,2]
; CHECK-NEXT: movw $27409, %ax # imm = 0x6B11
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermd %zmm1, %zmm2, %zmm0 {%k1}
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = mem[0],zero,mem[1],zero
; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm1 = [0,2,0,2,0,2,0,2,0,2,0,2,0,2,0,2]
; CHECK-NEXT: movw $27409, %ax # imm = 0x6B11
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermd %zmm0, %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = load <2 x i32>, <2 x i32>* %vp
; CHECK-LABEL: test_masked_4xi32_to_8xi32_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-87, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} ymm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_z_4xi32_to_8xi32_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-87, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_4xi32_to_8xi32_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} ymm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_z_4xi32_to_8xi32_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_4xi32_to_8xi32_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $114, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} ymm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_z_4xi32_to_8xi32_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $114, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_4xi32_to_8xi32_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $66, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} ymm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_z_4xi32_to_8xi32_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $66, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_4xi32_to_16xi32_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $10334, %ax # imm = 0x285E
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_z_4xi32_to_16xi32_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $10334, %ax # imm = 0x285E
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_4xi32_to_16xi32_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-30962, %ax # imm = 0x870E
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_z_4xi32_to_16xi32_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-30962, %ax # imm = 0x870E
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_4xi32_to_16xi32_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $31933, %ax # imm = 0x7CBD
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_z_4xi32_to_16xi32_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $31933, %ax # imm = 0x7CBD
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_4xi32_to_16xi32_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-28744, %ax # imm = 0x8FB8
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_z_4xi32_to_16xi32_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-28744, %ax # imm = 0x8FB8
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i32>, <4 x i32>* %vp
; CHECK-LABEL: test_masked_2xi64_to_4xi64_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} ymm0 {%k1} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_z_2xi64_to_4xi64_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_2xi64_to_4xi64_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} ymm0 {%k1} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_z_2xi64_to_4xi64_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_2xi64_to_4xi64_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} ymm0 {%k1} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_z_2xi64_to_4xi64_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_2xi64_to_4xi64_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} ymm0 {%k1} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_z_2xi64_to_4xi64_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} ymm0 {%k1} {z} = mem[0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_2xi64_to_8xi64_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $119, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} zmm0 {%k1} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_z_2xi64_to_8xi64_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $119, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_2xi64_to_8xi64_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-50, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} zmm0 {%k1} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_z_2xi64_to_8xi64_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-50, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_2xi64_to_8xi64_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-33, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} zmm0 {%k1} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_z_2xi64_to_8xi64_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-33, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_2xi64_to_8xi64_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-49, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} zmm0 {%k1} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_z_2xi64_to_8xi64_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-49, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x2 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,0,1,0,1,0,1]
; CHECK-NEXT: retq
%vec = load <2 x i64>, <2 x i64>* %vp
; CHECK-LABEL: test_masked_8xi32_to_16xi32_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $12321, %ax # imm = 0x3021
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x8 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x i32>, <8 x i32>* %vp
; CHECK-LABEL: test_masked_z_8xi32_to_16xi32_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $12321, %ax # imm = 0x3021
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x8 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x i32>, <8 x i32>* %vp
; CHECK-LABEL: test_masked_8xi32_to_16xi32_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-39, %ax
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x8 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x i32>, <8 x i32>* %vp
; CHECK-LABEL: test_masked_z_8xi32_to_16xi32_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-39, %ax
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x8 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x i32>, <8 x i32>* %vp
; CHECK-LABEL: test_masked_8xi32_to_16xi32_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-24047, %ax # imm = 0xA211
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x8 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x i32>, <8 x i32>* %vp
; CHECK-LABEL: test_masked_z_8xi32_to_16xi32_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-24047, %ax # imm = 0xA211
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x8 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x i32>, <8 x i32>* %vp
; CHECK-LABEL: test_masked_8xi32_to_16xi32_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $5470, %ax # imm = 0x155E
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x8 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x i32>, <8 x i32>* %vp
; CHECK-LABEL: test_masked_z_8xi32_to_16xi32_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $5470, %ax # imm = 0x155E
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti32x8 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x i32>, <8 x i32>* %vp
; CHECK-LABEL: test_masked_4xi64_to_8xi64_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-71, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i64>, <4 x i64>* %vp
; CHECK-LABEL: test_masked_z_4xi64_to_8xi64_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-71, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i64>, <4 x i64>* %vp
; CHECK-LABEL: test_masked_4xi64_to_8xi64_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i64>, <4 x i64>* %vp
; CHECK-LABEL: test_masked_z_4xi64_to_8xi64_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i64>, <4 x i64>* %vp
; CHECK-LABEL: test_masked_4xi64_to_8xi64_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $103, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i64>, <4 x i64>* %vp
; CHECK-LABEL: test_masked_z_4xi64_to_8xi64_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $103, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i64>, <4 x i64>* %vp
; CHECK-LABEL: test_masked_4xi64_to_8xi64_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-83, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x4 {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i64>, <4 x i64>* %vp
; CHECK-LABEL: test_masked_z_4xi64_to_8xi64_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-83, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vbroadcasti64x4 {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%vec = load <4 x i64>, <4 x i64>* %vp
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mcpu=skx %s -o - | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl %s -o - | FileCheck %s
define <4 x float> @test_4xfloat_dup_high(<4 x float> %vec) {
; CHECK-LABEL: test_4xfloat_dup_high:
; CHECK-LABEL: test_masked_4xfloat_dup_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm1 {%k1} = xmm0[1,1,3,3]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_dup_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} {z} = xmm0[1,1,3,3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
; CHECK-LABEL: test_masked_4xfloat_dup_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm1 {%k1} = xmm0[1,1,3,3]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_dup_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} {z} = xmm0[1,1,3,3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
; CHECK-LABEL: test_masked_4xfloat_dup_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm1 {%k1} = xmm0[1,1,3,3]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_dup_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} {z} = xmm0[1,1,3,3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
; CHECK-LABEL: test_masked_4xfloat_dup_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm1 {%k1} = xmm0[1,1,3,3]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_dup_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} {z} = xmm0[1,1,3,3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
; CHECK-LABEL: test_masked_4xfloat_dup_high_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm1 {%k1} = xmm0[1,1,3,3]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_dup_high_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} {z} = xmm0[1,1,3,3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
; CHECK-LABEL: test_masked_4xfloat_dup_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} = mem[1,1,3,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_dup_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} {z} = mem[1,1,3,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_dup_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} = mem[1,1,3,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_dup_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} {z} = mem[1,1,3,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_dup_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} = mem[1,1,3,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_dup_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} {z} = mem[1,1,3,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_dup_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} = mem[1,1,3,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_dup_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} {z} = mem[1,1,3,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_dup_high_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} = mem[1,1,3,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_dup_high_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} xmm0 {%k1} {z} = mem[1,1,3,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_dup_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-106, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm1 {%k1} = ymm0[1,1,3,3,5,5,7,7]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_dup_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-106, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} {z} = ymm0[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 1, i32 1, i32 3, i32 3, i32 5, i32 5, i32 7, i32 7>
; CHECK-LABEL: test_masked_8xfloat_dup_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $114, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm1 {%k1} = ymm0[1,1,3,3,5,5,7,7]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_dup_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $114, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} {z} = ymm0[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 1, i32 1, i32 3, i32 3, i32 5, i32 5, i32 7, i32 7>
; CHECK-LABEL: test_masked_8xfloat_dup_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-104, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm1 {%k1} = ymm0[1,1,3,3,5,5,7,7]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_dup_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-104, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} {z} = ymm0[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 1, i32 1, i32 3, i32 3, i32 5, i32 5, i32 7, i32 7>
; CHECK-LABEL: test_masked_8xfloat_dup_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm1 {%k1} = ymm0[1,1,3,3,5,5,7,7]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_dup_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} {z} = ymm0[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 1, i32 1, i32 3, i32 3, i32 5, i32 5, i32 7, i32 7>
; CHECK-LABEL: test_masked_8xfloat_dup_high_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-109, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm1 {%k1} = ymm0[1,1,3,3,5,5,7,7]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_dup_high_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-109, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} {z} = ymm0[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 1, i32 1, i32 3, i32 3, i32 5, i32 5, i32 7, i32 7>
; CHECK-LABEL: test_masked_8xfloat_dup_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $74, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} = mem[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_dup_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $74, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_dup_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $49, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} = mem[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_dup_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $49, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_dup_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $48, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} = mem[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_dup_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $48, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_dup_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-100, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} = mem[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_dup_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-100, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_dup_high_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-89, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} = mem[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_dup_high_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-89, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} ymm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_dup_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $31229, %ax # imm = 0x79FD
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm1 {%k1} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $31229, %ax # imm = 0x79FD
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 1, i32 1, i32 3, i32 3, i32 5, i32 5, i32 7, i32 7, i32 9, i32 9, i32 11, i32 11, i32 13, i32 13, i32 15, i32 15>
; CHECK-LABEL: test_masked_16xfloat_dup_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $5887, %ax # imm = 0x16FF
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm1 {%k1} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $5887, %ax # imm = 0x16FF
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 1, i32 1, i32 3, i32 3, i32 5, i32 5, i32 7, i32 7, i32 9, i32 9, i32 11, i32 11, i32 13, i32 13, i32 15, i32 15>
; CHECK-LABEL: test_masked_16xfloat_dup_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-15887, %ax # imm = 0xC1F1
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm1 {%k1} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-15887, %ax # imm = 0xC1F1
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 1, i32 1, i32 3, i32 3, i32 5, i32 5, i32 7, i32 7, i32 9, i32 9, i32 11, i32 11, i32 13, i32 13, i32 15, i32 15>
; CHECK-LABEL: test_masked_16xfloat_dup_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-8077, %ax # imm = 0xE073
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm1 {%k1} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-8077, %ax # imm = 0xE073
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 1, i32 1, i32 3, i32 3, i32 5, i32 5, i32 7, i32 7, i32 9, i32 9, i32 11, i32 11, i32 13, i32 13, i32 15, i32 15>
; CHECK-LABEL: test_masked_16xfloat_dup_high_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-5023, %ax # imm = 0xEC61
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm1 {%k1} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-5023, %ax # imm = 0xEC61
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 1, i32 1, i32 3, i32 3, i32 5, i32 5, i32 7, i32 7, i32 9, i32 9, i32 11, i32 11, i32 13, i32 13, i32 15, i32 15>
; CHECK-LABEL: test_masked_16xfloat_dup_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-10326, %ax # imm = 0xD7AA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-10326, %ax # imm = 0xD7AA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_dup_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-6675, %ax # imm = 0xE5ED
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-6675, %ax # imm = 0xE5ED
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_dup_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-5042, %ax # imm = 0xEC4E
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-5042, %ax # imm = 0xEC4E
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_dup_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-30108, %ax # imm = 0x8A64
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-30108, %ax # imm = 0x8A64
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_dup_high_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movw $25644, %ax # imm = 0x642C
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movw $25644, %ax # imm = 0x642C
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mcpu=skx %s -o - | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl %s -o - | FileCheck %s
define <2 x double> @test_2xdouble_dup_low(<2 x double> %vec) {
; CHECK-LABEL: test_2xdouble_dup_low:
; CHECK-LABEL: test_masked_2xdouble_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} xmm1 {%k1} = xmm0[0,0]
; CHECK-NEXT: vmovapd %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xdouble_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} xmm0 {%k1} {z} = xmm0[0,0]
; CHECK-NEXT: retq
%shuf = shufflevector <2 x double> %vec, <2 x double> undef, <2 x i32> <i32 0, i32 0>
; CHECK-LABEL: test_masked_2xdouble_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} xmm1 {%k1} = xmm0[0,0]
; CHECK-NEXT: vmovapd %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xdouble_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} xmm0 {%k1} {z} = xmm0[0,0]
; CHECK-NEXT: retq
%shuf = shufflevector <2 x double> %vec, <2 x double> undef, <2 x i32> <i32 0, i32 0>
; CHECK-LABEL: test_masked_2xdouble_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} xmm0 {%k1} = mem[0,0]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_z_2xdouble_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} xmm0 {%k1} {z} = mem[0,0]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_2xdouble_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} xmm0 {%k1} = mem[0,0]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_z_2xdouble_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} xmm0 {%k1} {z} = mem[0,0]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_4xdouble_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm1 {%k1} = ymm0[0,0,2,2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xdouble_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} {z} = ymm0[0,0,2,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec, <4 x double> undef, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
; CHECK-LABEL: test_masked_4xdouble_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm1 {%k1} = ymm0[0,0,2,2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xdouble_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} {z} = ymm0[0,0,2,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec, <4 x double> undef, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
; CHECK-LABEL: test_masked_4xdouble_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm1 {%k1} = ymm0[0,0,2,2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xdouble_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} {z} = ymm0[0,0,2,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec, <4 x double> undef, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
; CHECK-LABEL: test_masked_4xdouble_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm1 {%k1} = ymm0[0,0,2,2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xdouble_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} {z} = ymm0[0,0,2,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec, <4 x double> undef, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
; CHECK-LABEL: test_masked_4xdouble_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm1 {%k1} = ymm0[0,0,2,2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xdouble_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} {z} = ymm0[0,0,2,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec, <4 x double> undef, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
; CHECK-LABEL: test_masked_4xdouble_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $9, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_z_4xdouble_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $9, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} {z} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_4xdouble_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_z_4xdouble_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} {z} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_4xdouble_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_z_4xdouble_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} {z} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_4xdouble_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_z_4xdouble_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} {z} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_4xdouble_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_z_4xdouble_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} ymm0 {%k1} {z} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_8xdouble_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec, <8 x double> undef, <8 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6>
; CHECK-LABEL: test_masked_8xdouble_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $64, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $64, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec, <8 x double> undef, <8 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6>
; CHECK-LABEL: test_masked_8xdouble_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-24, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-24, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec, <8 x double> undef, <8 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6>
; CHECK-LABEL: test_masked_8xdouble_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec, <8 x double> undef, <8 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6>
; CHECK-LABEL: test_masked_8xdouble_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-50, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-50, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec, <8 x double> undef, <8 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6>
; CHECK-LABEL: test_masked_8xdouble_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-26, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-26, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_8xdouble_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $79, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $79, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_8xdouble_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-70, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-70, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_8xdouble_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-27, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-27, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_8xdouble_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-82, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-82, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_4xfloat_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm1 {%k1} = xmm0[0,0,2,2]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} {z} = xmm0[0,0,2,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
; CHECK-LABEL: test_masked_4xfloat_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm1 {%k1} = xmm0[0,0,2,2]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} {z} = xmm0[0,0,2,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
; CHECK-LABEL: test_masked_4xfloat_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm1 {%k1} = xmm0[0,0,2,2]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} {z} = xmm0[0,0,2,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
; CHECK-LABEL: test_masked_4xfloat_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm1 {%k1} = xmm0[0,0,2,2]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} {z} = xmm0[0,0,2,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
; CHECK-LABEL: test_masked_4xfloat_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm1 {%k1} = xmm0[0,0,2,2]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} {z} = xmm0[0,0,2,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
; CHECK-LABEL: test_masked_4xfloat_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} {z} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} {z} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} {z} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} {z} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $9, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $9, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} xmm0 {%k1} {z} = mem[0,0,2,2]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-116, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm1 {%k1} = ymm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-116, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} {z} = ymm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6>
; CHECK-LABEL: test_masked_8xfloat_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm1 {%k1} = ymm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} {z} = ymm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6>
; CHECK-LABEL: test_masked_8xfloat_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-73, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm1 {%k1} = ymm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-73, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} {z} = ymm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6>
; CHECK-LABEL: test_masked_8xfloat_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $102, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm1 {%k1} = ymm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $102, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} {z} = ymm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6>
; CHECK-LABEL: test_masked_8xfloat_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-46, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm1 {%k1} = ymm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-46, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} {z} = ymm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6>
; CHECK-LABEL: test_masked_8xfloat_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-86, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-86, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-35, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-35, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $62, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $62, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} ymm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $21312, %ax # imm = 0x5340
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $21312, %ax # imm = 0x5340
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6, i32 8, i32 8, i32 10, i32 10, i32 12, i32 12, i32 14, i32 14>
; CHECK-LABEL: test_masked_16xfloat_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-8490, %ax # imm = 0xDED6
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-8490, %ax # imm = 0xDED6
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6, i32 8, i32 8, i32 10, i32 10, i32 12, i32 12, i32 14, i32 14>
; CHECK-LABEL: test_masked_16xfloat_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $12522, %ax # imm = 0x30EA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $12522, %ax # imm = 0x30EA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6, i32 8, i32 8, i32 10, i32 10, i32 12, i32 12, i32 14, i32 14>
; CHECK-LABEL: test_masked_16xfloat_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-28344, %ax # imm = 0x9148
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-28344, %ax # imm = 0x9148
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6, i32 8, i32 8, i32 10, i32 10, i32 12, i32 12, i32 14, i32 14>
; CHECK-LABEL: test_masked_16xfloat_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movw $15638, %ax # imm = 0x3D16
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movw $15638, %ax # imm = 0x3D16
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6, i32 8, i32 8, i32 10, i32 10, i32 12, i32 12, i32 14, i32 14>
; CHECK-LABEL: test_masked_16xfloat_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-2129, %ax # imm = 0xF7AF
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-2129, %ax # imm = 0xF7AF
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-12900, %ax # imm = 0xCD9C
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-12900, %ax # imm = 0xCD9C
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $29358, %ax # imm = 0x72AE
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $29358, %ax # imm = 0x72AE
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $5272, %ax # imm = 0x1498
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $5272, %ax # imm = 0x1498
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movw $20975, %ax # imm = 0x51EF
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movw $20975, %ax # imm = 0x51EF
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mcpu=skx %s -o - | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl %s -o - | FileCheck %s
; FIXME: The non immediate <16 x float> test cases should be fixed by PR34382
; CHECK-LABEL: test_masked_4xfloat_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm1 {%k1} = xmm0[2,1,3,1]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm0 {%k1} {z} = xmm0[2,1,3,1]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 2, i32 1, i32 3, i32 1>
; CHECK-LABEL: test_masked_4xfloat_perm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm1 {%k1} = xmm0[1,2,3,2]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_perm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm0 {%k1} {z} = xmm0[1,2,3,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 1, i32 2, i32 3, i32 2>
; CHECK-LABEL: test_masked_4xfloat_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm1 {%k1} = xmm0[1,3,2,1]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm0 {%k1} {z} = xmm0[1,3,2,1]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 1, i32 3, i32 2, i32 1>
; CHECK-LABEL: test_masked_4xfloat_perm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm1 {%k1} = xmm0[1,2,3,2]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xfloat_perm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm0 {%k1} {z} = xmm0[1,2,3,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec, <4 x float> undef, <4 x i32> <i32 1, i32 2, i32 3, i32 2>
; CHECK-LABEL: test_masked_4xfloat_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm0 {%k1} = mem[3,3,1,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm0 {%k1} {z} = mem[3,3,1,3]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_perm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm0 {%k1} = mem[1,3,2,0]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_perm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm0 {%k1} {z} = mem[1,3,2,0]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm0 {%k1} = mem[2,1,3,2]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm0 {%k1} {z} = mem[2,1,3,2]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_4xfloat_perm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm0 {%k1} = mem[0,1,3,0]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_z_4xfloat_perm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} xmm0 {%k1} {z} = mem[0,1,3,0]
; CHECK-NEXT: retq
%vec = load <4 x float>, <4 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $83, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm1 {%k1} = ymm0[0,1,2,3,4,6,6,6]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $83, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = ymm0[0,1,2,3,4,6,6,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 6, i32 6, i32 6>
; CHECK-LABEL: test_masked_8xfloat_perm_imm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-34, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm1 {%k1} = ymm0[3,2,3,2,7,6,7,6]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_perm_imm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-34, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = ymm0[3,2,3,2,7,6,7,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 3, i32 2, i32 3, i32 2, i32 7, i32 6, i32 7, i32 6>
; CHECK-LABEL: test_masked_8xfloat_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $49, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm1 {%k1} = ymm0[2,1,2,1,6,5,4,4]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $49, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = ymm0[2,1,2,1,6,5,4,4]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 2, i32 1, i32 2, i32 1, i32 6, i32 5, i32 4, i32 4>
; CHECK-LABEL: test_masked_8xfloat_perm_imm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-111, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm1 {%k1} = ymm0[2,2,1,0,6,6,5,4]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_perm_imm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-111, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = ymm0[2,2,1,0,6,6,5,4]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 2, i32 2, i32 1, i32 0, i32 6, i32 6, i32 5, i32 4>
; CHECK-LABEL: test_masked_8xfloat_perm_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $61, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm1 {%k1} = ymm0[3,3,3,3,7,7,6,5]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_perm_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: movb $61, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = ymm0[3,3,3,3,7,7,6,5]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 3, i32 3, i32 3, i32 3, i32 7, i32 7, i32 6, i32 5>
; CHECK-LABEL: test_masked_8xfloat_perm_imm_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm1 {%k1} = ymm0[2,1,3,3,6,5,7,7]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_perm_imm_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = ymm0[2,1,3,3,6,5,7,7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 2, i32 1, i32 3, i32 3, i32 6, i32 5, i32 7, i32 7>
; CHECK-LABEL: test_masked_8xfloat_perm_mask6:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-51, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm1 {%k1} = ymm0[3,2,3,2,5,6,7,7]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_perm_mask6:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-51, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = ymm0[3,2,3,2,5,6,7,7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 3, i32 2, i32 3, i32 2, i32 5, i32 6, i32 7, i32 7>
; CHECK-LABEL: test_masked_8xfloat_perm_imm_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: movb $114, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm1 {%k1} = ymm0[3,0,2,1,7,4,6,5]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xfloat_perm_imm_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: movb $114, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = ymm0[3,0,2,1,7,4,6,5]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec, <8 x float> undef, <8 x i32> <i32 3, i32 0, i32 2, i32 1, i32 7, i32 4, i32 6, i32 5>
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %ymm1
; CHECK-NEXT: movb $-95, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} = ymm1[3,0,0,2,4,6,7,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %ymm0
; CHECK-NEXT: movb $-95, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = ymm0[3,0,0,2,4,6,7,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_perm_imm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-41, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} = mem[2,0,2,2,6,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_perm_imm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-41, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = mem[2,0,2,2,6,4,6,6]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %ymm1
; CHECK-NEXT: movb $62, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} = ymm1[2,1,1,3,4,4,7,4]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %ymm0
; CHECK-NEXT: movb $62, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = ymm0[2,1,1,3,4,4,7,4]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_perm_imm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-70, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} = mem[0,0,3,3,4,4,7,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_perm_imm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-70, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = mem[0,0,3,3,4,4,7,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %ymm1
; CHECK-NEXT: movb $30, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} = ymm1[0,1,0,1,4,6,5,4]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %ymm0
; CHECK-NEXT: movb $30, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = ymm0[0,1,0,1,4,6,5,4]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_perm_imm_mem_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: movb $56, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} = mem[2,0,0,3,6,4,4,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_perm_imm_mem_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: movb $56, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = mem[2,0,0,3,6,4,4,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %ymm1
; CHECK-NEXT: movb $-54, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} = ymm1[0,1,2,3,7,4,6,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %ymm0
; CHECK-NEXT: movb $-54, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = ymm0[0,1,2,3,7,4,6,7]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_8xfloat_perm_imm_mem_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: movb $85, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} = mem[0,2,3,1,4,6,7,5]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK-LABEL: test_masked_z_8xfloat_perm_imm_mem_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: movb $85, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} ymm0 {%k1} {z} = mem[0,2,3,1,4,6,7,5]
; CHECK-NEXT: retq
%vec = load <8 x float>, <8 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm2 = [1,1,3,1,6,4,6,5,8,9,8,11,13,13,13,15]
; CHECK-NEXT: movw $16429, %ax # imm = 0x402D
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm2, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm1 = [1,1,3,1,6,4,6,5,8,9,8,11,13,13,13,15]
; CHECK-NEXT: movw $16429, %ax # imm = 0x402D
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 1, i32 1, i32 3, i32 1, i32 6, i32 4, i32 6, i32 5, i32 8, i32 9, i32 8, i32 11, i32 13, i32 13, i32 13, i32 15>
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-26425, %ax # imm = 0x98C7
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm1 {%k1} = zmm0[2,2,2,1,6,6,6,5,10,10,10,9,14,14,14,13]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-26425, %ax # imm = 0x98C7
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm0[2,2,2,1,6,6,6,5,10,10,10,9,14,14,14,13]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 2, i32 2, i32 2, i32 1, i32 6, i32 6, i32 6, i32 5, i32 10, i32 10, i32 10, i32 9, i32 14, i32 14, i32 14, i32 13>
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm2 = [1,2,0,0,5,4,6,5,11,10,9,9,14,13,14,12]
; CHECK-NEXT: movw $28987, %ax # imm = 0x713B
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm2, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm1 = [1,2,0,0,5,4,6,5,11,10,9,9,14,13,14,12]
; CHECK-NEXT: movw $28987, %ax # imm = 0x713B
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 1, i32 2, i32 0, i32 0, i32 5, i32 4, i32 6, i32 5, i32 11, i32 10, i32 9, i32 9, i32 14, i32 13, i32 14, i32 12>
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $11457, %ax # imm = 0x2CC1
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm1 {%k1} = zmm0[1,1,0,2,5,5,4,6,9,9,8,10,13,13,12,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $11457, %ax # imm = 0x2CC1
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm0[1,1,0,2,5,5,4,6,9,9,8,10,13,13,12,14]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 1, i32 1, i32 0, i32 2, i32 5, i32 5, i32 4, i32 6, i32 9, i32 9, i32 8, i32 10, i32 13, i32 13, i32 12, i32 14>
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm2 = [1,2,3,3,5,5,5,7,11,11,8,11,14,12,14,15]
; CHECK-NEXT: movw $30908, %ax # imm = 0x78BC
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm2, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm1 = [1,2,3,3,5,5,5,7,11,11,8,11,14,12,14,15]
; CHECK-NEXT: movw $30908, %ax # imm = 0x78BC
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 1, i32 2, i32 3, i32 3, i32 5, i32 5, i32 5, i32 7, i32 11, i32 11, i32 8, i32 11, i32 14, i32 12, i32 14, i32 15>
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: movw $26863, %ax # imm = 0x68EF
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm1 {%k1} = zmm0[1,2,1,0,5,6,5,4,9,10,9,8,13,14,13,12]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: movw $26863, %ax # imm = 0x68EF
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2,1,0,5,6,5,4,9,10,9,8,13,14,13,12]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 1, i32 2, i32 1, i32 0, i32 5, i32 6, i32 5, i32 4, i32 9, i32 10, i32 9, i32 8, i32 13, i32 14, i32 13, i32 12>
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm2 = [2,0,3,2,4,4,6,7,9,11,8,11,13,12,13,13]
; CHECK-NEXT: movw $-28239, %ax # imm = 0x91B1
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm2, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm1 = [2,0,3,2,4,4,6,7,9,11,8,11,13,12,13,13]
; CHECK-NEXT: movw $-28239, %ax # imm = 0x91B1
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 2, i32 0, i32 3, i32 2, i32 4, i32 4, i32 6, i32 7, i32 9, i32 11, i32 8, i32 11, i32 13, i32 12, i32 13, i32 13>
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-32205, %ax # imm = 0x8233
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm1 {%k1} = zmm0[3,3,0,2,7,7,4,6,11,11,8,10,15,15,12,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-32205, %ax # imm = 0x8233
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm0[3,3,0,2,7,7,4,6,11,11,8,10,15,15,12,14]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec, <16 x float> undef, <16 x i32> <i32 3, i32 3, i32 0, i32 2, i32 7, i32 7, i32 4, i32 6, i32 11, i32 11, i32 8, i32 10, i32 15, i32 15, i32 12, i32 14>
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm1 = [3,3,3,0,6,6,6,6,11,10,9,10,12,14,12,12]
; CHECK-NEXT: movw $-22887, %ax # imm = 0xA699
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm1, %zmm0 {%k1}
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm0 = [3,3,3,0,6,6,6,6,11,10,9,10,12,14,12,12]
; CHECK-NEXT: movw $-22887, %ax # imm = 0xA699
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $22744, %ax # imm = 0x58D8
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} = mem[1,3,2,1,5,7,6,5,9,11,10,9,13,15,14,13]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $22744, %ax # imm = 0x58D8
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = mem[1,3,2,1,5,7,6,5,9,11,10,9,13,15,14,13]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm1 = [2,0,0,3,5,5,6,5,9,8,8,8,14,12,13,13]
; CHECK-NEXT: movw $-8399, %ax # imm = 0xDF31
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm1, %zmm0 {%k1}
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm0 = [2,0,0,3,5,5,6,5,9,8,8,8,14,12,13,13]
; CHECK-NEXT: movw $-8399, %ax # imm = 0xDF31
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $18246, %ax # imm = 0x4746
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} = mem[1,0,3,1,5,4,7,5,9,8,11,9,13,12,15,13]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $18246, %ax # imm = 0x4746
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = mem[1,0,3,1,5,4,7,5,9,8,11,9,13,12,15,13]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm1 = [3,3,1,1,6,5,5,6,11,11,10,9,15,14,12,12]
; CHECK-NEXT: movw $1218, %ax # imm = 0x4C2
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm1, %zmm0 {%k1}
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm0 = [3,3,1,1,6,5,5,6,11,11,10,9,15,14,12,12]
; CHECK-NEXT: movw $1218, %ax # imm = 0x4C2
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mem_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: movw $2665, %ax # imm = 0xA69
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} = mem[2,0,0,1,6,4,4,5,10,8,8,9,14,12,12,13]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mem_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: movw $2665, %ax # imm = 0xA69
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = mem[2,0,0,1,6,4,4,5,10,8,8,9,14,12,12,13]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm1 = [2,1,1,2,6,5,5,7,9,11,9,9,12,15,14,15]
; CHECK-NEXT: movw $-20907, %ax # imm = 0xAE55
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm1, %zmm0 {%k1}
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm0 = [2,1,1,2,6,5,5,7,9,11,9,9,12,15,14,15]
; CHECK-NEXT: movw $-20907, %ax # imm = 0xAE55
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mem_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-28944, %ax # imm = 0x8EF0
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} = mem[1,2,0,1,5,6,4,5,9,10,8,9,13,14,12,13]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mem_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-28944, %ax # imm = 0x8EF0
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = mem[1,2,0,1,5,6,4,5,9,10,8,9,13,14,12,13]
; CHECK-NEXT: retq
%vec = load <16 x float>, <16 x float>* %vp
; CHECK-LABEL: test_masked_2xdouble_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} xmm1 {%k1} = xmm0[1,0]
; CHECK-NEXT: vmovapd %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xdouble_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} xmm0 {%k1} {z} = xmm0[1,0]
; CHECK-NEXT: retq
%shuf = shufflevector <2 x double> %vec, <2 x double> undef, <2 x i32> <i32 1, i32 0>
; CHECK-LABEL: test_masked_2xdouble_perm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} xmm1 {%k1} = xmm0[1,0]
; CHECK-NEXT: vmovapd %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_2xdouble_perm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} xmm0 {%k1} {z} = xmm0[1,0]
; CHECK-NEXT: retq
%shuf = shufflevector <2 x double> %vec, <2 x double> undef, <2 x i32> <i32 1, i32 0>
; CHECK-LABEL: test_masked_2xdouble_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} xmm0 {%k1} = mem[1,0]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_z_2xdouble_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} xmm0 {%k1} {z} = mem[1,0]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_2xdouble_perm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} xmm0 {%k1} = mem[1,0]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_z_2xdouble_perm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} xmm0 {%k1} {z} = mem[1,0]
; CHECK-NEXT: retq
%vec = load <2 x double>, <2 x double>* %vp
; CHECK-LABEL: test_masked_4xdouble_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm1 {%k1} = ymm0[1,0,2,3]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xdouble_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1,0,2,3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec, <4 x double> undef, <4 x i32> <i32 1, i32 0, i32 2, i32 3>
; CHECK-LABEL: test_masked_4xdouble_perm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm1 {%k1} = ymm0[1,1,2,2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xdouble_perm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1,1,2,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec, <4 x double> undef, <4 x i32> <i32 1, i32 1, i32 2, i32 2>
; CHECK-LABEL: test_masked_4xdouble_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $9, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm1 {%k1} = ymm0[0,1,3,3]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xdouble_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $9, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0,1,3,3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec, <4 x double> undef, <4 x i32> <i32 0, i32 1, i32 3, i32 3>
; CHECK-LABEL: test_masked_4xdouble_perm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm1 {%k1} = ymm0[1,1,2,2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_4xdouble_perm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1,1,2,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec, <4 x double> undef, <4 x i32> <i32 1, i32 1, i32 2, i32 2>
; CHECK-LABEL: test_masked_4xdouble_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm0 {%k1} = mem[0,1,2,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_z_4xdouble_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm0 {%k1} {z} = mem[0,1,2,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_4xdouble_perm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm0 {%k1} = mem[0,1,3,3]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_z_4xdouble_perm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm0 {%k1} {z} = mem[0,1,3,3]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_4xdouble_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm0 {%k1} = mem[1,0,3,3]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_z_4xdouble_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm0 {%k1} {z} = mem[1,0,3,3]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_4xdouble_perm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm0 {%k1} = mem[1,0,3,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_z_4xdouble_perm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} ymm0 {%k1} {z} = mem[1,0,3,2]
; CHECK-NEXT: retq
%vec = load <4 x double>, <4 x double>* %vp
; CHECK-LABEL: test_masked_8xdouble_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-107, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm1 {%k1} = zmm0[0,0,3,2,4,5,7,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xdouble_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-107, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,3,2,4,5,7,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec, <8 x double> undef, <8 x i32> <i32 0, i32 0, i32 3, i32 2, i32 4, i32 5, i32 7, i32 6>
; CHECK-LABEL: test_masked_8xdouble_perm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-39, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm1 {%k1} = zmm0[0,1,2,3,4,4,7,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xdouble_perm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-39, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0,1,2,3,4,4,7,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec, <8 x double> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 4, i32 7, i32 6>
; CHECK-LABEL: test_masked_8xdouble_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-53, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,3,5,5,6,7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xdouble_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-53, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,3,5,5,6,7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec, <8 x double> undef, <8 x i32> <i32 0, i32 0, i32 2, i32 3, i32 5, i32 5, i32 6, i32 7>
; CHECK-LABEL: test_masked_8xdouble_perm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-89, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm1 {%k1} = zmm0[0,1,2,2,4,4,6,7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_masked_z_8xdouble_perm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-89, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0,1,2,2,4,4,6,7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec, <8 x double> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 2, i32 4, i32 4, i32 6, i32 7>
; CHECK-LABEL: test_masked_8xdouble_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-95, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,5,4,7,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_z_8xdouble_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-95, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,5,4,7,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_8xdouble_perm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $27, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} = mem[0,1,3,3,4,5,7,7]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_z_8xdouble_perm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $27, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = mem[0,1,3,3,4,5,7,7]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_8xdouble_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-116, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} = mem[1,1,3,3,5,4,7,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_z_8xdouble_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-116, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = mem[1,1,3,3,5,4,7,6]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_8xdouble_perm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $89, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} = mem[1,0,3,2,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; CHECK-LABEL: test_masked_z_8xdouble_perm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $89, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = mem[1,0,3,2,4,5,6,7]
; CHECK-NEXT: retq
%vec = load <8 x double>, <8 x double>* %vp
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mcpu=skx %s -o - | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl,+avx512bw %s -o - | FileCheck %s
; FIXME: All cases here should be fixed by PR34380
; CHECK: # BB#0:
; CHECK-NEXT: vextractf64x4 $1, %zmm0, %ymm2
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm2[1],ymm0[3],ymm2[3]
+; CHECK-NEXT: vextractf128 $1, %ymm0, %xmm0
; CHECK-NEXT: movb $1, %al
; CHECK-NEXT: kmovd %eax, %k1
-; CHECK-NEXT: vextractf64x2 $1, %ymm0, %xmm1 {%k1}
-; CHECK-NEXT: vmovapd %xmm1, %xmm0
+; CHECK-NEXT: vblendmpd %xmm0, %xmm1, %xmm0 {%k1}
; CHECK-NEXT: vzeroupper
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec, <8 x double> undef, <2 x i32> <i32 3, i32 7>
; CHECK: # BB#0:
; CHECK-NEXT: vextractf64x4 $1, %zmm0, %ymm1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
+; CHECK-NEXT: vextractf128 $1, %ymm0, %xmm0
; CHECK-NEXT: movb $1, %al
; CHECK-NEXT: kmovd %eax, %k1
-; CHECK-NEXT: vextractf64x2 $1, %ymm0, %xmm0 {%k1} {z}
+; CHECK-NEXT: vmovapd %xmm0, %xmm0 {%k1} {z}
; CHECK-NEXT: vzeroupper
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec, <8 x double> undef, <2 x i32> <i32 3, i32 7>
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mcpu=skx %s -o - | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl,+avx512bw %s -o - | FileCheck %s
define <16 x i16> @test_16xi16_perm_mask0(<16 x i16> %vec) {
; CHECK-LABEL: test_16xi16_perm_mask0:
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mcpu=skx %s -o - | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl %s -o - | FileCheck %s
define <4 x float> @test_4xfloat_shuff_mask0(<4 x float> %vec1, <4 x float> %vec2) {
; CHECK-LABEL: test_4xfloat_shuff_mask0:
; CHECK-LABEL: test_4xfloat_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm2 {%k1} = xmm0[2,1],xmm1[3,1]
; CHECK-NEXT: vmovaps %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm0 {%k1} {z} = xmm0[2,1],xmm1[3,1]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec1, <4 x float> %vec2, <4 x i32> <i32 2, i32 1, i32 7, i32 5>
; CHECK-LABEL: test_4xfloat_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm2 {%k1} = xmm0[1,2],xmm1[3,2]
; CHECK-NEXT: vmovaps %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm0 {%k1} {z} = xmm0[1,2],xmm1[3,2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec1, <4 x float> %vec2, <4 x i32> <i32 1, i32 2, i32 7, i32 6>
; CHECK-LABEL: test_4xfloat_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm2 {%k1} = xmm0[1,3],xmm1[2,1]
; CHECK-NEXT: vmovaps %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm0 {%k1} {z} = xmm0[1,3],xmm1[2,1]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec1, <4 x float> %vec2, <4 x i32> <i32 1, i32 3, i32 6, i32 5>
; CHECK-LABEL: test_4xfloat_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm2 {%k1} = xmm0[3,3],xmm1[3,3]
; CHECK-NEXT: vmovaps %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm0 {%k1} {z} = xmm0[3,3],xmm1[3,3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec1, <4 x float> %vec2, <4 x i32> <i32 3, i32 3, i32 7, i32 7>
; CHECK-LABEL: test_4xfloat_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm1 {%k1} = xmm0[1,0],mem[1,2]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm0 {%k1} {z} = xmm0[1,0],mem[1,2]
; CHECK-NEXT: retq
%vec2 = load <4 x float>, <4 x float>* %vec2p
; CHECK-LABEL: test_4xfloat_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm1 {%k1} = xmm0[3,3],mem[1,3]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm0 {%k1} {z} = xmm0[3,3],mem[1,3]
; CHECK-NEXT: retq
%vec2 = load <4 x float>, <4 x float>* %vec2p
; CHECK-LABEL: test_4xfloat_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm1 {%k1} = xmm0[1,3],mem[2,0]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm0 {%k1} {z} = xmm0[1,3],mem[2,0]
; CHECK-NEXT: retq
%vec2 = load <4 x float>, <4 x float>* %vec2p
; CHECK-LABEL: test_4xfloat_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm1 {%k1} = xmm0[2,1],mem[3,2]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} xmm0 {%k1} {z} = xmm0[2,1],mem[3,2]
; CHECK-NEXT: retq
%vec2 = load <4 x float>, <4 x float>* %vec2p
; CHECK-LABEL: test_8xfloat_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm2 {%k1} = ymm0[1,3],ymm1[0,2],ymm0[5,7],ymm1[4,6]
; CHECK-NEXT: vmovaps %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm0 {%k1} {z} = ymm0[1,3],ymm1[0,2],ymm0[5,7],ymm1[4,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 1, i32 3, i32 8, i32 10, i32 5, i32 7, i32 12, i32 14>
; CHECK-LABEL: test_8xfloat_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm2 {%k1} = ymm0[0,3],ymm1[3,1],ymm0[4,7],ymm1[7,5]
; CHECK-NEXT: vmovaps %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm0 {%k1} {z} = ymm0[0,3],ymm1[3,1],ymm0[4,7],ymm1[7,5]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 0, i32 3, i32 11, i32 9, i32 4, i32 7, i32 15, i32 13>
; CHECK-LABEL: test_8xfloat_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-35, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm2 {%k1} = ymm0[0,2],ymm1[2,2],ymm0[4,6],ymm1[6,6]
; CHECK-NEXT: vmovaps %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-35, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm0 {%k1} {z} = ymm0[0,2],ymm1[2,2],ymm0[4,6],ymm1[6,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 0, i32 2, i32 10, i32 10, i32 4, i32 6, i32 14, i32 14>
; CHECK-LABEL: test_8xfloat_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $62, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm2 {%k1} = ymm0[3,2],ymm1[3,2],ymm0[7,6],ymm1[7,6]
; CHECK-NEXT: vmovaps %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $62, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm0 {%k1} {z} = ymm0[3,2],ymm1[3,2],ymm0[7,6],ymm1[7,6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 3, i32 2, i32 11, i32 10, i32 7, i32 6, i32 15, i32 14>
; CHECK-LABEL: test_8xfloat_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-106, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm1 {%k1} = ymm0[2,1],mem[0,0],ymm0[6,5],mem[4,4]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-106, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm0 {%k1} {z} = ymm0[2,1],mem[0,0],ymm0[6,5],mem[4,4]
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK-LABEL: test_8xfloat_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $114, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm1 {%k1} = ymm0[2,2],mem[1,0],ymm0[6,6],mem[5,4]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $114, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm0 {%k1} {z} = ymm0[2,2],mem[1,0],ymm0[6,6],mem[5,4]
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK-LABEL: test_8xfloat_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-104, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm1 {%k1} = ymm0[3,3],mem[3,3],ymm0[7,7],mem[7,7]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-104, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm0 {%k1} {z} = ymm0[3,3],mem[3,3],ymm0[7,7],mem[7,7]
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK-LABEL: test_8xfloat_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm1 {%k1} = ymm0[3,3],mem[2,1],ymm0[7,7],mem[6,5]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} ymm0 {%k1} {z} = ymm0[3,3],mem[2,1],ymm0[7,7],mem[6,5]
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-19315, %ax # imm = 0xB48D
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm2 {%k1} = zmm0[3,2],zmm1[3,2],zmm0[7,6],zmm1[7,6],zmm0[11,10],zmm1[11,10],zmm0[15,14],zmm1[15,14]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-19315, %ax # imm = 0xB48D
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[3,2],zmm1[3,2],zmm0[7,6],zmm1[7,6],zmm0[11,10],zmm1[11,10],zmm0[15,14],zmm1[15,14]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 3, i32 2, i32 19, i32 18, i32 7, i32 6, i32 23, i32 22, i32 11, i32 10, i32 27, i32 26, i32 15, i32 14, i32 31, i32 30>
; CHECK-LABEL: test_16xfloat_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $18064, %ax # imm = 0x4690
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm2 {%k1} = zmm0[1,2],zmm1[3,3],zmm0[5,6],zmm1[7,7],zmm0[9,10],zmm1[11,11],zmm0[13,14],zmm1[15,15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $18064, %ax # imm = 0x4690
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2],zmm1[3,3],zmm0[5,6],zmm1[7,7],zmm0[9,10],zmm1[11,11],zmm0[13,14],zmm1[15,15]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 1, i32 2, i32 19, i32 19, i32 5, i32 6, i32 23, i32 23, i32 9, i32 10, i32 27, i32 27, i32 13, i32 14, i32 31, i32 31>
; CHECK-LABEL: test_16xfloat_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-12346, %ax # imm = 0xCFC6
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm2 {%k1} = zmm0[3,0],zmm1[2,1],zmm0[7,4],zmm1[6,5],zmm0[11,8],zmm1[10,9],zmm0[15,12],zmm1[14,13]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-12346, %ax # imm = 0xCFC6
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[3,0],zmm1[2,1],zmm0[7,4],zmm1[6,5],zmm0[11,8],zmm1[10,9],zmm0[15,12],zmm1[14,13]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 3, i32 0, i32 18, i32 17, i32 7, i32 4, i32 22, i32 21, i32 11, i32 8, i32 26, i32 25, i32 15, i32 12, i32 30, i32 29>
; CHECK-LABEL: test_16xfloat_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-9865, %ax # imm = 0xD977
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm2 {%k1} = zmm0[2,3],zmm1[0,2],zmm0[6,7],zmm1[4,6],zmm0[10,11],zmm1[8,10],zmm0[14,15],zmm1[12,14]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-9865, %ax # imm = 0xD977
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[2,3],zmm1[0,2],zmm0[6,7],zmm1[4,6],zmm0[10,11],zmm1[8,10],zmm0[14,15],zmm1[12,14]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 2, i32 3, i32 16, i32 18, i32 6, i32 7, i32 20, i32 22, i32 10, i32 11, i32 24, i32 26, i32 14, i32 15, i32 28, i32 30>
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $7677, %ax # imm = 0x1DFD
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm1 {%k1} = zmm0[3,0],mem[0,2],zmm0[7,4],mem[4,6],zmm0[11,8],mem[8,10],zmm0[15,12],mem[12,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $7677, %ax # imm = 0x1DFD
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[3,0],mem[0,2],zmm0[7,4],mem[4,6],zmm0[11,8],mem[8,10],zmm0[15,12],mem[12,14]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $14448, %ax # imm = 0x3870
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm1 {%k1} = zmm0[0,2],mem[3,2],zmm0[4,6],mem[7,6],zmm0[8,10],mem[11,10],zmm0[12,14],mem[15,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $14448, %ax # imm = 0x3870
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[0,2],mem[3,2],zmm0[4,6],mem[7,6],zmm0[8,10],mem[11,10],zmm0[12,14],mem[15,14]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-13463, %ax # imm = 0xCB69
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm1 {%k1} = zmm0[2,0],mem[2,2],zmm0[6,4],mem[6,6],zmm0[10,8],mem[10,10],zmm0[14,12],mem[14,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-13463, %ax # imm = 0xCB69
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[2,0],mem[2,2],zmm0[6,4],mem[6,6],zmm0[10,8],mem[10,10],zmm0[14,12],mem[14,14]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $21793, %ax # imm = 0x5521
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm1 {%k1} = zmm0[2,1],mem[1,3],zmm0[6,5],mem[5,7],zmm0[10,9],mem[9,11],zmm0[14,13],mem[13,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $21793, %ax # imm = 0x5521
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[2,1],mem[1,3],zmm0[6,5],mem[5,7],zmm0[10,9],mem[9,11],zmm0[14,13],mem[13,15]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_2xdouble_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} xmm2 {%k1} = xmm0[1],xmm1[0]
; CHECK-NEXT: vmovapd %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_2xdouble_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} xmm0 {%k1} {z} = xmm0[1],xmm1[0]
; CHECK-NEXT: retq
%shuf = shufflevector <2 x double> %vec1, <2 x double> %vec2, <2 x i32> <i32 1, i32 2>
; CHECK-LABEL: test_2xdouble_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} xmm2 {%k1} = xmm0[1],xmm1[0]
; CHECK-NEXT: vmovapd %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_2xdouble_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} xmm0 {%k1} {z} = xmm0[1],xmm1[0]
; CHECK-NEXT: retq
%shuf = shufflevector <2 x double> %vec1, <2 x double> %vec2, <2 x i32> <i32 1, i32 2>
; CHECK-LABEL: test_2xdouble_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} xmm1 {%k1} = xmm0[1],mem[0]
; CHECK-NEXT: vmovapd %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_2xdouble_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} xmm0 {%k1} {z} = xmm0[1],mem[0]
; CHECK-NEXT: retq
%vec2 = load <2 x double>, <2 x double>* %vec2p
; CHECK-LABEL: test_2xdouble_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} xmm1 {%k1} = xmm0[1],mem[0]
; CHECK-NEXT: vmovapd %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_2xdouble_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} xmm0 {%k1} {z} = xmm0[1],mem[0]
; CHECK-NEXT: retq
%vec2 = load <2 x double>, <2 x double>* %vec2p
; CHECK-LABEL: test_4xdouble_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm2 {%k1} = ymm0[0],ymm1[0],ymm0[3],ymm1[3]
; CHECK-NEXT: vmovapd %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0],ymm1[0],ymm0[3],ymm1[3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 0, i32 4, i32 3, i32 7>
; CHECK-LABEL: test_4xdouble_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm2 {%k1} = ymm0[0],ymm1[0],ymm0[3],ymm1[2]
; CHECK-NEXT: vmovapd %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0],ymm1[0],ymm0[3],ymm1[2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 0, i32 4, i32 3, i32 6>
; CHECK-LABEL: test_4xdouble_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm2 {%k1} = ymm0[1],ymm1[0],ymm0[3],ymm1[2]
; CHECK-NEXT: vmovapd %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1],ymm1[0],ymm0[3],ymm1[2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 1, i32 4, i32 3, i32 6>
; CHECK-LABEL: test_4xdouble_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm2 {%k1} = ymm0[0],ymm1[0],ymm0[2],ymm1[3]
; CHECK-NEXT: vmovapd %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0],ymm1[0],ymm0[2],ymm1[3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 0, i32 4, i32 2, i32 7>
; CHECK-LABEL: test_4xdouble_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm1 {%k1} = ymm0[1],mem[1],ymm0[3],mem[2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1],mem[1],ymm0[3],mem[2]
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK-LABEL: test_4xdouble_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm1 {%k1} = ymm0[0],mem[1],ymm0[2],mem[2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0],mem[1],ymm0[2],mem[2]
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK-LABEL: test_4xdouble_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm1 {%k1} = ymm0[0],mem[0],ymm0[3],mem[2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0],mem[0],ymm0[3],mem[2]
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK-LABEL: test_4xdouble_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm1 {%k1} = ymm0[1],mem[1],ymm0[2],mem[2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1],mem[1],ymm0[2],mem[2]
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-77, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[2],zmm1[3],zmm0[4],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-77, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[2],zmm1[3],zmm0[4],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 0, i32 8, i32 2, i32 11, i32 4, i32 13, i32 7, i32 15>
; CHECK-LABEL: test_8xdouble_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $107, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[2],zmm1[3],zmm0[5],zmm1[5],zmm0[6],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $107, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[2],zmm1[3],zmm0[5],zmm1[5],zmm0[6],zmm1[7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 0, i32 8, i32 2, i32 11, i32 5, i32 13, i32 6, i32 15>
; CHECK-LABEL: test_8xdouble_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-87, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm2 {%k1} = zmm0[1],zmm1[0],zmm0[3],zmm1[3],zmm0[4],zmm1[5],zmm0[6],zmm1[6]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-87, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],zmm1[0],zmm0[3],zmm1[3],zmm0[4],zmm1[5],zmm0[6],zmm1[6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 1, i32 8, i32 3, i32 11, i32 4, i32 13, i32 6, i32 14>
; CHECK-LABEL: test_8xdouble_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm2 {%k1} = zmm0[1],zmm1[0],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[7],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],zmm1[0],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[7],zmm1[7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 1, i32 8, i32 3, i32 11, i32 4, i32 12, i32 7, i32 15>
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $72, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[5],mem[5],zmm0[6],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $72, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[5],mem[5],zmm0[6],mem[7]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[0],zmm0[3],mem[2],zmm0[4],mem[4],zmm0[7],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[0],zmm0[3],mem[2],zmm0[4],mem[4],zmm0[7],mem[7]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $26, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[1],zmm0[3],mem[2],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $26, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[1],zmm0[3],mem[2],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-39, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[1],zmm0[2],mem[3],zmm0[4],mem[5],zmm0[6],mem[6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-39, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[1],zmm0[2],mem[3],zmm0[4],mem[5],zmm0[6],mem[6]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mcpu=skx %s -o - | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl %s -o - | FileCheck %s
; FIXME: 128-bit shuffles of 256-bit vectors cases should be fixed by PR34359
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $-41, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmps %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $-41, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovaps %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $-63, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmps %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $-63, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovaps %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $107, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmps %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 12, i32 13, i32 14, i32 15>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $107, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovaps %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 12, i32 13, i32 14, i32 15>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $66, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmps %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $66, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovaps %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $-24, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmps %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $-24, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovaps %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $-6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmps %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $-6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovaps %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $-50, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmps %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $-50, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovaps %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $-26, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmps %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $-26, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovaps %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-11480, %ax # imm = 0xD328
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm2 {%k1} = zmm0[12,13,14,15,0,1,2,3],zmm1[4,5,6,7,12,13,14,15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-11480, %ax # imm = 0xD328
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[12,13,14,15,0,1,2,3],zmm1[4,5,6,7,12,13,14,15]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 12, i32 13, i32 14, i32 15, i32 0, i32 1, i32 2, i32 3, i32 20, i32 21, i32 22, i32 23, i32 28, i32 29, i32 30, i32 31>
; CHECK-LABEL: test_16xfloat_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-21749, %ax # imm = 0xAB0B
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm2 {%k1} = zmm0[0,1,2,3,8,9,10,11],zmm1[0,1,2,3,12,13,14,15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-21749, %ax # imm = 0xAB0B
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[0,1,2,3,8,9,10,11],zmm1[0,1,2,3,12,13,14,15]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19, i32 28, i32 29, i32 30, i32 31>
; CHECK-LABEL: test_16xfloat_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $75, %ax
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm2 {%k1} = zmm0[12,13,14,15,4,5,6,7],zmm1[0,1,2,3,4,5,6,7]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $75, %ax
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[12,13,14,15,4,5,6,7],zmm1[0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 12, i32 13, i32 14, i32 15, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
; CHECK-LABEL: test_16xfloat_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $32347, %ax # imm = 0x7E5B
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm2 {%k1} = zmm0[8,9,10,11,12,13,14,15],zmm1[0,1,2,3,8,9,10,11]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $32347, %ax # imm = 0x7E5B
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[8,9,10,11,12,13,14,15],zmm1[0,1,2,3,8,9,10,11]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 24, i32 25, i32 26, i32 27>
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-19232, %ax # imm = 0xB4E0
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm1 {%k1} = zmm0[12,13,14,15,8,9,10,11],mem[8,9,10,11,4,5,6,7]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-19232, %ax # imm = 0xB4E0
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[12,13,14,15,8,9,10,11],mem[8,9,10,11,4,5,6,7]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-29660, %ax # imm = 0x8C24
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm1 {%k1} = zmm0[8,9,10,11,4,5,6,7],mem[8,9,10,11,4,5,6,7]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-29660, %ax # imm = 0x8C24
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[8,9,10,11,4,5,6,7],mem[8,9,10,11,4,5,6,7]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-12160, %ax # imm = 0xD080
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm1 {%k1} = zmm0[0,1,2,3,0,1,2,3],mem[8,9,10,11,8,9,10,11]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-12160, %ax # imm = 0xD080
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[0,1,2,3,0,1,2,3],mem[8,9,10,11,8,9,10,11]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-30129, %ax # imm = 0x8A4F
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm1 {%k1} = zmm0[4,5,6,7,0,1,2,3],mem[12,13,14,15,12,13,14,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-30129, %ax # imm = 0x8A4F
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[4,5,6,7,0,1,2,3],mem[12,13,14,15,12,13,14,15]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmpd %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovapd %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmpd %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovapd %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmpd %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 2, i32 3, i32 6, i32 7>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovapd %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 2, i32 3, i32 6, i32 7>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmpd %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 2, i32 3, i32 6, i32 7>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovapd %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 2, i32 3, i32 6, i32 7>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmpd %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovapd %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmpd %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovapd %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmpd %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovapd %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vblendmpd %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovapd %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $62, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm2 {%k1} = zmm0[6,7,2,3],zmm1[6,7,0,1]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $62, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[6,7,2,3],zmm1[6,7,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 6, i32 7, i32 2, i32 3, i32 14, i32 15, i32 8, i32 9>
; CHECK-LABEL: test_8xdouble_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-70, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm2 {%k1} = zmm0[0,1,4,5],zmm1[0,1,4,5]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-70, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[0,1,4,5],zmm1[0,1,4,5]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 0, i32 1, i32 4, i32 5, i32 8, i32 9, i32 12, i32 13>
; CHECK-LABEL: test_8xdouble_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $30, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm2 {%k1} = zmm0[6,7,4,5],zmm1[4,5,0,1]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $30, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[6,7,4,5],zmm1[4,5,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 6, i32 7, i32 4, i32 5, i32 12, i32 13, i32 8, i32 9>
; CHECK-LABEL: test_8xdouble_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $56, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm2 {%k1} = zmm0[4,5,4,5],zmm1[4,5,2,3]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $56, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[4,5,4,5],zmm1[4,5,2,3]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 4, i32 5, i32 4, i32 5, i32 12, i32 13, i32 10, i32 11>
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $95, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm1 {%k1} = zmm0[6,7,0,1],mem[0,1,0,1]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $95, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[6,7,0,1],mem[0,1,0,1]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm1 {%k1} = zmm0[6,7,6,7],mem[0,1,2,3]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[6,7,6,7],mem[0,1,2,3]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm1 {%k1} = zmm0[0,1,2,3],mem[0,1,4,5]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[0,1,2,3],mem[0,1,4,5]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm1 {%k1} = zmm0[2,3,0,1],mem[4,5,0,1]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[2,3,0,1],mem[4,5,0,1]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $26, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmd %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i32> %vec1, <8 x i32> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 12, i32 13, i32 14, i32 15>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $26, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa32 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i32> %vec1, <8 x i32> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 12, i32 13, i32 14, i32 15>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $-4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmd %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i32> %vec1, <8 x i32> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $-4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa32 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i32> %vec1, <8 x i32> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $51, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmd %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i32> %vec1, <8 x i32> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 12, i32 13, i32 14, i32 15>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $51, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa32 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i32> %vec1, <8 x i32> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 12, i32 13, i32 14, i32 15>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $92, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmd %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i32> %vec1, <8 x i32> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $92, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa32 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i32> %vec1, <8 x i32> %vec2, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $64, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmd %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <8 x i32>, <8 x i32>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $64, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa32 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <8 x i32>, <8 x i32>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $-104, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmd %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <8 x i32>, <8 x i32>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $-104, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa32 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <8 x i32>, <8 x i32>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $113, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmd %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <8 x i32>, <8 x i32>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $113, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa32 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <8 x i32>, <8 x i32>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $45, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmd %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <8 x i32>, <8 x i32>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $45, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa32 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <8 x i32>, <8 x i32>* %vec2p
; CHECK-LABEL: test_16xi32_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $2995, %ax # imm = 0xBB3
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm2 {%k1} = zmm0[4,5,6,7,4,5,6,7],zmm1[4,5,6,7,12,13,14,15]
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xi32_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $2995, %ax # imm = 0xBB3
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[4,5,6,7,4,5,6,7],zmm1[4,5,6,7,12,13,14,15]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x i32> %vec1, <16 x i32> %vec2, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 4, i32 5, i32 6, i32 7, i32 20, i32 21, i32 22, i32 23, i32 28, i32 29, i32 30, i32 31>
; CHECK-LABEL: test_16xi32_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $18408, %ax # imm = 0x47E8
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm2 {%k1} = zmm0[8,9,10,11,8,9,10,11],zmm1[8,9,10,11,4,5,6,7]
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xi32_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $18408, %ax # imm = 0x47E8
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[8,9,10,11,8,9,10,11],zmm1[8,9,10,11,4,5,6,7]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x i32> %vec1, <16 x i32> %vec2, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 8, i32 9, i32 10, i32 11, i32 24, i32 25, i32 26, i32 27, i32 20, i32 21, i32 22, i32 23>
; CHECK-LABEL: test_16xi32_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $15737, %ax # imm = 0x3D79
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm2 {%k1} = zmm0[4,5,6,7,8,9,10,11],zmm1[0,1,2,3,0,1,2,3]
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xi32_zero_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $15737, %ax # imm = 0x3D79
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[4,5,6,7,8,9,10,11],zmm1[0,1,2,3,0,1,2,3]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x i32> %vec1, <16 x i32> %vec2, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19, i32 16, i32 17, i32 18, i32 19>
; CHECK-LABEL: test_16xi32_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-3073, %ax # imm = 0xF3FF
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm2 {%k1} = zmm0[4,5,6,7,0,1,2,3],zmm1[8,9,10,11,4,5,6,7]
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xi32_zero_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-3073, %ax # imm = 0xF3FF
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[4,5,6,7,0,1,2,3],zmm1[8,9,10,11,4,5,6,7]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x i32> %vec1, <16 x i32> %vec2, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 24, i32 25, i32 26, i32 27, i32 20, i32 21, i32 22, i32 23>
; CHECK-LABEL: test_16xi32_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-8166, %ax # imm = 0xE01A
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm1 {%k1} = zmm0[8,9,10,11,4,5,6,7],mem[8,9,10,11,0,1,2,3]
; CHECK-NEXT: vmovdqa64 %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xi32_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-8166, %ax # imm = 0xE01A
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[8,9,10,11,4,5,6,7],mem[8,9,10,11,0,1,2,3]
; CHECK-NEXT: retq
%vec2 = load <16 x i32>, <16 x i32>* %vec2p
; CHECK-LABEL: test_16xi32_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-28302, %ax # imm = 0x9172
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm1 {%k1} = zmm0[4,5,6,7,4,5,6,7],mem[0,1,2,3,8,9,10,11]
; CHECK-NEXT: vmovdqa64 %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xi32_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-28302, %ax # imm = 0x9172
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[4,5,6,7,4,5,6,7],mem[0,1,2,3,8,9,10,11]
; CHECK-NEXT: retq
%vec2 = load <16 x i32>, <16 x i32>* %vec2p
; CHECK-LABEL: test_16xi32_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $27158, %ax # imm = 0x6A16
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm1 {%k1} = zmm0[4,5,6,7,8,9,10,11],mem[12,13,14,15,12,13,14,15]
; CHECK-NEXT: vmovdqa64 %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xi32_zero_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $27158, %ax # imm = 0x6A16
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[4,5,6,7,8,9,10,11],mem[12,13,14,15,12,13,14,15]
; CHECK-NEXT: retq
%vec2 = load <16 x i32>, <16 x i32>* %vec2p
; CHECK-LABEL: test_16xi32_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $26363, %ax # imm = 0x66FB
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm1 {%k1} = zmm0[4,5,6,7,4,5,6,7],mem[4,5,6,7,12,13,14,15]
; CHECK-NEXT: vmovdqa64 %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xi32_zero_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $26363, %ax # imm = 0x66FB
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[4,5,6,7,4,5,6,7],mem[4,5,6,7,12,13,14,15]
; CHECK-NEXT: retq
%vec2 = load <16 x i32>, <16 x i32>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmq %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x i64> %vec1, <4 x i64> %vec2, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa64 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x i64> %vec1, <4 x i64> %vec2, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmq %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x i64> %vec1, <4 x i64> %vec2, <4 x i32> <i32 2, i32 3, i32 6, i32 7>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa64 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x i64> %vec1, <4 x i64> %vec2, <4 x i32> <i32 2, i32 3, i32 6, i32 7>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmq %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x i64> %vec1, <4 x i64> %vec2, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa64 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x i64> %vec1, <4 x i64> %vec2, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmq %ymm0, %ymm2, %ymm0 {%k1}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x i64> %vec1, <4 x i64> %vec2, <4 x i32> <i32 2, i32 3, i32 6, i32 7>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa64 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%shuf = shufflevector <4 x i64> %vec1, <4 x i64> %vec2, <4 x i32> <i32 2, i32 3, i32 6, i32 7>
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmq %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <4 x i64>, <4 x i64>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa64 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <4 x i64>, <4 x i64>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmq %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <4 x i64>, <4 x i64>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa64 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <4 x i64>, <4 x i64>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmq %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <4 x i64>, <4 x i64>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1]
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa64 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <4 x i64>, <4 x i64>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vpblendmq %ymm0, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: retq
%vec2 = load <4 x i64>, <4 x i64>* %vec2p
; CHECK: # BB#0:
; CHECK-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vmovdqa64 %ymm0, %ymm0 {%k1} {z}
; CHECK-NEXT: retq
%vec2 = load <4 x i64>, <4 x i64>* %vec2p
; CHECK-LABEL: test_8xi64_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-15, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm2 {%k1} = zmm0[4,5,4,5],zmm1[4,5,4,5]
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xi64_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-15, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[4,5,4,5],zmm1[4,5,4,5]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i64> %vec1, <8 x i64> %vec2, <8 x i32> <i32 4, i32 5, i32 4, i32 5, i32 12, i32 13, i32 12, i32 13>
; CHECK-LABEL: test_8xi64_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-17, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm2 {%k1} = zmm0[6,7,4,5],zmm1[2,3,4,5]
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xi64_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-17, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[6,7,4,5],zmm1[2,3,4,5]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i64> %vec1, <8 x i64> %vec2, <8 x i32> <i32 6, i32 7, i32 4, i32 5, i32 10, i32 11, i32 12, i32 13>
; CHECK-LABEL: test_8xi64_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-24, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm2 {%k1} = zmm0[0,1,4,5],zmm1[0,1,0,1]
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xi64_zero_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-24, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[0,1,4,5],zmm1[0,1,0,1]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i64> %vec1, <8 x i64> %vec2, <8 x i32> <i32 0, i32 1, i32 4, i32 5, i32 8, i32 9, i32 8, i32 9>
; CHECK-LABEL: test_8xi64_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm2 {%k1} = zmm0[2,3,6,7],zmm1[4,5,2,3]
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xi64_zero_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[2,3,6,7],zmm1[4,5,2,3]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x i64> %vec1, <8 x i64> %vec2, <8 x i32> <i32 2, i32 3, i32 6, i32 7, i32 12, i32 13, i32 10, i32 11>
; CHECK-LABEL: test_8xi64_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm1 {%k1} = zmm0[2,3,2,3],mem[4,5,2,3]
; CHECK-NEXT: vmovdqa64 %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xi64_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[2,3,2,3],mem[4,5,2,3]
; CHECK-NEXT: retq
%vec2 = load <8 x i64>, <8 x i64>* %vec2p
; CHECK-LABEL: test_8xi64_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm1 {%k1} = zmm0[2,3,0,1],mem[0,1,0,1]
; CHECK-NEXT: vmovdqa64 %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xi64_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[2,3,0,1],mem[0,1,0,1]
; CHECK-NEXT: retq
%vec2 = load <8 x i64>, <8 x i64>* %vec2p
; CHECK-LABEL: test_8xi64_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $42, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm1 {%k1} = zmm0[4,5,0,1],mem[2,3,2,3]
; CHECK-NEXT: vmovdqa64 %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xi64_zero_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $42, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[4,5,0,1],mem[2,3,2,3]
; CHECK-NEXT: retq
%vec2 = load <8 x i64>, <8 x i64>* %vec2p
; CHECK-LABEL: test_8xi64_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm1 {%k1} = zmm0[2,3,0,1],mem[6,7,2,3]
; CHECK-NEXT: vmovdqa64 %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xi64_zero_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vshufi64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[2,3,0,1],mem[6,7,2,3]
; CHECK-NEXT: retq
%vec2 = load <8 x i64>, <8 x i64>* %vec2p
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mcpu=skx %s -o - | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl,+avx512bw %s -o - | FileCheck %s
define <16 x i8> @test_16xi8_perm_mask0(<16 x i8> %vec) {
; CHECK-LABEL: test_16xi8_perm_mask0:
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mcpu=skx %s -o - | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl %s -o - | FileCheck %s
define <4 x float> @test_4xfloat_unpack_low_mask0(<4 x float> %vec1, <4 x float> %vec2) {
; CHECK-LABEL: test_4xfloat_unpack_low_mask0:
; CHECK-LABEL: test_4xfloat_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm2 {%k1} = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; CHECK-NEXT: vmovaps %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm0 {%k1} {z} = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec1, <4 x float> %vec2, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
; CHECK-LABEL: test_4xfloat_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm2 {%k1} = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; CHECK-NEXT: vmovaps %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm0 {%k1} {z} = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec1, <4 x float> %vec2, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
; CHECK-LABEL: test_4xfloat_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm2 {%k1} = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; CHECK-NEXT: vmovaps %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm0 {%k1} {z} = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec1, <4 x float> %vec2, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
; CHECK-LABEL: test_4xfloat_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm2 {%k1} = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; CHECK-NEXT: vmovaps %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm0 {%k1} {z} = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec1, <4 x float> %vec2, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
; CHECK-LABEL: test_4xfloat_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm1 {%k1} = xmm0[0],mem[0],xmm0[1],mem[1]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $8, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm0 {%k1} {z} = xmm0[0],mem[0],xmm0[1],mem[1]
; CHECK-NEXT: retq
%vec2 = load <4 x float>, <4 x float>* %vec2p
; CHECK-LABEL: test_4xfloat_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm1 {%k1} = xmm0[0],mem[0],xmm0[1],mem[1]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm0 {%k1} {z} = xmm0[0],mem[0],xmm0[1],mem[1]
; CHECK-NEXT: retq
%vec2 = load <4 x float>, <4 x float>* %vec2p
; CHECK-LABEL: test_4xfloat_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm1 {%k1} = xmm0[0],mem[0],xmm0[1],mem[1]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm0 {%k1} {z} = xmm0[0],mem[0],xmm0[1],mem[1]
; CHECK-NEXT: retq
%vec2 = load <4 x float>, <4 x float>* %vec2p
; CHECK-LABEL: test_4xfloat_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm1 {%k1} = xmm0[0],mem[0],xmm0[1],mem[1]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} xmm0 {%k1} {z} = xmm0[0],mem[0],xmm0[1],mem[1]
; CHECK-NEXT: retq
%vec2 = load <4 x float>, <4 x float>* %vec2p
; CHECK-LABEL: test_8xfloat_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $122, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm2 {%k1} = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5]
; CHECK-NEXT: vmovaps %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $122, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm0 {%k1} {z} = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 4, i32 12, i32 5, i32 13>
; CHECK-LABEL: test_8xfloat_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-107, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm2 {%k1} = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5]
; CHECK-NEXT: vmovaps %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-107, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm0 {%k1} {z} = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 4, i32 12, i32 5, i32 13>
; CHECK-LABEL: test_8xfloat_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-25, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm2 {%k1} = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5]
; CHECK-NEXT: vmovaps %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-25, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm0 {%k1} {z} = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 4, i32 12, i32 5, i32 13>
; CHECK-LABEL: test_8xfloat_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-127, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm2 {%k1} = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5]
; CHECK-NEXT: vmovaps %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-127, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm0 {%k1} {z} = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 4, i32 12, i32 5, i32 13>
; CHECK-LABEL: test_8xfloat_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $72, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm1 {%k1} = ymm0[0],mem[0],ymm0[1],mem[1],ymm0[4],mem[4],ymm0[5],mem[5]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $72, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm0 {%k1} {z} = ymm0[0],mem[0],ymm0[1],mem[1],ymm0[4],mem[4],ymm0[5],mem[5]
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK-LABEL: test_8xfloat_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-64, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm1 {%k1} = ymm0[0],mem[0],ymm0[1],mem[1],ymm0[4],mem[4],ymm0[5],mem[5]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-64, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm0 {%k1} {z} = ymm0[0],mem[0],ymm0[1],mem[1],ymm0[4],mem[4],ymm0[5],mem[5]
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK-LABEL: test_8xfloat_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm1 {%k1} = ymm0[0],mem[0],ymm0[1],mem[1],ymm0[4],mem[4],ymm0[5],mem[5]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-98, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm0 {%k1} {z} = ymm0[0],mem[0],ymm0[1],mem[1],ymm0[4],mem[4],ymm0[5],mem[5]
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK-LABEL: test_8xfloat_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $64, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm1 {%k1} = ymm0[0],mem[0],ymm0[1],mem[1],ymm0[4],mem[4],ymm0[5],mem[5]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $64, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} ymm0 {%k1} {z} = ymm0[0],mem[0],ymm0[1],mem[1],ymm0[4],mem[4],ymm0[5],mem[5]
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-5916, %ax # imm = 0xE8E4
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-5916, %ax # imm = 0xE8E4
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 4, i32 20, i32 5, i32 21, i32 8, i32 24, i32 9, i32 25, i32 12, i32 28, i32 13, i32 29>
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-1130, %ax # imm = 0xFB96
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-1130, %ax # imm = 0xFB96
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 4, i32 20, i32 5, i32 21, i32 8, i32 24, i32 9, i32 25, i32 12, i32 28, i32 13, i32 29>
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-12439, %ax # imm = 0xCF69
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-12439, %ax # imm = 0xCF69
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 4, i32 20, i32 5, i32 21, i32 8, i32 24, i32 9, i32 25, i32 12, i32 28, i32 13, i32 29>
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-6413, %ax # imm = 0xE6F3
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-6413, %ax # imm = 0xE6F3
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 4, i32 20, i32 5, i32 21, i32 8, i32 24, i32 9, i32 25, i32 12, i32 28, i32 13, i32 29>
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $20326, %ax # imm = 0x4F66
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $20326, %ax # imm = 0x4F66
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-17707, %ax # imm = 0xBAD5
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-17707, %ax # imm = 0xBAD5
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-6631, %ax # imm = 0xE619
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-6631, %ax # imm = 0xE619
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-20711, %ax # imm = 0xAF19
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-20711, %ax # imm = 0xAF19
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_2xdouble_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} xmm2 {%k1} = xmm0[0],xmm1[0]
; CHECK-NEXT: vmovapd %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_2xdouble_zero_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} xmm0 {%k1} {z} = xmm0[0],xmm1[0]
; CHECK-NEXT: retq
%shuf = shufflevector <2 x double> %vec1, <2 x double> %vec2, <2 x i32> <i32 0, i32 2>
; CHECK-LABEL: test_2xdouble_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} xmm2 {%k1} = xmm0[0],xmm1[0]
; CHECK-NEXT: vmovapd %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_2xdouble_zero_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} xmm0 {%k1} {z} = xmm0[0],xmm1[0]
; CHECK-NEXT: retq
%shuf = shufflevector <2 x double> %vec1, <2 x double> %vec2, <2 x i32> <i32 0, i32 2>
; CHECK-LABEL: test_2xdouble_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} xmm1 {%k1} = xmm0[0],mem[0]
; CHECK-NEXT: vmovapd %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_2xdouble_zero_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} xmm0 {%k1} {z} = xmm0[0],mem[0]
; CHECK-NEXT: retq
%vec2 = load <2 x double>, <2 x double>* %vec2p
; CHECK-LABEL: test_2xdouble_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} xmm1 {%k1} = xmm0[0],mem[0]
; CHECK-NEXT: vmovapd %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_2xdouble_zero_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} xmm0 {%k1} {z} = xmm0[0],mem[0]
; CHECK-NEXT: retq
%vec2 = load <2 x double>, <2 x double>* %vec2p
; CHECK-LABEL: test_4xdouble_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm2 {%k1} = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
; CHECK-NEXT: vmovapd %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
; CHECK-LABEL: test_4xdouble_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm2 {%k1} = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
; CHECK-NEXT: vmovapd %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
; CHECK-LABEL: test_4xdouble_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm2 {%k1} = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
; CHECK-NEXT: vmovapd %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
; CHECK-LABEL: test_4xdouble_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm2 {%k1} = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
; CHECK-NEXT: vmovapd %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
; CHECK-LABEL: test_4xdouble_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm1 {%k1} = ymm0[0],mem[0],ymm0[2],mem[2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0],mem[0],ymm0[2],mem[2]
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK-LABEL: test_4xdouble_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm1 {%k1} = ymm0[0],mem[0],ymm0[2],mem[2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0],mem[0],ymm0[2],mem[2]
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK-LABEL: test_4xdouble_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm1 {%k1} = ymm0[0],mem[0],ymm0[2],mem[2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0],mem[0],ymm0[2],mem[2]
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK-LABEL: test_4xdouble_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm1 {%k1} = ymm0[0],mem[0],ymm0[2],mem[2]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0],mem[0],ymm0[2],mem[2]
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-73, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-73, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $102, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $102, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-46, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-46, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-86, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-86, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-35, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-35, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $62, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $62, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK-LABEL: test_4xfloat_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm2 {%k1} = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; CHECK-NEXT: vmovaps %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm0 {%k1} {z} = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec1, <4 x float> %vec2, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
; CHECK-LABEL: test_4xfloat_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm2 {%k1} = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; CHECK-NEXT: vmovaps %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm0 {%k1} {z} = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec1, <4 x float> %vec2, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
; CHECK-LABEL: test_4xfloat_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm2 {%k1} = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; CHECK-NEXT: vmovaps %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $3, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm0 {%k1} {z} = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec1, <4 x float> %vec2, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
; CHECK-LABEL: test_4xfloat_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm2 {%k1} = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; CHECK-NEXT: vmovaps %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $7, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm0 {%k1} {z} = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x float> %vec1, <4 x float> %vec2, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
; CHECK-LABEL: test_4xfloat_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm1 {%k1} = xmm0[2],mem[2],xmm0[3],mem[3]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $4, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm0 {%k1} {z} = xmm0[2],mem[2],xmm0[3],mem[3]
; CHECK-NEXT: retq
%vec2 = load <4 x float>, <4 x float>* %vec2p
; CHECK-LABEL: test_4xfloat_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm1 {%k1} = xmm0[2],mem[2],xmm0[3],mem[3]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm0 {%k1} {z} = xmm0[2],mem[2],xmm0[3],mem[3]
; CHECK-NEXT: retq
%vec2 = load <4 x float>, <4 x float>* %vec2p
; CHECK-LABEL: test_4xfloat_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm1 {%k1} = xmm0[2],mem[2],xmm0[3],mem[3]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm0 {%k1} {z} = xmm0[2],mem[2],xmm0[3],mem[3]
; CHECK-NEXT: retq
%vec2 = load <4 x float>, <4 x float>* %vec2p
; CHECK-LABEL: test_4xfloat_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm1 {%k1} = xmm0[2],mem[2],xmm0[3],mem[3]
; CHECK-NEXT: vmovaps %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xfloat_zero_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $5, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} xmm0 {%k1} {z} = xmm0[2],mem[2],xmm0[3],mem[3]
; CHECK-NEXT: retq
%vec2 = load <4 x float>, <4 x float>* %vec2p
; CHECK-LABEL: test_8xfloat_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $21, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm2 {%k1} = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7]
; CHECK-NEXT: vmovaps %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $21, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm0 {%k1} {z} = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 2, i32 10, i32 3, i32 11, i32 6, i32 14, i32 7, i32 15>
; CHECK-LABEL: test_8xfloat_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $82, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm2 {%k1} = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7]
; CHECK-NEXT: vmovaps %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $82, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm0 {%k1} {z} = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 2, i32 10, i32 3, i32 11, i32 6, i32 14, i32 7, i32 15>
; CHECK-LABEL: test_8xfloat_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm2 {%k1} = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7]
; CHECK-NEXT: vmovaps %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-126, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm0 {%k1} {z} = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 2, i32 10, i32 3, i32 11, i32 6, i32 14, i32 7, i32 15>
; CHECK-LABEL: test_8xfloat_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-19, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm2 {%k1} = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7]
; CHECK-NEXT: vmovaps %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-19, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm0 {%k1} {z} = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x float> %vec1, <8 x float> %vec2, <8 x i32> <i32 2, i32 10, i32 3, i32 11, i32 6, i32 14, i32 7, i32 15>
; CHECK-LABEL: test_8xfloat_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $28, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm1 {%k1} = ymm0[2],mem[2],ymm0[3],mem[3],ymm0[6],mem[6],ymm0[7],mem[7]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $28, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm0 {%k1} {z} = ymm0[2],mem[2],ymm0[3],mem[3],ymm0[6],mem[6],ymm0[7],mem[7]
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK-LABEL: test_8xfloat_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-115, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm1 {%k1} = ymm0[2],mem[2],ymm0[3],mem[3],ymm0[6],mem[6],ymm0[7],mem[7]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-115, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm0 {%k1} {z} = ymm0[2],mem[2],ymm0[3],mem[3],ymm0[6],mem[6],ymm0[7],mem[7]
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK-LABEL: test_8xfloat_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-76, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm1 {%k1} = ymm0[2],mem[2],ymm0[3],mem[3],ymm0[6],mem[6],ymm0[7],mem[7]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-76, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm0 {%k1} {z} = ymm0[2],mem[2],ymm0[3],mem[3],ymm0[6],mem[6],ymm0[7],mem[7]
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK-LABEL: test_8xfloat_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-116, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm1 {%k1} = ymm0[2],mem[2],ymm0[3],mem[3],ymm0[6],mem[6],ymm0[7],mem[7]
; CHECK-NEXT: vmovaps %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xfloat_zero_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-116, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} ymm0 {%k1} {z} = ymm0[2],mem[2],ymm0[3],mem[3],ymm0[6],mem[6],ymm0[7],mem[7]
; CHECK-NEXT: retq
%vec2 = load <8 x float>, <8 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-12160, %ax # imm = 0xD080
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm2 {%k1} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-12160, %ax # imm = 0xD080
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 2, i32 18, i32 3, i32 19, i32 6, i32 22, i32 7, i32 23, i32 10, i32 26, i32 11, i32 27, i32 14, i32 30, i32 15, i32 31>
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-30129, %ax # imm = 0x8A4F
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm2 {%k1} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-30129, %ax # imm = 0x8A4F
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 2, i32 18, i32 3, i32 19, i32 6, i32 22, i32 7, i32 23, i32 10, i32 26, i32 11, i32 27, i32 14, i32 30, i32 15, i32 31>
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-2371, %ax # imm = 0xF6BD
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm2 {%k1} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-2371, %ax # imm = 0xF6BD
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 2, i32 18, i32 3, i32 19, i32 6, i32 22, i32 7, i32 23, i32 10, i32 26, i32 11, i32 27, i32 14, i32 30, i32 15, i32 31>
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-26006, %ax # imm = 0x9A6A
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm2 {%k1} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-26006, %ax # imm = 0x9A6A
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: retq
%shuf = shufflevector <16 x float> %vec1, <16 x float> %vec2, <16 x i32> <i32 2, i32 18, i32 3, i32 19, i32 6, i32 22, i32 7, i32 23, i32 10, i32 26, i32 11, i32 27, i32 14, i32 30, i32 15, i32 31>
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-27027, %ax # imm = 0x966D
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm1 {%k1} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-27027, %ax # imm = 0x966D
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $29162, %ax # imm = 0x71EA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm1 {%k1} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movw $29162, %ax # imm = 0x71EA
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-26458, %ax # imm = 0x98A6
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm1 {%k1} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movw $-26458, %ax # imm = 0x98A6
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $25225, %ax # imm = 0x6289
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm1 {%k1} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movw $25225, %ax # imm = 0x6289
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: retq
%vec2 = load <16 x float>, <16 x float>* %vec2p
; CHECK-LABEL: test_2xdouble_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} xmm2 {%k1} = xmm0[1],xmm1[1]
; CHECK-NEXT: vmovapd %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_2xdouble_zero_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} xmm0 {%k1} {z} = xmm0[1],xmm1[1]
; CHECK-NEXT: retq
%shuf = shufflevector <2 x double> %vec1, <2 x double> %vec2, <2 x i32> <i32 1, i32 3>
; CHECK-LABEL: test_2xdouble_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} xmm2 {%k1} = xmm0[1],xmm1[1]
; CHECK-NEXT: vmovapd %xmm2, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_2xdouble_zero_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} xmm0 {%k1} {z} = xmm0[1],xmm1[1]
; CHECK-NEXT: retq
%shuf = shufflevector <2 x double> %vec1, <2 x double> %vec2, <2 x i32> <i32 1, i32 3>
; CHECK-LABEL: test_2xdouble_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} xmm1 {%k1} = xmm0[1],mem[1]
; CHECK-NEXT: vmovapd %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_2xdouble_zero_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} xmm0 {%k1} {z} = xmm0[1],mem[1]
; CHECK-NEXT: retq
%vec2 = load <2 x double>, <2 x double>* %vec2p
; CHECK-LABEL: test_2xdouble_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} xmm1 {%k1} = xmm0[1],mem[1]
; CHECK-NEXT: vmovapd %xmm1, %xmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_2xdouble_zero_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $2, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} xmm0 {%k1} {z} = xmm0[1],mem[1]
; CHECK-NEXT: retq
%vec2 = load <2 x double>, <2 x double>* %vec2p
; CHECK-LABEL: test_4xdouble_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $9, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm2 {%k1} = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
; CHECK-NEXT: vmovapd %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $9, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
; CHECK-LABEL: test_4xdouble_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm2 {%k1} = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
; CHECK-NEXT: vmovapd %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $14, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
; CHECK-LABEL: test_4xdouble_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm2 {%k1} = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
; CHECK-NEXT: vmovapd %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $6, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
; CHECK-LABEL: test_4xdouble_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm2 {%k1} = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
; CHECK-NEXT: vmovapd %ymm2, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $1, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
; CHECK-NEXT: retq
%shuf = shufflevector <4 x double> %vec1, <4 x double> %vec2, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
; CHECK-LABEL: test_4xdouble_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm1 {%k1} = ymm0[1],mem[1],ymm0[3],mem[3]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $11, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1],mem[1],ymm0[3],mem[3]
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK-LABEL: test_4xdouble_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm1 {%k1} = ymm0[1],mem[1],ymm0[3],mem[3]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $12, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1],mem[1],ymm0[3],mem[3]
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK-LABEL: test_4xdouble_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm1 {%k1} = ymm0[1],mem[1],ymm0[3],mem[3]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $13, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1],mem[1],ymm0[3],mem[3]
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK-LABEL: test_4xdouble_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm1 {%k1} = ymm0[1],mem[1],ymm0[3],mem[3]
; CHECK-NEXT: vmovapd %ymm1, %ymm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_4xdouble_zero_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $10, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm0 {%k1} {z} = ymm0[1],mem[1],ymm0[3],mem[3]
; CHECK-NEXT: retq
%vec2 = load <4 x double>, <4 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-27, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm2 {%k1} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-27, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-21, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm2 {%k1} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-21, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-118, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm2 {%k1} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-118, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $100, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm2 {%k1} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $100, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: retq
%shuf = shufflevector <8 x double> %vec1, <8 x double> %vec2, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-76, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-76, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $71, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: movb $71, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-49, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-49, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-40, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
; CHECK-NEXT: retq
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: movb $-40, %al
-; CHECK-NEXT: kmovd %eax, %k1
+; CHECK-NEXT: kmovw %eax, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: retq
%vec2 = load <8 x double>, <8 x double>* %vec2p