1; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=core-avx2 -mattr=avx2,+fma -fp-contract=fast | FileCheck %s
2; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=bdver2 -mattr=-fma4 -fp-contract=fast | FileCheck %s
3; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=bdver1 -fp-contract=fast | FileCheck %s --check-prefix=CHECK_FMA4
4
5; CHECK-LABEL: test_x86_fmadd_ps_y_wide
6; CHECK: vfmadd213ps
7; CHECK: vfmadd213ps
8; CHECK: ret
9; CHECK_FMA4-LABEL: test_x86_fmadd_ps_y_wide
10; CHECK_FMA4: vfmaddps
11; CHECK_FMA4: vfmaddps
12; CHECK_FMA4: ret
13define <16 x float> @test_x86_fmadd_ps_y_wide(<16 x float> %a0, <16 x float> %a1, <16 x float> %a2) {
14  %x = fmul <16 x float> %a0, %a1
15  %res = fadd <16 x float> %x, %a2
16  ret <16 x float> %res
17}
18
19; CHECK-LABEL: test_x86_fmsub_ps_y_wide
20; CHECK: vfmsub213ps
21; CHECK: vfmsub213ps
22; CHECK: ret
23; CHECK_FMA4-LABEL: test_x86_fmsub_ps_y_wide
24; CHECK_FMA4: vfmsubps
25; CHECK_FMA4: vfmsubps
26; CHECK_FMA4: ret
27define <16 x float> @test_x86_fmsub_ps_y_wide(<16 x float> %a0, <16 x float> %a1, <16 x float> %a2) {
28  %x = fmul <16 x float> %a0, %a1
29  %res = fsub <16 x float> %x, %a2
30  ret <16 x float> %res
31}
32
33; CHECK-LABEL: test_x86_fnmadd_ps_y_wide
34; CHECK: vfnmadd213ps
35; CHECK: vfnmadd213ps
36; CHECK: ret
37; CHECK_FMA4-LABEL: test_x86_fnmadd_ps_y_wide
38; CHECK_FMA4: vfnmaddps
39; CHECK_FMA4: vfnmaddps
40; CHECK_FMA4: ret
41define <16 x float> @test_x86_fnmadd_ps_y_wide(<16 x float> %a0, <16 x float> %a1, <16 x float> %a2) {
42  %x = fmul <16 x float> %a0, %a1
43  %res = fsub <16 x float> %a2, %x
44  ret <16 x float> %res
45}
46
47; CHECK-LABEL: test_x86_fnmsub_ps_y_wide
48; CHECK: vfnmsub213ps
49; CHECK: vfnmsub213ps
50; CHECK: ret
51define <16 x float> @test_x86_fnmsub_ps_y_wide(<16 x float> %a0, <16 x float> %a1, <16 x float> %a2) {
52  %x = fmul <16 x float> %a0, %a1
53  %y = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %x
54  %res = fsub <16 x float> %y, %a2
55  ret <16 x float> %res
56}
57
58; CHECK-LABEL: test_x86_fmadd_pd_y_wide
59; CHECK: vfmadd213pd
60; CHECK: vfmadd213pd
61; CHECK: ret
62; CHECK_FMA4-LABEL: test_x86_fmadd_pd_y_wide
63; CHECK_FMA4: vfmaddpd
64; CHECK_FMA4: vfmaddpd
65; CHECK_FMA4: ret
66define <8 x double> @test_x86_fmadd_pd_y_wide(<8 x double> %a0, <8 x double> %a1, <8 x double> %a2) {
67  %x = fmul <8 x double> %a0, %a1
68  %res = fadd <8 x double> %x, %a2
69  ret <8 x double> %res
70}
71
72; CHECK-LABEL: test_x86_fmsub_pd_y_wide
73; CHECK: vfmsub213pd
74; CHECK: vfmsub213pd
75; CHECK: ret
76; CHECK_FMA4-LABEL: test_x86_fmsub_pd_y_wide
77; CHECK_FMA4: vfmsubpd
78; CHECK_FMA4: vfmsubpd
79; CHECK_FMA4: ret
80define <8 x double> @test_x86_fmsub_pd_y_wide(<8 x double> %a0, <8 x double> %a1, <8 x double> %a2) {
81  %x = fmul <8 x double> %a0, %a1
82  %res = fsub <8 x double> %x, %a2
83  ret <8 x double> %res
84}
85