1; RUN: llc < %s -march=x86-64 -mattr=+avx512f,+avx512bw | FileCheck %s
2
3define void @f(<16 x float> %A, <16 x float> %AA, i8* %B, <8 x double> %C, <8 x double> %CC, <8 x i64> %E, <8 x i64> %EE, <16 x i32> %F, <16 x i32> %FF, <32 x i16> %G, <32 x i16> %GG, <64 x i8> %H, <64 x i8> %HH) {
4; CHECK: vmovntps %z
5  %cast = bitcast i8* %B to <16 x float>*
6  %A2 = fadd <16 x float> %A, %AA
7  store <16 x float> %A2, <16 x float>* %cast, align 64, !nontemporal !0
8; CHECK: vmovntdq %z
9  %cast1 = bitcast i8* %B to <8 x i64>*
10  %E2 = add <8 x i64> %E, %EE
11  store <8 x i64> %E2, <8 x i64>* %cast1, align 64, !nontemporal !0
12; CHECK: vmovntpd %z
13  %cast2 = bitcast i8* %B to <8 x double>*
14  %C2 = fadd <8 x double> %C, %CC
15  store <8 x double> %C2, <8 x double>* %cast2, align 64, !nontemporal !0
16; CHECK: vmovntdq %z
17  %cast3 = bitcast i8* %B to <16 x i32>*
18  %F2 = add <16 x i32> %F, %FF
19  store <16 x i32> %F2, <16 x i32>* %cast3, align 64, !nontemporal !0
20; CHECK: vmovntdq %z
21  %cast4 = bitcast i8* %B to <32 x i16>*
22  %G2 = add <32 x i16> %G, %GG
23  store <32 x i16> %G2, <32 x i16>* %cast4, align 64, !nontemporal !0
24; CHECK: vmovntdq %z
25  %cast5 = bitcast i8* %B to <64 x i8>*
26  %H2 = add <64 x i8> %H, %HH
27  store <64 x i8> %H2, <64 x i8>* %cast5, align 64, !nontemporal !0
28  ret void
29}
30
31!0 = !{i32 1}
32