1 ; RUN: opt < %s -cost-model -costmodel-reduxcost=true -analyze -mcpu=core2 -mtriple=x86_64-apple-darwin | FileCheck %s
3 define fastcc float @reduction_cost_float(<4 x float> %rdx) {
4 %rdx.shuf = shufflevector <4 x float> %rdx, <4 x float> undef, <4 x i32> <i32 2, i32 3, i32 undef, i32 undef>
5 %bin.rdx = fadd <4 x float> %rdx, %rdx.shuf
6 %rdx.shuf7 = shufflevector <4 x float> %bin.rdx, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
7 %bin.rdx8 = fadd <4 x float> %bin.rdx, %rdx.shuf7
9 ; Check that we recognize the tree starting at the extractelement as a
11 ; CHECK-LABEL: reduction_cost
12 ; CHECK: cost of 9 {{.*}} extractelement
14 %r = extractelement <4 x float> %bin.rdx8, i32 0
18 define fastcc i32 @reduction_cost_int(<8 x i32> %rdx) {
19 %rdx.shuf = shufflevector <8 x i32> %rdx, <8 x i32> undef,
20 <8 x i32> <i32 4 , i32 5, i32 6, i32 7,
21 i32 undef, i32 undef, i32 undef, i32 undef>
22 %bin.rdx = add <8 x i32> %rdx, %rdx.shuf
23 %rdx.shuf.2 = shufflevector <8 x i32> %bin.rdx, <8 x i32> undef,
24 <8 x i32> <i32 2 , i32 3, i32 undef, i32 undef,
25 i32 undef, i32 undef, i32 undef, i32 undef>
26 %bin.rdx.2 = add <8 x i32> %bin.rdx, %rdx.shuf.2
27 %rdx.shuf.3 = shufflevector <8 x i32> %bin.rdx.2, <8 x i32> undef,
28 <8 x i32> <i32 1 , i32 undef, i32 undef, i32 undef,
29 i32 undef, i32 undef, i32 undef, i32 undef>
30 %bin.rdx.3 = add <8 x i32> %bin.rdx.2, %rdx.shuf.3
32 ; CHECK-LABEL: reduction_cost_int
33 ; CHECK: cost of 23 {{.*}} extractelement
35 %r = extractelement <8 x i32> %bin.rdx.3, i32 0
39 define fastcc float @pairwise_hadd(<4 x float> %rdx, float %f1) {
40 %rdx.shuf.0.0 = shufflevector <4 x float> %rdx, <4 x float> undef,
41 <4 x i32> <i32 0, i32 2 , i32 undef, i32 undef>
42 %rdx.shuf.0.1 = shufflevector <4 x float> %rdx, <4 x float> undef,
43 <4 x i32> <i32 1, i32 3, i32 undef, i32 undef>
44 %bin.rdx.0 = fadd <4 x float> %rdx.shuf.0.0, %rdx.shuf.0.1
45 %rdx.shuf.1.0 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef,
46 <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
47 %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef,
48 <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
49 %bin.rdx.1 = fadd <4 x float> %rdx.shuf.1.0, %rdx.shuf.1.1
51 ; CHECK-LABEL: pairwise_hadd
52 ; CHECK: cost of 11 {{.*}} extractelement
54 %r = extractelement <4 x float> %bin.rdx.1, i32 0
55 %r2 = fadd float %r, %f1
58 define fastcc float @pairwise_hadd_assoc(<4 x float> %rdx, float %f1) {
59 %rdx.shuf.0.0 = shufflevector <4 x float> %rdx, <4 x float> undef,
60 <4 x i32> <i32 0, i32 2 , i32 undef, i32 undef>
61 %rdx.shuf.0.1 = shufflevector <4 x float> %rdx, <4 x float> undef,
62 <4 x i32> <i32 1, i32 3, i32 undef, i32 undef>
63 %bin.rdx.0 = fadd <4 x float> %rdx.shuf.0.1, %rdx.shuf.0.0
64 %rdx.shuf.1.0 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef,
65 <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
66 %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef,
67 <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
68 %bin.rdx.1 = fadd <4 x float> %rdx.shuf.1.0, %rdx.shuf.1.1
70 ; CHECK-LABEL: pairwise_hadd_assoc
71 ; CHECK: cost of 11 {{.*}} extractelement
73 %r = extractelement <4 x float> %bin.rdx.1, i32 0
74 %r2 = fadd float %r, %f1
78 define fastcc float @pairwise_hadd_skip_first(<4 x float> %rdx, float %f1) {
79 %rdx.shuf.0.0 = shufflevector <4 x float> %rdx, <4 x float> undef,
80 <4 x i32> <i32 0, i32 2 , i32 undef, i32 undef>
81 %rdx.shuf.0.1 = shufflevector <4 x float> %rdx, <4 x float> undef,
82 <4 x i32> <i32 1, i32 3, i32 undef, i32 undef>
83 %bin.rdx.0 = fadd <4 x float> %rdx.shuf.0.0, %rdx.shuf.0.1
84 %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef,
85 <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
86 %bin.rdx.1 = fadd <4 x float> %bin.rdx.0, %rdx.shuf.1.1
88 ; CHECK-LABEL: pairwise_hadd_skip_first
89 ; CHECK: cost of 11 {{.*}} extractelement
91 %r = extractelement <4 x float> %bin.rdx.1, i32 0
92 %r2 = fadd float %r, %f1