1 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7 -mattr=+sse2,-sse4.1 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=SSE2
2 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=SSE41
3 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7-avx -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX
4 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=core-avx2 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX2
7 ; Verify the cost of vector select instructions.
9 ; SSE41 added blend instructions with an immediate for <2 x double> and
10 ; <4 x float>. Integers of the same size should also use those instructions.
12 define <2 x i64> @test_2i64(<2 x i64> %a, <2 x i64> %b) {
13 ; CHECK: Printing analysis 'Cost Model Analysis' for function 'test_2i64':
14 ; SSE2: Cost Model: {{.*}} 4 for instruction: %sel = select <2 x i1>
15 ; SSE41: Cost Model: {{.*}} 1 for instruction: %sel = select <2 x i1>
16 ; AVX: Cost Model: {{.*}} 1 for instruction: %sel = select <2 x i1>
17 ; AVX2: Cost Model: {{.*}} 1 for instruction: %sel = select <2 x i1>
18 %sel = select <2 x i1> <i1 true, i1 false>, <2 x i64> %a, <2 x i64> %b
22 define <2 x double> @test_2double(<2 x double> %a, <2 x double> %b) {
23 ; CHECK: Printing analysis 'Cost Model Analysis' for function 'test_2double':
24 ; SSE2: Cost Model: {{.*}} 3 for instruction: %sel = select <2 x i1>
25 ; SSE41: Cost Model: {{.*}} 1 for instruction: %sel = select <2 x i1>
26 ; AVX: Cost Model: {{.*}} 1 for instruction: %sel = select <2 x i1>
27 ; AVX2: Cost Model: {{.*}} 1 for instruction: %sel = select <2 x i1>
28 %sel = select <2 x i1> <i1 true, i1 false>, <2 x double> %a, <2 x double> %b
32 define <4 x i32> @test_4i32(<4 x i32> %a, <4 x i32> %b) {
33 ; CHECK: Printing analysis 'Cost Model Analysis' for function 'test_4i32':
34 ; SSE2: Cost Model: {{.*}} 8 for instruction: %sel = select <4 x i1>
35 ; SSE41: Cost Model: {{.*}} 1 for instruction: %sel = select <4 x i1>
36 ; AVX: Cost Model: {{.*}} 1 for instruction: %sel = select <4 x i1>
37 ; AVX2: Cost Model: {{.*}} 1 for instruction: %sel = select <4 x i1>
38 %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x i32> %a, <4 x i32> %b
42 define <4 x float> @test_4float(<4 x float> %a, <4 x float> %b) {
43 ; CHECK: Printing analysis 'Cost Model Analysis' for function 'test_4float':
44 ; SSE2: Cost Model: {{.*}} 7 for instruction: %sel = select <4 x i1>
45 ; SSE41: Cost Model: {{.*}} 1 for instruction: %sel = select <4 x i1>
46 ; AVX: Cost Model: {{.*}} 1 for instruction: %sel = select <4 x i1>
47 ; AVX2: Cost Model: {{.*}} 1 for instruction: %sel = select <4 x i1>
48 %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 true>, <4 x float> %a, <4 x float> %b
52 define <16 x i8> @test_16i8(<16 x i8> %a, <16 x i8> %b) {
53 ; CHECK: Printing analysis 'Cost Model Analysis' for function 'test_16i8':
54 ; SSE2: Cost Model: {{.*}} 32 for instruction: %sel = select <16 x i1>
55 ; SSE41: Cost Model: {{.*}} 1 for instruction: %sel = select <16 x i1>
56 ; AVX: Cost Model: {{.*}} 1 for instruction: %sel = select <16 x i1>
57 ; AVX2: Cost Model: {{.*}} 1 for instruction: %sel = select <16 x i1>
58 %sel = select <16 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true>, <16 x i8> %a, <16 x i8> %b
62 ; AVX added blend instructions with an immediate for <4 x double> and
63 ; <8 x float>. Integers of the same size should also use those instructions.
64 define <4 x i64> @test_4i64(<4 x i64> %a, <4 x i64> %b) {
65 ; CHECK: Printing analysis 'Cost Model Analysis' for function 'test_4i64':
66 ; SSE2: Cost Model: {{.*}} 8 for instruction: %sel = select <4 x i1>
67 ; SSE41: Cost Model: {{.*}} 2 for instruction: %sel = select <4 x i1>
68 ; AVX: Cost Model: {{.*}} 1 for instruction: %sel = select <4 x i1>
69 ; AVX2: Cost Model: {{.*}} 1 for instruction: %sel = select <4 x i1>
70 %sel = select <4 x i1> <i1 true, i1 false, i1 false, i1 true>, <4 x i64> %a, <4 x i64> %b
74 define <4 x double> @test_4double(<4 x double> %a, <4 x double> %b) {
75 ; CHECK: Printing analysis 'Cost Model Analysis' for function 'test_4double':
76 ; SSE2: Cost Model: {{.*}} 6 for instruction: %sel = select <4 x i1>
77 ; SSE41: Cost Model: {{.*}} 2 for instruction: %sel = select <4 x i1>
78 ; AVX: Cost Model: {{.*}} 1 for instruction: %sel = select <4 x i1>
79 ; AVX2: Cost Model: {{.*}} 1 for instruction: %sel = select <4 x i1>
80 %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x double> %a, <4 x double> %b
84 define <8 x i32> @test_8i32(<8 x i32> %a, <8 x i32> %b) {
85 ; CHECK: Printing analysis 'Cost Model Analysis' for function 'test_8i32':
86 ; SSE2: Cost Model: {{.*}} 16 for instruction: %sel = select <8 x i1>
87 ; SSE41: Cost Model: {{.*}} 2 for instruction: %sel = select <8 x i1>
88 ; AVX: Cost Model: {{.*}} 1 for instruction: %sel = select <8 x i1>
89 ; AVX2: Cost Model: {{.*}} 1 for instruction: %sel = select <8 x i1>
90 %sel = select <8 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 false>, <8 x i32> %a, <8 x i32> %b
94 define <8 x float> @test_8float(<8 x float> %a, <8 x float> %b) {
95 ; CHECK: Printing analysis 'Cost Model Analysis' for function 'test_8float':
96 ; SSE2: Cost Model: {{.*}} 14 for instruction: %sel = select <8 x i1>
97 ; SSE41: Cost Model: {{.*}} 2 for instruction: %sel = select <8 x i1>
98 ; AVX: Cost Model: {{.*}} 1 for instruction: %sel = select <8 x i1>
99 ; AVX2: Cost Model: {{.*}} 1 for instruction: %sel = select <8 x i1>
100 %sel = select <8 x i1> <i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false>, <8 x float> %a, <8 x float> %b
105 define <16 x i16> @test_16i16(<16 x i16> %a, <16 x i16> %b) {
106 ; CHECK:Printing analysis 'Cost Model Analysis' for function 'test_16i16':
107 ; SSE2: Cost Model: {{.*}} 32 for instruction: %sel = select <16 x i1>
108 ; SSE41: Cost Model: {{.*}} 2 for instruction: %sel = select <16 x i1>
109 ;;; FIXME: This AVX cost is obviously wrong. We shouldn't be scalarizing.
110 ; AVX: Cost Model: {{.*}} 32 for instruction: %sel = select <16 x i1>
111 ; AVX2: Cost Model: {{.*}} 1 for instruction: %sel = select <16 x i1>
112 %sel = select <16 x i1> <i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false>, <16 x i16> %a, <16 x i16> %b
116 define <32 x i8> @test_32i8(<32 x i8> %a, <32 x i8> %b) {
117 ; CHECK: Printing analysis 'Cost Model Analysis' for function 'test_32i8':
118 ; SSE2: Cost Model: {{.*}} 64 for instruction: %sel = select <32 x i1>
119 ; SSE41: Cost Model: {{.*}} 2 for instruction: %sel = select <32 x i1>
120 ;;; FIXME: This AVX cost is obviously wrong. We shouldn't be scalarizing.
121 ; AVX: Cost Model: {{.*}} 64 for instruction: %sel = select <32 x i1>
122 ; AVX2: Cost Model: {{.*}} 1 for instruction: %sel = select <32 x i1>
123 %sel = select <32 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true>, <32 x i8> %a, <32 x i8> %b