--- /dev/null
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=thumbv8.1m.main-arm-none-eabi -mattr=+mve,+fullfp16 -verify-machineinstrs %s -o - | FileCheck %s --check-prefix=CHECK --check-prefix=CHECK-MVE
+; RUN: llc -mtriple=thumbv8.1m.main-arm-none-eabi -mattr=+mve.fp -verify-machineinstrs %s -o - | FileCheck %s --check-prefix=CHECK --check-prefix=CHECK-MVEFP
+
+define arm_aapcs_vfpcc <4 x float> @sqrt_float32_t(<4 x float> %src) {
+; CHECK-MVE-LABEL: sqrt_float32_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: vsqrt.f32 s4, s0
+; CHECK-MVE-NEXT: movs r0, #0
+; CHECK-MVE-NEXT: vsqrt.f32 s6, s1
+; CHECK-MVE-NEXT: vsqrt.f32 s8, s2
+; CHECK-MVE-NEXT: vsqrt.f32 s10, s3
+; CHECK-MVE-NEXT: vdup.32 q0, r0
+; CHECK-MVE-NEXT: vmov.f32 s0, s4
+; CHECK-MVE-NEXT: vmov.f32 s1, s6
+; CHECK-MVE-NEXT: vmov.f32 s2, s8
+; CHECK-MVE-NEXT: vmov.f32 s3, s10
+; CHECK-MVE-NEXT: bx lr
+;
+; CHECK-MVEFP-LABEL: sqrt_float32_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: vsqrt.f32 s4, s0
+; CHECK-MVEFP-NEXT: vsqrt.f32 s5, s1
+; CHECK-MVEFP-NEXT: vsqrt.f32 s6, s2
+; CHECK-MVEFP-NEXT: vsqrt.f32 s7, s3
+; CHECK-MVEFP-NEXT: vmov q0, q1
+; CHECK-MVEFP-NEXT: bx lr
+entry:
+ %0 = call fast <4 x float> @llvm.sqrt.v4f32(<4 x float> %src)
+ ret <4 x float> %0
+}
+
+define arm_aapcs_vfpcc <8 x half> @sqrt_float16_t(<8 x half> %src) {
+; CHECK-MVE-LABEL: sqrt_float16_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[0]
+; CHECK-MVE-NEXT: vmov.u16 r1, q0[1]
+; CHECK-MVE-NEXT: vmov s4, r0
+; CHECK-MVE-NEXT: movs r2, #0
+; CHECK-MVE-NEXT: vsqrt.f16 s4, s4
+; CHECK-MVE-NEXT: vmov r0, s4
+; CHECK-MVE-NEXT: vmov s4, r1
+; CHECK-MVE-NEXT: vsqrt.f16 s4, s4
+; CHECK-MVE-NEXT: vmov r1, s4
+; CHECK-MVE-NEXT: vdup.16 q1, r2
+; CHECK-MVE-NEXT: vmov.16 q1[0], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[2]
+; CHECK-MVE-NEXT: vmov s8, r0
+; CHECK-MVE-NEXT: vmov.16 q1[1], r1
+; CHECK-MVE-NEXT: vsqrt.f16 s8, s8
+; CHECK-MVE-NEXT: vmov r0, s8
+; CHECK-MVE-NEXT: vmov.16 q1[2], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[3]
+; CHECK-MVE-NEXT: vmov s8, r0
+; CHECK-MVE-NEXT: vsqrt.f16 s8, s8
+; CHECK-MVE-NEXT: vmov r0, s8
+; CHECK-MVE-NEXT: vmov.16 q1[3], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[4]
+; CHECK-MVE-NEXT: vmov s8, r0
+; CHECK-MVE-NEXT: vsqrt.f16 s8, s8
+; CHECK-MVE-NEXT: vmov r0, s8
+; CHECK-MVE-NEXT: vmov.16 q1[4], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[5]
+; CHECK-MVE-NEXT: vmov s8, r0
+; CHECK-MVE-NEXT: vsqrt.f16 s8, s8
+; CHECK-MVE-NEXT: vmov r0, s8
+; CHECK-MVE-NEXT: vmov.16 q1[5], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[6]
+; CHECK-MVE-NEXT: vmov s8, r0
+; CHECK-MVE-NEXT: vsqrt.f16 s8, s8
+; CHECK-MVE-NEXT: vmov r0, s8
+; CHECK-MVE-NEXT: vmov.16 q1[6], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[7]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vsqrt.f16 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q1[7], r0
+; CHECK-MVE-NEXT: vmov q0, q1
+; CHECK-MVE-NEXT: bx lr
+;
+; CHECK-MVEFP-LABEL: sqrt_float16_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[0]
+; CHECK-MVEFP-NEXT: vmov.u16 r1, q0[1]
+; CHECK-MVEFP-NEXT: vmov s4, r0
+; CHECK-MVEFP-NEXT: vsqrt.f16 s4, s4
+; CHECK-MVEFP-NEXT: vmov r0, s4
+; CHECK-MVEFP-NEXT: vmov s4, r1
+; CHECK-MVEFP-NEXT: vsqrt.f16 s4, s4
+; CHECK-MVEFP-NEXT: vmov r1, s4
+; CHECK-MVEFP-NEXT: vmov.16 q1[0], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[2]
+; CHECK-MVEFP-NEXT: vmov.16 q1[1], r1
+; CHECK-MVEFP-NEXT: vmov s8, r0
+; CHECK-MVEFP-NEXT: vsqrt.f16 s8, s8
+; CHECK-MVEFP-NEXT: vmov r0, s8
+; CHECK-MVEFP-NEXT: vmov.16 q1[2], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[3]
+; CHECK-MVEFP-NEXT: vmov s8, r0
+; CHECK-MVEFP-NEXT: vsqrt.f16 s8, s8
+; CHECK-MVEFP-NEXT: vmov r0, s8
+; CHECK-MVEFP-NEXT: vmov.16 q1[3], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[4]
+; CHECK-MVEFP-NEXT: vmov s8, r0
+; CHECK-MVEFP-NEXT: vsqrt.f16 s8, s8
+; CHECK-MVEFP-NEXT: vmov r0, s8
+; CHECK-MVEFP-NEXT: vmov.16 q1[4], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[5]
+; CHECK-MVEFP-NEXT: vmov s8, r0
+; CHECK-MVEFP-NEXT: vsqrt.f16 s8, s8
+; CHECK-MVEFP-NEXT: vmov r0, s8
+; CHECK-MVEFP-NEXT: vmov.16 q1[5], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[6]
+; CHECK-MVEFP-NEXT: vmov s8, r0
+; CHECK-MVEFP-NEXT: vsqrt.f16 s8, s8
+; CHECK-MVEFP-NEXT: vmov r0, s8
+; CHECK-MVEFP-NEXT: vmov.16 q1[6], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[7]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vsqrt.f16 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q1[7], r0
+; CHECK-MVEFP-NEXT: vmov q0, q1
+; CHECK-MVEFP-NEXT: bx lr
+entry:
+ %0 = call fast <8 x half> @llvm.sqrt.v8f16(<8 x half> %src)
+ ret <8 x half> %0
+}
+
+define arm_aapcs_vfpcc <4 x float> @cos_float32_t(<4 x float> %src) {
+; CHECK-MVE-LABEL: cos_float32_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10}
+; CHECK-MVE-NEXT: .pad #16
+; CHECK-MVE-NEXT: sub sp, #16
+; CHECK-MVE-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl cosf
+; CHECK-MVE-NEXT: mov r4, r0
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl cosf
+; CHECK-MVE-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVE-NEXT: vmov s16, r0
+; CHECK-MVE-NEXT: vmov s18, r4
+; CHECK-MVE-NEXT: mov r0, r1
+; CHECK-MVE-NEXT: bl cosf
+; CHECK-MVE-NEXT: vmov s20, r0
+; CHECK-MVE-NEXT: mov r0, r5
+; CHECK-MVE-NEXT: bl cosf
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vmov s4, r0
+; CHECK-MVE-NEXT: vdup.32 q0, r1
+; CHECK-MVE-NEXT: vmov.f32 s0, s16
+; CHECK-MVE-NEXT: vmov.f32 s1, s18
+; CHECK-MVE-NEXT: vmov.f32 s2, s20
+; CHECK-MVE-NEXT: vmov.f32 s3, s4
+; CHECK-MVE-NEXT: add sp, #16
+; CHECK-MVE-NEXT: vpop {d8, d9, d10}
+; CHECK-MVE-NEXT: pop {r4, r5, r7, pc}
+;
+; CHECK-MVEFP-LABEL: cos_float32_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9}
+; CHECK-MVEFP-NEXT: vpush {d8, d9}
+; CHECK-MVEFP-NEXT: .pad #16
+; CHECK-MVEFP-NEXT: sub sp, #16
+; CHECK-MVEFP-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl cosf
+; CHECK-MVEFP-NEXT: mov r4, r0
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl cosf
+; CHECK-MVEFP-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVEFP-NEXT: vmov s16, r0
+; CHECK-MVEFP-NEXT: vmov s17, r4
+; CHECK-MVEFP-NEXT: mov r0, r1
+; CHECK-MVEFP-NEXT: bl cosf
+; CHECK-MVEFP-NEXT: vmov s18, r0
+; CHECK-MVEFP-NEXT: mov r0, r5
+; CHECK-MVEFP-NEXT: bl cosf
+; CHECK-MVEFP-NEXT: vmov s19, r0
+; CHECK-MVEFP-NEXT: vmov q0, q4
+; CHECK-MVEFP-NEXT: add sp, #16
+; CHECK-MVEFP-NEXT: vpop {d8, d9}
+; CHECK-MVEFP-NEXT: pop {r4, r5, r7, pc}
+entry:
+ %0 = call fast <4 x float> @llvm.cos.v4f32(<4 x float> %src)
+ ret <4 x float> %0
+}
+
+define arm_aapcs_vfpcc <8 x half> @cos_float16_t(<8 x half> %src) {
+; CHECK-MVE-LABEL: cos_float16_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, lr}
+; CHECK-MVE-NEXT: push {r4, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: .pad #32
+; CHECK-MVE-NEXT: sub sp, #32
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVE-NEXT: vmov q4, q0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #24]
+; CHECK-MVE-NEXT: ldr r0, [sp, #24]
+; CHECK-MVE-NEXT: bl cosf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r4, s0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #28]
+; CHECK-MVE-NEXT: ldr r0, [sp, #28]
+; CHECK-MVE-NEXT: bl cosf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vdup.16 q5, r1
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[0], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.16 q5[1], r4
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #20]
+; CHECK-MVE-NEXT: ldr r0, [sp, #20]
+; CHECK-MVE-NEXT: bl cosf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[2], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #16]
+; CHECK-MVE-NEXT: ldr r0, [sp, #16]
+; CHECK-MVE-NEXT: bl cosf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[3], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #12]
+; CHECK-MVE-NEXT: ldr r0, [sp, #12]
+; CHECK-MVE-NEXT: bl cosf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[4], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #8]
+; CHECK-MVE-NEXT: ldr r0, [sp, #8]
+; CHECK-MVE-NEXT: bl cosf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[5], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #4]
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl cosf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[6], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp]
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl cosf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[7], r0
+; CHECK-MVE-NEXT: vmov q0, q5
+; CHECK-MVE-NEXT: add sp, #32
+; CHECK-MVE-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: pop {r4, pc}
+;
+; CHECK-MVEFP-LABEL: cos_float16_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, lr}
+; CHECK-MVEFP-NEXT: push {r4, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: .pad #32
+; CHECK-MVEFP-NEXT: sub sp, #32
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVEFP-NEXT: vmov q4, q0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #24]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #24]
+; CHECK-MVEFP-NEXT: bl cosf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r4, s0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #28]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #28]
+; CHECK-MVEFP-NEXT: bl cosf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[0], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.16 q5[1], r4
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #20]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #20]
+; CHECK-MVEFP-NEXT: bl cosf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[2], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #16]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #16]
+; CHECK-MVEFP-NEXT: bl cosf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[3], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #12]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #12]
+; CHECK-MVEFP-NEXT: bl cosf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[4], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #8]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #8]
+; CHECK-MVEFP-NEXT: bl cosf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[5], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #4]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl cosf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[6], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp]
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl cosf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[7], r0
+; CHECK-MVEFP-NEXT: vmov q0, q5
+; CHECK-MVEFP-NEXT: add sp, #32
+; CHECK-MVEFP-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: pop {r4, pc}
+entry:
+ %0 = call fast <8 x half> @llvm.cos.v8f16(<8 x half> %src)
+ ret <8 x half> %0
+}
+
+define arm_aapcs_vfpcc <4 x float> @sin_float32_t(<4 x float> %src) {
+; CHECK-MVE-LABEL: sin_float32_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10}
+; CHECK-MVE-NEXT: .pad #16
+; CHECK-MVE-NEXT: sub sp, #16
+; CHECK-MVE-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl sinf
+; CHECK-MVE-NEXT: mov r4, r0
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl sinf
+; CHECK-MVE-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVE-NEXT: vmov s16, r0
+; CHECK-MVE-NEXT: vmov s18, r4
+; CHECK-MVE-NEXT: mov r0, r1
+; CHECK-MVE-NEXT: bl sinf
+; CHECK-MVE-NEXT: vmov s20, r0
+; CHECK-MVE-NEXT: mov r0, r5
+; CHECK-MVE-NEXT: bl sinf
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vmov s4, r0
+; CHECK-MVE-NEXT: vdup.32 q0, r1
+; CHECK-MVE-NEXT: vmov.f32 s0, s16
+; CHECK-MVE-NEXT: vmov.f32 s1, s18
+; CHECK-MVE-NEXT: vmov.f32 s2, s20
+; CHECK-MVE-NEXT: vmov.f32 s3, s4
+; CHECK-MVE-NEXT: add sp, #16
+; CHECK-MVE-NEXT: vpop {d8, d9, d10}
+; CHECK-MVE-NEXT: pop {r4, r5, r7, pc}
+;
+; CHECK-MVEFP-LABEL: sin_float32_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9}
+; CHECK-MVEFP-NEXT: vpush {d8, d9}
+; CHECK-MVEFP-NEXT: .pad #16
+; CHECK-MVEFP-NEXT: sub sp, #16
+; CHECK-MVEFP-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl sinf
+; CHECK-MVEFP-NEXT: mov r4, r0
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl sinf
+; CHECK-MVEFP-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVEFP-NEXT: vmov s16, r0
+; CHECK-MVEFP-NEXT: vmov s17, r4
+; CHECK-MVEFP-NEXT: mov r0, r1
+; CHECK-MVEFP-NEXT: bl sinf
+; CHECK-MVEFP-NEXT: vmov s18, r0
+; CHECK-MVEFP-NEXT: mov r0, r5
+; CHECK-MVEFP-NEXT: bl sinf
+; CHECK-MVEFP-NEXT: vmov s19, r0
+; CHECK-MVEFP-NEXT: vmov q0, q4
+; CHECK-MVEFP-NEXT: add sp, #16
+; CHECK-MVEFP-NEXT: vpop {d8, d9}
+; CHECK-MVEFP-NEXT: pop {r4, r5, r7, pc}
+entry:
+ %0 = call fast <4 x float> @llvm.sin.v4f32(<4 x float> %src)
+ ret <4 x float> %0
+}
+
+define arm_aapcs_vfpcc <8 x half> @sin_float16_t(<8 x half> %src) {
+; CHECK-MVE-LABEL: sin_float16_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, lr}
+; CHECK-MVE-NEXT: push {r4, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: .pad #32
+; CHECK-MVE-NEXT: sub sp, #32
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVE-NEXT: vmov q4, q0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #24]
+; CHECK-MVE-NEXT: ldr r0, [sp, #24]
+; CHECK-MVE-NEXT: bl sinf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r4, s0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #28]
+; CHECK-MVE-NEXT: ldr r0, [sp, #28]
+; CHECK-MVE-NEXT: bl sinf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vdup.16 q5, r1
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[0], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.16 q5[1], r4
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #20]
+; CHECK-MVE-NEXT: ldr r0, [sp, #20]
+; CHECK-MVE-NEXT: bl sinf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[2], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #16]
+; CHECK-MVE-NEXT: ldr r0, [sp, #16]
+; CHECK-MVE-NEXT: bl sinf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[3], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #12]
+; CHECK-MVE-NEXT: ldr r0, [sp, #12]
+; CHECK-MVE-NEXT: bl sinf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[4], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #8]
+; CHECK-MVE-NEXT: ldr r0, [sp, #8]
+; CHECK-MVE-NEXT: bl sinf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[5], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #4]
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl sinf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[6], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp]
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl sinf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[7], r0
+; CHECK-MVE-NEXT: vmov q0, q5
+; CHECK-MVE-NEXT: add sp, #32
+; CHECK-MVE-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: pop {r4, pc}
+;
+; CHECK-MVEFP-LABEL: sin_float16_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, lr}
+; CHECK-MVEFP-NEXT: push {r4, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: .pad #32
+; CHECK-MVEFP-NEXT: sub sp, #32
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVEFP-NEXT: vmov q4, q0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #24]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #24]
+; CHECK-MVEFP-NEXT: bl sinf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r4, s0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #28]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #28]
+; CHECK-MVEFP-NEXT: bl sinf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[0], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.16 q5[1], r4
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #20]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #20]
+; CHECK-MVEFP-NEXT: bl sinf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[2], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #16]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #16]
+; CHECK-MVEFP-NEXT: bl sinf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[3], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #12]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #12]
+; CHECK-MVEFP-NEXT: bl sinf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[4], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #8]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #8]
+; CHECK-MVEFP-NEXT: bl sinf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[5], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #4]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl sinf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[6], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp]
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl sinf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[7], r0
+; CHECK-MVEFP-NEXT: vmov q0, q5
+; CHECK-MVEFP-NEXT: add sp, #32
+; CHECK-MVEFP-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: pop {r4, pc}
+entry:
+ %0 = call fast <8 x half> @llvm.sin.v8f16(<8 x half> %src)
+ ret <8 x half> %0
+}
+
+define arm_aapcs_vfpcc <4 x float> @exp_float32_t(<4 x float> %src) {
+; CHECK-MVE-LABEL: exp_float32_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10}
+; CHECK-MVE-NEXT: .pad #16
+; CHECK-MVE-NEXT: sub sp, #16
+; CHECK-MVE-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl expf
+; CHECK-MVE-NEXT: mov r4, r0
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl expf
+; CHECK-MVE-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVE-NEXT: vmov s16, r0
+; CHECK-MVE-NEXT: vmov s18, r4
+; CHECK-MVE-NEXT: mov r0, r1
+; CHECK-MVE-NEXT: bl expf
+; CHECK-MVE-NEXT: vmov s20, r0
+; CHECK-MVE-NEXT: mov r0, r5
+; CHECK-MVE-NEXT: bl expf
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vmov s4, r0
+; CHECK-MVE-NEXT: vdup.32 q0, r1
+; CHECK-MVE-NEXT: vmov.f32 s0, s16
+; CHECK-MVE-NEXT: vmov.f32 s1, s18
+; CHECK-MVE-NEXT: vmov.f32 s2, s20
+; CHECK-MVE-NEXT: vmov.f32 s3, s4
+; CHECK-MVE-NEXT: add sp, #16
+; CHECK-MVE-NEXT: vpop {d8, d9, d10}
+; CHECK-MVE-NEXT: pop {r4, r5, r7, pc}
+;
+; CHECK-MVEFP-LABEL: exp_float32_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9}
+; CHECK-MVEFP-NEXT: vpush {d8, d9}
+; CHECK-MVEFP-NEXT: .pad #16
+; CHECK-MVEFP-NEXT: sub sp, #16
+; CHECK-MVEFP-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl expf
+; CHECK-MVEFP-NEXT: mov r4, r0
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl expf
+; CHECK-MVEFP-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVEFP-NEXT: vmov s16, r0
+; CHECK-MVEFP-NEXT: vmov s17, r4
+; CHECK-MVEFP-NEXT: mov r0, r1
+; CHECK-MVEFP-NEXT: bl expf
+; CHECK-MVEFP-NEXT: vmov s18, r0
+; CHECK-MVEFP-NEXT: mov r0, r5
+; CHECK-MVEFP-NEXT: bl expf
+; CHECK-MVEFP-NEXT: vmov s19, r0
+; CHECK-MVEFP-NEXT: vmov q0, q4
+; CHECK-MVEFP-NEXT: add sp, #16
+; CHECK-MVEFP-NEXT: vpop {d8, d9}
+; CHECK-MVEFP-NEXT: pop {r4, r5, r7, pc}
+entry:
+ %0 = call fast <4 x float> @llvm.exp.v4f32(<4 x float> %src)
+ ret <4 x float> %0
+}
+
+define arm_aapcs_vfpcc <8 x half> @exp_float16_t(<8 x half> %src) {
+; CHECK-MVE-LABEL: exp_float16_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, lr}
+; CHECK-MVE-NEXT: push {r4, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: .pad #32
+; CHECK-MVE-NEXT: sub sp, #32
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVE-NEXT: vmov q4, q0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #24]
+; CHECK-MVE-NEXT: ldr r0, [sp, #24]
+; CHECK-MVE-NEXT: bl expf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r4, s0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #28]
+; CHECK-MVE-NEXT: ldr r0, [sp, #28]
+; CHECK-MVE-NEXT: bl expf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vdup.16 q5, r1
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[0], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.16 q5[1], r4
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #20]
+; CHECK-MVE-NEXT: ldr r0, [sp, #20]
+; CHECK-MVE-NEXT: bl expf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[2], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #16]
+; CHECK-MVE-NEXT: ldr r0, [sp, #16]
+; CHECK-MVE-NEXT: bl expf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[3], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #12]
+; CHECK-MVE-NEXT: ldr r0, [sp, #12]
+; CHECK-MVE-NEXT: bl expf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[4], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #8]
+; CHECK-MVE-NEXT: ldr r0, [sp, #8]
+; CHECK-MVE-NEXT: bl expf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[5], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #4]
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl expf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[6], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp]
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl expf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[7], r0
+; CHECK-MVE-NEXT: vmov q0, q5
+; CHECK-MVE-NEXT: add sp, #32
+; CHECK-MVE-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: pop {r4, pc}
+;
+; CHECK-MVEFP-LABEL: exp_float16_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, lr}
+; CHECK-MVEFP-NEXT: push {r4, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: .pad #32
+; CHECK-MVEFP-NEXT: sub sp, #32
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVEFP-NEXT: vmov q4, q0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #24]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #24]
+; CHECK-MVEFP-NEXT: bl expf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r4, s0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #28]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #28]
+; CHECK-MVEFP-NEXT: bl expf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[0], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.16 q5[1], r4
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #20]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #20]
+; CHECK-MVEFP-NEXT: bl expf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[2], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #16]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #16]
+; CHECK-MVEFP-NEXT: bl expf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[3], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #12]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #12]
+; CHECK-MVEFP-NEXT: bl expf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[4], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #8]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #8]
+; CHECK-MVEFP-NEXT: bl expf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[5], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #4]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl expf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[6], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp]
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl expf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[7], r0
+; CHECK-MVEFP-NEXT: vmov q0, q5
+; CHECK-MVEFP-NEXT: add sp, #32
+; CHECK-MVEFP-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: pop {r4, pc}
+entry:
+ %0 = call fast <8 x half> @llvm.exp.v8f16(<8 x half> %src)
+ ret <8 x half> %0
+}
+
+define arm_aapcs_vfpcc <4 x float> @exp2_float32_t(<4 x float> %src) {
+; CHECK-MVE-LABEL: exp2_float32_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10}
+; CHECK-MVE-NEXT: .pad #16
+; CHECK-MVE-NEXT: sub sp, #16
+; CHECK-MVE-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl exp2f
+; CHECK-MVE-NEXT: mov r4, r0
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl exp2f
+; CHECK-MVE-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVE-NEXT: vmov s16, r0
+; CHECK-MVE-NEXT: vmov s18, r4
+; CHECK-MVE-NEXT: mov r0, r1
+; CHECK-MVE-NEXT: bl exp2f
+; CHECK-MVE-NEXT: vmov s20, r0
+; CHECK-MVE-NEXT: mov r0, r5
+; CHECK-MVE-NEXT: bl exp2f
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vmov s4, r0
+; CHECK-MVE-NEXT: vdup.32 q0, r1
+; CHECK-MVE-NEXT: vmov.f32 s0, s16
+; CHECK-MVE-NEXT: vmov.f32 s1, s18
+; CHECK-MVE-NEXT: vmov.f32 s2, s20
+; CHECK-MVE-NEXT: vmov.f32 s3, s4
+; CHECK-MVE-NEXT: add sp, #16
+; CHECK-MVE-NEXT: vpop {d8, d9, d10}
+; CHECK-MVE-NEXT: pop {r4, r5, r7, pc}
+;
+; CHECK-MVEFP-LABEL: exp2_float32_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9}
+; CHECK-MVEFP-NEXT: vpush {d8, d9}
+; CHECK-MVEFP-NEXT: .pad #16
+; CHECK-MVEFP-NEXT: sub sp, #16
+; CHECK-MVEFP-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl exp2f
+; CHECK-MVEFP-NEXT: mov r4, r0
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl exp2f
+; CHECK-MVEFP-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVEFP-NEXT: vmov s16, r0
+; CHECK-MVEFP-NEXT: vmov s17, r4
+; CHECK-MVEFP-NEXT: mov r0, r1
+; CHECK-MVEFP-NEXT: bl exp2f
+; CHECK-MVEFP-NEXT: vmov s18, r0
+; CHECK-MVEFP-NEXT: mov r0, r5
+; CHECK-MVEFP-NEXT: bl exp2f
+; CHECK-MVEFP-NEXT: vmov s19, r0
+; CHECK-MVEFP-NEXT: vmov q0, q4
+; CHECK-MVEFP-NEXT: add sp, #16
+; CHECK-MVEFP-NEXT: vpop {d8, d9}
+; CHECK-MVEFP-NEXT: pop {r4, r5, r7, pc}
+entry:
+ %0 = call fast <4 x float> @llvm.exp2.v4f32(<4 x float> %src)
+ ret <4 x float> %0
+}
+
+define arm_aapcs_vfpcc <8 x half> @exp2_float16_t(<8 x half> %src) {
+; CHECK-MVE-LABEL: exp2_float16_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, lr}
+; CHECK-MVE-NEXT: push {r4, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: .pad #32
+; CHECK-MVE-NEXT: sub sp, #32
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVE-NEXT: vmov q4, q0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #24]
+; CHECK-MVE-NEXT: ldr r0, [sp, #24]
+; CHECK-MVE-NEXT: bl exp2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r4, s0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #28]
+; CHECK-MVE-NEXT: ldr r0, [sp, #28]
+; CHECK-MVE-NEXT: bl exp2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vdup.16 q5, r1
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[0], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.16 q5[1], r4
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #20]
+; CHECK-MVE-NEXT: ldr r0, [sp, #20]
+; CHECK-MVE-NEXT: bl exp2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[2], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #16]
+; CHECK-MVE-NEXT: ldr r0, [sp, #16]
+; CHECK-MVE-NEXT: bl exp2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[3], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #12]
+; CHECK-MVE-NEXT: ldr r0, [sp, #12]
+; CHECK-MVE-NEXT: bl exp2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[4], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #8]
+; CHECK-MVE-NEXT: ldr r0, [sp, #8]
+; CHECK-MVE-NEXT: bl exp2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[5], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #4]
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl exp2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[6], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp]
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl exp2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[7], r0
+; CHECK-MVE-NEXT: vmov q0, q5
+; CHECK-MVE-NEXT: add sp, #32
+; CHECK-MVE-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: pop {r4, pc}
+;
+; CHECK-MVEFP-LABEL: exp2_float16_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, lr}
+; CHECK-MVEFP-NEXT: push {r4, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: .pad #32
+; CHECK-MVEFP-NEXT: sub sp, #32
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVEFP-NEXT: vmov q4, q0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #24]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #24]
+; CHECK-MVEFP-NEXT: bl exp2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r4, s0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #28]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #28]
+; CHECK-MVEFP-NEXT: bl exp2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[0], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.16 q5[1], r4
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #20]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #20]
+; CHECK-MVEFP-NEXT: bl exp2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[2], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #16]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #16]
+; CHECK-MVEFP-NEXT: bl exp2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[3], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #12]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #12]
+; CHECK-MVEFP-NEXT: bl exp2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[4], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #8]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #8]
+; CHECK-MVEFP-NEXT: bl exp2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[5], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #4]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl exp2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[6], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp]
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl exp2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[7], r0
+; CHECK-MVEFP-NEXT: vmov q0, q5
+; CHECK-MVEFP-NEXT: add sp, #32
+; CHECK-MVEFP-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: pop {r4, pc}
+entry:
+ %0 = call fast <8 x half> @llvm.exp2.v8f16(<8 x half> %src)
+ ret <8 x half> %0
+}
+
+define arm_aapcs_vfpcc <4 x float> @log_float32_t(<4 x float> %src) {
+; CHECK-MVE-LABEL: log_float32_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10}
+; CHECK-MVE-NEXT: .pad #16
+; CHECK-MVE-NEXT: sub sp, #16
+; CHECK-MVE-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl logf
+; CHECK-MVE-NEXT: mov r4, r0
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl logf
+; CHECK-MVE-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVE-NEXT: vmov s16, r0
+; CHECK-MVE-NEXT: vmov s18, r4
+; CHECK-MVE-NEXT: mov r0, r1
+; CHECK-MVE-NEXT: bl logf
+; CHECK-MVE-NEXT: vmov s20, r0
+; CHECK-MVE-NEXT: mov r0, r5
+; CHECK-MVE-NEXT: bl logf
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vmov s4, r0
+; CHECK-MVE-NEXT: vdup.32 q0, r1
+; CHECK-MVE-NEXT: vmov.f32 s0, s16
+; CHECK-MVE-NEXT: vmov.f32 s1, s18
+; CHECK-MVE-NEXT: vmov.f32 s2, s20
+; CHECK-MVE-NEXT: vmov.f32 s3, s4
+; CHECK-MVE-NEXT: add sp, #16
+; CHECK-MVE-NEXT: vpop {d8, d9, d10}
+; CHECK-MVE-NEXT: pop {r4, r5, r7, pc}
+;
+; CHECK-MVEFP-LABEL: log_float32_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9}
+; CHECK-MVEFP-NEXT: vpush {d8, d9}
+; CHECK-MVEFP-NEXT: .pad #16
+; CHECK-MVEFP-NEXT: sub sp, #16
+; CHECK-MVEFP-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl logf
+; CHECK-MVEFP-NEXT: mov r4, r0
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl logf
+; CHECK-MVEFP-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVEFP-NEXT: vmov s16, r0
+; CHECK-MVEFP-NEXT: vmov s17, r4
+; CHECK-MVEFP-NEXT: mov r0, r1
+; CHECK-MVEFP-NEXT: bl logf
+; CHECK-MVEFP-NEXT: vmov s18, r0
+; CHECK-MVEFP-NEXT: mov r0, r5
+; CHECK-MVEFP-NEXT: bl logf
+; CHECK-MVEFP-NEXT: vmov s19, r0
+; CHECK-MVEFP-NEXT: vmov q0, q4
+; CHECK-MVEFP-NEXT: add sp, #16
+; CHECK-MVEFP-NEXT: vpop {d8, d9}
+; CHECK-MVEFP-NEXT: pop {r4, r5, r7, pc}
+entry:
+ %0 = call fast <4 x float> @llvm.log.v4f32(<4 x float> %src)
+ ret <4 x float> %0
+}
+
+define arm_aapcs_vfpcc <8 x half> @log_float16_t(<8 x half> %src) {
+; CHECK-MVE-LABEL: log_float16_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, lr}
+; CHECK-MVE-NEXT: push {r4, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: .pad #32
+; CHECK-MVE-NEXT: sub sp, #32
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVE-NEXT: vmov q4, q0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #24]
+; CHECK-MVE-NEXT: ldr r0, [sp, #24]
+; CHECK-MVE-NEXT: bl logf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r4, s0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #28]
+; CHECK-MVE-NEXT: ldr r0, [sp, #28]
+; CHECK-MVE-NEXT: bl logf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vdup.16 q5, r1
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[0], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.16 q5[1], r4
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #20]
+; CHECK-MVE-NEXT: ldr r0, [sp, #20]
+; CHECK-MVE-NEXT: bl logf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[2], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #16]
+; CHECK-MVE-NEXT: ldr r0, [sp, #16]
+; CHECK-MVE-NEXT: bl logf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[3], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #12]
+; CHECK-MVE-NEXT: ldr r0, [sp, #12]
+; CHECK-MVE-NEXT: bl logf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[4], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #8]
+; CHECK-MVE-NEXT: ldr r0, [sp, #8]
+; CHECK-MVE-NEXT: bl logf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[5], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #4]
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl logf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[6], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp]
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl logf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[7], r0
+; CHECK-MVE-NEXT: vmov q0, q5
+; CHECK-MVE-NEXT: add sp, #32
+; CHECK-MVE-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: pop {r4, pc}
+;
+; CHECK-MVEFP-LABEL: log_float16_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, lr}
+; CHECK-MVEFP-NEXT: push {r4, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: .pad #32
+; CHECK-MVEFP-NEXT: sub sp, #32
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVEFP-NEXT: vmov q4, q0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #24]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #24]
+; CHECK-MVEFP-NEXT: bl logf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r4, s0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #28]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #28]
+; CHECK-MVEFP-NEXT: bl logf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[0], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.16 q5[1], r4
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #20]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #20]
+; CHECK-MVEFP-NEXT: bl logf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[2], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #16]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #16]
+; CHECK-MVEFP-NEXT: bl logf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[3], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #12]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #12]
+; CHECK-MVEFP-NEXT: bl logf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[4], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #8]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #8]
+; CHECK-MVEFP-NEXT: bl logf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[5], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #4]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl logf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[6], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp]
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl logf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[7], r0
+; CHECK-MVEFP-NEXT: vmov q0, q5
+; CHECK-MVEFP-NEXT: add sp, #32
+; CHECK-MVEFP-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: pop {r4, pc}
+entry:
+ %0 = call fast <8 x half> @llvm.log.v8f16(<8 x half> %src)
+ ret <8 x half> %0
+}
+
+define arm_aapcs_vfpcc <4 x float> @log2_float32_t(<4 x float> %src) {
+; CHECK-MVE-LABEL: log2_float32_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10}
+; CHECK-MVE-NEXT: .pad #16
+; CHECK-MVE-NEXT: sub sp, #16
+; CHECK-MVE-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl log2f
+; CHECK-MVE-NEXT: mov r4, r0
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl log2f
+; CHECK-MVE-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVE-NEXT: vmov s16, r0
+; CHECK-MVE-NEXT: vmov s18, r4
+; CHECK-MVE-NEXT: mov r0, r1
+; CHECK-MVE-NEXT: bl log2f
+; CHECK-MVE-NEXT: vmov s20, r0
+; CHECK-MVE-NEXT: mov r0, r5
+; CHECK-MVE-NEXT: bl log2f
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vmov s4, r0
+; CHECK-MVE-NEXT: vdup.32 q0, r1
+; CHECK-MVE-NEXT: vmov.f32 s0, s16
+; CHECK-MVE-NEXT: vmov.f32 s1, s18
+; CHECK-MVE-NEXT: vmov.f32 s2, s20
+; CHECK-MVE-NEXT: vmov.f32 s3, s4
+; CHECK-MVE-NEXT: add sp, #16
+; CHECK-MVE-NEXT: vpop {d8, d9, d10}
+; CHECK-MVE-NEXT: pop {r4, r5, r7, pc}
+;
+; CHECK-MVEFP-LABEL: log2_float32_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9}
+; CHECK-MVEFP-NEXT: vpush {d8, d9}
+; CHECK-MVEFP-NEXT: .pad #16
+; CHECK-MVEFP-NEXT: sub sp, #16
+; CHECK-MVEFP-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl log2f
+; CHECK-MVEFP-NEXT: mov r4, r0
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl log2f
+; CHECK-MVEFP-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVEFP-NEXT: vmov s16, r0
+; CHECK-MVEFP-NEXT: vmov s17, r4
+; CHECK-MVEFP-NEXT: mov r0, r1
+; CHECK-MVEFP-NEXT: bl log2f
+; CHECK-MVEFP-NEXT: vmov s18, r0
+; CHECK-MVEFP-NEXT: mov r0, r5
+; CHECK-MVEFP-NEXT: bl log2f
+; CHECK-MVEFP-NEXT: vmov s19, r0
+; CHECK-MVEFP-NEXT: vmov q0, q4
+; CHECK-MVEFP-NEXT: add sp, #16
+; CHECK-MVEFP-NEXT: vpop {d8, d9}
+; CHECK-MVEFP-NEXT: pop {r4, r5, r7, pc}
+entry:
+ %0 = call fast <4 x float> @llvm.log2.v4f32(<4 x float> %src)
+ ret <4 x float> %0
+}
+
+define arm_aapcs_vfpcc <8 x half> @log2_float16_t(<8 x half> %src) {
+; CHECK-MVE-LABEL: log2_float16_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, lr}
+; CHECK-MVE-NEXT: push {r4, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: .pad #32
+; CHECK-MVE-NEXT: sub sp, #32
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVE-NEXT: vmov q4, q0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #24]
+; CHECK-MVE-NEXT: ldr r0, [sp, #24]
+; CHECK-MVE-NEXT: bl log2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r4, s0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #28]
+; CHECK-MVE-NEXT: ldr r0, [sp, #28]
+; CHECK-MVE-NEXT: bl log2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vdup.16 q5, r1
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[0], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.16 q5[1], r4
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #20]
+; CHECK-MVE-NEXT: ldr r0, [sp, #20]
+; CHECK-MVE-NEXT: bl log2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[2], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #16]
+; CHECK-MVE-NEXT: ldr r0, [sp, #16]
+; CHECK-MVE-NEXT: bl log2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[3], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #12]
+; CHECK-MVE-NEXT: ldr r0, [sp, #12]
+; CHECK-MVE-NEXT: bl log2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[4], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #8]
+; CHECK-MVE-NEXT: ldr r0, [sp, #8]
+; CHECK-MVE-NEXT: bl log2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[5], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #4]
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl log2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[6], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp]
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl log2f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[7], r0
+; CHECK-MVE-NEXT: vmov q0, q5
+; CHECK-MVE-NEXT: add sp, #32
+; CHECK-MVE-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: pop {r4, pc}
+;
+; CHECK-MVEFP-LABEL: log2_float16_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, lr}
+; CHECK-MVEFP-NEXT: push {r4, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: .pad #32
+; CHECK-MVEFP-NEXT: sub sp, #32
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVEFP-NEXT: vmov q4, q0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #24]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #24]
+; CHECK-MVEFP-NEXT: bl log2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r4, s0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #28]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #28]
+; CHECK-MVEFP-NEXT: bl log2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[0], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.16 q5[1], r4
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #20]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #20]
+; CHECK-MVEFP-NEXT: bl log2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[2], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #16]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #16]
+; CHECK-MVEFP-NEXT: bl log2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[3], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #12]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #12]
+; CHECK-MVEFP-NEXT: bl log2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[4], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #8]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #8]
+; CHECK-MVEFP-NEXT: bl log2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[5], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #4]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl log2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[6], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp]
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl log2f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[7], r0
+; CHECK-MVEFP-NEXT: vmov q0, q5
+; CHECK-MVEFP-NEXT: add sp, #32
+; CHECK-MVEFP-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: pop {r4, pc}
+entry:
+ %0 = call fast <8 x half> @llvm.log2.v8f16(<8 x half> %src)
+ ret <8 x half> %0
+}
+
+define arm_aapcs_vfpcc <4 x float> @log10_float32_t(<4 x float> %src) {
+; CHECK-MVE-LABEL: log10_float32_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10}
+; CHECK-MVE-NEXT: .pad #16
+; CHECK-MVE-NEXT: sub sp, #16
+; CHECK-MVE-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl log10f
+; CHECK-MVE-NEXT: mov r4, r0
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl log10f
+; CHECK-MVE-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVE-NEXT: vmov s16, r0
+; CHECK-MVE-NEXT: vmov s18, r4
+; CHECK-MVE-NEXT: mov r0, r1
+; CHECK-MVE-NEXT: bl log10f
+; CHECK-MVE-NEXT: vmov s20, r0
+; CHECK-MVE-NEXT: mov r0, r5
+; CHECK-MVE-NEXT: bl log10f
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vmov s4, r0
+; CHECK-MVE-NEXT: vdup.32 q0, r1
+; CHECK-MVE-NEXT: vmov.f32 s0, s16
+; CHECK-MVE-NEXT: vmov.f32 s1, s18
+; CHECK-MVE-NEXT: vmov.f32 s2, s20
+; CHECK-MVE-NEXT: vmov.f32 s3, s4
+; CHECK-MVE-NEXT: add sp, #16
+; CHECK-MVE-NEXT: vpop {d8, d9, d10}
+; CHECK-MVE-NEXT: pop {r4, r5, r7, pc}
+;
+; CHECK-MVEFP-LABEL: log10_float32_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: push {r4, r5, r7, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9}
+; CHECK-MVEFP-NEXT: vpush {d8, d9}
+; CHECK-MVEFP-NEXT: .pad #16
+; CHECK-MVEFP-NEXT: sub sp, #16
+; CHECK-MVEFP-NEXT: vstmia sp, {s0, s1, s2, s3}
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl log10f
+; CHECK-MVEFP-NEXT: mov r4, r0
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl log10f
+; CHECK-MVEFP-NEXT: ldrd r1, r5, [sp, #8]
+; CHECK-MVEFP-NEXT: vmov s16, r0
+; CHECK-MVEFP-NEXT: vmov s17, r4
+; CHECK-MVEFP-NEXT: mov r0, r1
+; CHECK-MVEFP-NEXT: bl log10f
+; CHECK-MVEFP-NEXT: vmov s18, r0
+; CHECK-MVEFP-NEXT: mov r0, r5
+; CHECK-MVEFP-NEXT: bl log10f
+; CHECK-MVEFP-NEXT: vmov s19, r0
+; CHECK-MVEFP-NEXT: vmov q0, q4
+; CHECK-MVEFP-NEXT: add sp, #16
+; CHECK-MVEFP-NEXT: vpop {d8, d9}
+; CHECK-MVEFP-NEXT: pop {r4, r5, r7, pc}
+entry:
+ %0 = call fast <4 x float> @llvm.log10.v4f32(<4 x float> %src)
+ ret <4 x float> %0
+}
+
+define arm_aapcs_vfpcc <8 x half> @log10_float16_t(<8 x half> %src) {
+; CHECK-MVE-LABEL: log10_float16_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, lr}
+; CHECK-MVE-NEXT: push {r4, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: .pad #32
+; CHECK-MVE-NEXT: sub sp, #32
+; CHECK-MVE-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVE-NEXT: vmov q4, q0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #24]
+; CHECK-MVE-NEXT: ldr r0, [sp, #24]
+; CHECK-MVE-NEXT: bl log10f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r4, s0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #28]
+; CHECK-MVE-NEXT: ldr r0, [sp, #28]
+; CHECK-MVE-NEXT: bl log10f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vdup.16 q5, r1
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[0], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.16 q5[1], r4
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #20]
+; CHECK-MVE-NEXT: ldr r0, [sp, #20]
+; CHECK-MVE-NEXT: bl log10f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[2], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #16]
+; CHECK-MVE-NEXT: ldr r0, [sp, #16]
+; CHECK-MVE-NEXT: bl log10f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[3], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #12]
+; CHECK-MVE-NEXT: ldr r0, [sp, #12]
+; CHECK-MVE-NEXT: bl log10f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[4], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #8]
+; CHECK-MVE-NEXT: ldr r0, [sp, #8]
+; CHECK-MVE-NEXT: bl log10f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[5], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #4]
+; CHECK-MVE-NEXT: ldr r0, [sp, #4]
+; CHECK-MVE-NEXT: bl log10f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[6], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp]
+; CHECK-MVE-NEXT: ldr r0, [sp]
+; CHECK-MVE-NEXT: bl log10f
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q5[7], r0
+; CHECK-MVE-NEXT: vmov q0, q5
+; CHECK-MVE-NEXT: add sp, #32
+; CHECK-MVE-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVE-NEXT: pop {r4, pc}
+;
+; CHECK-MVEFP-LABEL: log10_float16_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, lr}
+; CHECK-MVEFP-NEXT: push {r4, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: .pad #32
+; CHECK-MVEFP-NEXT: sub sp, #32
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q0[1]
+; CHECK-MVEFP-NEXT: vmov q4, q0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #24]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #24]
+; CHECK-MVEFP-NEXT: bl log10f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r4, s0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #28]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #28]
+; CHECK-MVEFP-NEXT: bl log10f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[0], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.16 q5[1], r4
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #20]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #20]
+; CHECK-MVEFP-NEXT: bl log10f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[2], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #16]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #16]
+; CHECK-MVEFP-NEXT: bl log10f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[3], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #12]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #12]
+; CHECK-MVEFP-NEXT: bl log10f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[4], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #8]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #8]
+; CHECK-MVEFP-NEXT: bl log10f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[5], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #4]
+; CHECK-MVEFP-NEXT: ldr r0, [sp, #4]
+; CHECK-MVEFP-NEXT: bl log10f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[6], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp]
+; CHECK-MVEFP-NEXT: ldr r0, [sp]
+; CHECK-MVEFP-NEXT: bl log10f
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q5[7], r0
+; CHECK-MVEFP-NEXT: vmov q0, q5
+; CHECK-MVEFP-NEXT: add sp, #32
+; CHECK-MVEFP-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-MVEFP-NEXT: pop {r4, pc}
+entry:
+ %0 = call fast <8 x half> @llvm.log10.v8f16(<8 x half> %src)
+ ret <8 x half> %0
+}
+
+define arm_aapcs_vfpcc <4 x float> @pow_float32_t(<4 x float> %src1, <4 x float> %src2) {
+; CHECK-MVE-LABEL: pow_float32_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, r5, r6, lr}
+; CHECK-MVE-NEXT: push {r4, r5, r6, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10}
+; CHECK-MVE-NEXT: .pad #32
+; CHECK-MVE-NEXT: sub sp, #32
+; CHECK-MVE-NEXT: vstr s3, [sp, #24]
+; CHECK-MVE-NEXT: ldr r4, [sp, #24]
+; CHECK-MVE-NEXT: vstr s1, [sp, #8]
+; CHECK-MVE-NEXT: vstr s5, [sp, #12]
+; CHECK-MVE-NEXT: vstr s0, [sp]
+; CHECK-MVE-NEXT: vstr s4, [sp, #4]
+; CHECK-MVE-NEXT: vstr s7, [sp, #28]
+; CHECK-MVE-NEXT: vstr s2, [sp, #16]
+; CHECK-MVE-NEXT: vstr s6, [sp, #20]
+; CHECK-MVE-NEXT: ldrd r0, r1, [sp, #8]
+; CHECK-MVE-NEXT: bl powf
+; CHECK-MVE-NEXT: mov r5, r0
+; CHECK-MVE-NEXT: ldrd r0, r1, [sp]
+; CHECK-MVE-NEXT: bl powf
+; CHECK-MVE-NEXT: ldrd r2, r1, [sp, #16]
+; CHECK-MVE-NEXT: vmov s16, r0
+; CHECK-MVE-NEXT: ldr r6, [sp, #28]
+; CHECK-MVE-NEXT: vmov s18, r5
+; CHECK-MVE-NEXT: mov r0, r2
+; CHECK-MVE-NEXT: bl powf
+; CHECK-MVE-NEXT: vmov s20, r0
+; CHECK-MVE-NEXT: mov r0, r4
+; CHECK-MVE-NEXT: mov r1, r6
+; CHECK-MVE-NEXT: bl powf
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vmov s4, r0
+; CHECK-MVE-NEXT: vdup.32 q0, r1
+; CHECK-MVE-NEXT: vmov.f32 s0, s16
+; CHECK-MVE-NEXT: vmov.f32 s1, s18
+; CHECK-MVE-NEXT: vmov.f32 s2, s20
+; CHECK-MVE-NEXT: vmov.f32 s3, s4
+; CHECK-MVE-NEXT: add sp, #32
+; CHECK-MVE-NEXT: vpop {d8, d9, d10}
+; CHECK-MVE-NEXT: pop {r4, r5, r6, pc}
+;
+; CHECK-MVEFP-LABEL: pow_float32_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, r5, r6, lr}
+; CHECK-MVEFP-NEXT: push {r4, r5, r6, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9}
+; CHECK-MVEFP-NEXT: vpush {d8, d9}
+; CHECK-MVEFP-NEXT: .pad #32
+; CHECK-MVEFP-NEXT: sub sp, #32
+; CHECK-MVEFP-NEXT: vstr s3, [sp, #24]
+; CHECK-MVEFP-NEXT: ldr r4, [sp, #24]
+; CHECK-MVEFP-NEXT: vstr s1, [sp, #8]
+; CHECK-MVEFP-NEXT: vstr s5, [sp, #12]
+; CHECK-MVEFP-NEXT: vstr s0, [sp]
+; CHECK-MVEFP-NEXT: vstr s4, [sp, #4]
+; CHECK-MVEFP-NEXT: vstr s7, [sp, #28]
+; CHECK-MVEFP-NEXT: vstr s2, [sp, #16]
+; CHECK-MVEFP-NEXT: vstr s6, [sp, #20]
+; CHECK-MVEFP-NEXT: ldrd r0, r1, [sp, #8]
+; CHECK-MVEFP-NEXT: bl powf
+; CHECK-MVEFP-NEXT: mov r5, r0
+; CHECK-MVEFP-NEXT: ldrd r0, r1, [sp]
+; CHECK-MVEFP-NEXT: bl powf
+; CHECK-MVEFP-NEXT: ldrd r2, r1, [sp, #16]
+; CHECK-MVEFP-NEXT: vmov s16, r0
+; CHECK-MVEFP-NEXT: ldr r6, [sp, #28]
+; CHECK-MVEFP-NEXT: vmov s17, r5
+; CHECK-MVEFP-NEXT: mov r0, r2
+; CHECK-MVEFP-NEXT: bl powf
+; CHECK-MVEFP-NEXT: vmov s18, r0
+; CHECK-MVEFP-NEXT: mov r0, r4
+; CHECK-MVEFP-NEXT: mov r1, r6
+; CHECK-MVEFP-NEXT: bl powf
+; CHECK-MVEFP-NEXT: vmov s19, r0
+; CHECK-MVEFP-NEXT: vmov q0, q4
+; CHECK-MVEFP-NEXT: add sp, #32
+; CHECK-MVEFP-NEXT: vpop {d8, d9}
+; CHECK-MVEFP-NEXT: pop {r4, r5, r6, pc}
+entry:
+ %0 = call fast <4 x float> @llvm.pow.v4f32(<4 x float> %src1, <4 x float> %src2)
+ ret <4 x float> %0
+}
+
+define arm_aapcs_vfpcc <8 x half> @pow_float16_t(<8 x half> %src1, <8 x half> %src2) {
+; CHECK-MVE-LABEL: pow_float16_t:
+; CHECK-MVE: @ %bb.0: @ %entry
+; CHECK-MVE-NEXT: .save {r4, lr}
+; CHECK-MVE-NEXT: push {r4, lr}
+; CHECK-MVE-NEXT: .vsave {d8, d9, d10, d11, d12, d13}
+; CHECK-MVE-NEXT: vpush {d8, d9, d10, d11, d12, d13}
+; CHECK-MVE-NEXT: .pad #64
+; CHECK-MVE-NEXT: sub sp, #64
+; CHECK-MVE-NEXT: vmov q5, q1
+; CHECK-MVE-NEXT: vmov q4, q0
+; CHECK-MVE-NEXT: vmov.u16 r0, q1[0]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVE-NEXT: vmov s2, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVE-NEXT: vstr s2, [sp, #56]
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #60]
+; CHECK-MVE-NEXT: ldrd r0, r1, [sp, #56]
+; CHECK-MVE-NEXT: bl powf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q5[1]
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r4, s0
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[1]
+; CHECK-MVE-NEXT: vmov s2, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVE-NEXT: vstr s2, [sp, #48]
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #52]
+; CHECK-MVE-NEXT: ldrd r0, r1, [sp, #48]
+; CHECK-MVE-NEXT: bl powf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: movs r1, #0
+; CHECK-MVE-NEXT: vdup.16 q6, r1
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q6[0], r4
+; CHECK-MVE-NEXT: vmov.16 q6[1], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q5[2]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVE-NEXT: vmov s2, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVE-NEXT: vstr s2, [sp, #40]
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #44]
+; CHECK-MVE-NEXT: ldrd r0, r1, [sp, #40]
+; CHECK-MVE-NEXT: bl powf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q6[2], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q5[3]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVE-NEXT: vmov s2, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVE-NEXT: vstr s2, [sp, #32]
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #36]
+; CHECK-MVE-NEXT: ldrd r0, r1, [sp, #32]
+; CHECK-MVE-NEXT: bl powf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q6[3], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q5[4]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVE-NEXT: vmov s2, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVE-NEXT: vstr s2, [sp, #24]
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #28]
+; CHECK-MVE-NEXT: ldrd r0, r1, [sp, #24]
+; CHECK-MVE-NEXT: bl powf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q6[4], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q5[5]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVE-NEXT: vmov s2, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVE-NEXT: vstr s2, [sp, #16]
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #20]
+; CHECK-MVE-NEXT: ldrd r0, r1, [sp, #16]
+; CHECK-MVE-NEXT: bl powf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q6[5], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q5[6]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVE-NEXT: vmov s2, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVE-NEXT: vstr s2, [sp, #8]
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #12]
+; CHECK-MVE-NEXT: ldrd r0, r1, [sp, #8]
+; CHECK-MVE-NEXT: bl powf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q6[6], r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q5[7]
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVE-NEXT: vmov s2, r0
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVE-NEXT: vstr s2, [sp]
+; CHECK-MVE-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVE-NEXT: vstr s0, [sp, #4]
+; CHECK-MVE-NEXT: ldrd r0, r1, [sp]
+; CHECK-MVE-NEXT: bl powf
+; CHECK-MVE-NEXT: vmov s0, r0
+; CHECK-MVE-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVE-NEXT: vmov r0, s0
+; CHECK-MVE-NEXT: vmov.16 q6[7], r0
+; CHECK-MVE-NEXT: vmov q0, q6
+; CHECK-MVE-NEXT: add sp, #64
+; CHECK-MVE-NEXT: vpop {d8, d9, d10, d11, d12, d13}
+; CHECK-MVE-NEXT: pop {r4, pc}
+;
+; CHECK-MVEFP-LABEL: pow_float16_t:
+; CHECK-MVEFP: @ %bb.0: @ %entry
+; CHECK-MVEFP-NEXT: .save {r4, lr}
+; CHECK-MVEFP-NEXT: push {r4, lr}
+; CHECK-MVEFP-NEXT: .vsave {d8, d9, d10, d11, d12, d13}
+; CHECK-MVEFP-NEXT: vpush {d8, d9, d10, d11, d12, d13}
+; CHECK-MVEFP-NEXT: .pad #64
+; CHECK-MVEFP-NEXT: sub sp, #64
+; CHECK-MVEFP-NEXT: vmov q5, q1
+; CHECK-MVEFP-NEXT: vmov q4, q0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q1[0]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[0]
+; CHECK-MVEFP-NEXT: vmov s2, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVEFP-NEXT: vstr s2, [sp, #56]
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #60]
+; CHECK-MVEFP-NEXT: ldrd r0, r1, [sp, #56]
+; CHECK-MVEFP-NEXT: bl powf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q5[1]
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r4, s0
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[1]
+; CHECK-MVEFP-NEXT: vmov s2, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVEFP-NEXT: vstr s2, [sp, #48]
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #52]
+; CHECK-MVEFP-NEXT: ldrd r0, r1, [sp, #48]
+; CHECK-MVEFP-NEXT: bl powf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.16 q6[0], r4
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q6[1], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q5[2]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[2]
+; CHECK-MVEFP-NEXT: vmov s2, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVEFP-NEXT: vstr s2, [sp, #40]
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #44]
+; CHECK-MVEFP-NEXT: ldrd r0, r1, [sp, #40]
+; CHECK-MVEFP-NEXT: bl powf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q6[2], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q5[3]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[3]
+; CHECK-MVEFP-NEXT: vmov s2, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVEFP-NEXT: vstr s2, [sp, #32]
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #36]
+; CHECK-MVEFP-NEXT: ldrd r0, r1, [sp, #32]
+; CHECK-MVEFP-NEXT: bl powf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q6[3], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q5[4]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[4]
+; CHECK-MVEFP-NEXT: vmov s2, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVEFP-NEXT: vstr s2, [sp, #24]
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #28]
+; CHECK-MVEFP-NEXT: ldrd r0, r1, [sp, #24]
+; CHECK-MVEFP-NEXT: bl powf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q6[4], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q5[5]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[5]
+; CHECK-MVEFP-NEXT: vmov s2, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVEFP-NEXT: vstr s2, [sp, #16]
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #20]
+; CHECK-MVEFP-NEXT: ldrd r0, r1, [sp, #16]
+; CHECK-MVEFP-NEXT: bl powf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q6[5], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q5[6]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[6]
+; CHECK-MVEFP-NEXT: vmov s2, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVEFP-NEXT: vstr s2, [sp, #8]
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #12]
+; CHECK-MVEFP-NEXT: ldrd r0, r1, [sp, #8]
+; CHECK-MVEFP-NEXT: bl powf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q6[6], r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q5[7]
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vmov.u16 r0, q4[7]
+; CHECK-MVEFP-NEXT: vmov s2, r0
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s2, s2
+; CHECK-MVEFP-NEXT: vstr s2, [sp]
+; CHECK-MVEFP-NEXT: vcvtb.f32.f16 s0, s0
+; CHECK-MVEFP-NEXT: vstr s0, [sp, #4]
+; CHECK-MVEFP-NEXT: ldrd r0, r1, [sp]
+; CHECK-MVEFP-NEXT: bl powf
+; CHECK-MVEFP-NEXT: vmov s0, r0
+; CHECK-MVEFP-NEXT: vcvtb.f16.f32 s0, s0
+; CHECK-MVEFP-NEXT: vmov r0, s0
+; CHECK-MVEFP-NEXT: vmov.16 q6[7], r0
+; CHECK-MVEFP-NEXT: vmov q0, q6
+; CHECK-MVEFP-NEXT: add sp, #64
+; CHECK-MVEFP-NEXT: vpop {d8, d9, d10, d11, d12, d13}
+; CHECK-MVEFP-NEXT: pop {r4, pc}
+entry:
+ %0 = call fast <8 x half> @llvm.pow.v8f16(<8 x half> %src1, <8 x half> %src2)
+ ret <8 x half> %0
+}
+
+declare <4 x float> @llvm.sqrt.v4f32(<4 x float>)
+declare <4 x float> @llvm.cos.v4f32(<4 x float>)
+declare <4 x float> @llvm.sin.v4f32(<4 x float>)
+declare <4 x float> @llvm.exp.v4f32(<4 x float>)
+declare <4 x float> @llvm.exp2.v4f32(<4 x float>)
+declare <4 x float> @llvm.log.v4f32(<4 x float>)
+declare <4 x float> @llvm.log2.v4f32(<4 x float>)
+declare <4 x float> @llvm.log10.v4f32(<4 x float>)
+declare <4 x float> @llvm.pow.v4f32(<4 x float>, <4 x float>)
+declare <8 x half> @llvm.sqrt.v8f16(<8 x half>)
+declare <8 x half> @llvm.cos.v8f16(<8 x half>)
+declare <8 x half> @llvm.sin.v8f16(<8 x half>)
+declare <8 x half> @llvm.exp.v8f16(<8 x half>)
+declare <8 x half> @llvm.exp2.v8f16(<8 x half>)
+declare <8 x half> @llvm.log.v8f16(<8 x half>)
+declare <8 x half> @llvm.log2.v8f16(<8 x half>)
+declare <8 x half> @llvm.log10.v8f16(<8 x half>)
+declare <8 x half> @llvm.pow.v8f16(<8 x half>, <8 x half>)
+