1
; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s
3
define void @vst4i8(i8* %A, <8 x i8>* %B) nounwind {
6
%tmp1 = load <8 x i8>* %B
7
call void @llvm.arm.neon.vst4.v8i8(i8* %A, <8 x i8> %tmp1, <8 x i8> %tmp1, <8 x i8> %tmp1, <8 x i8> %tmp1)
11
define void @vst4i16(i16* %A, <4 x i16>* %B) nounwind {
14
%tmp1 = load <4 x i16>* %B
15
call void @llvm.arm.neon.vst4.v4i16(i16* %A, <4 x i16> %tmp1, <4 x i16> %tmp1, <4 x i16> %tmp1, <4 x i16> %tmp1)
19
define void @vst4i32(i32* %A, <2 x i32>* %B) nounwind {
22
%tmp1 = load <2 x i32>* %B
23
call void @llvm.arm.neon.vst4.v2i32(i32* %A, <2 x i32> %tmp1, <2 x i32> %tmp1, <2 x i32> %tmp1, <2 x i32> %tmp1)
27
define void @vst4f(float* %A, <2 x float>* %B) nounwind {
30
%tmp1 = load <2 x float>* %B
31
call void @llvm.arm.neon.vst4.v2f32(float* %A, <2 x float> %tmp1, <2 x float> %tmp1, <2 x float> %tmp1, <2 x float> %tmp1)
35
define void @vst4i64(i64* %A, <1 x i64>* %B) nounwind {
38
%tmp1 = load <1 x i64>* %B
39
call void @llvm.arm.neon.vst4.v1i64(i64* %A, <1 x i64> %tmp1, <1 x i64> %tmp1, <1 x i64> %tmp1, <1 x i64> %tmp1)
43
define void @vst4Qi8(i8* %A, <16 x i8>* %B) nounwind {
47
%tmp1 = load <16 x i8>* %B
48
call void @llvm.arm.neon.vst4.v16i8(i8* %A, <16 x i8> %tmp1, <16 x i8> %tmp1, <16 x i8> %tmp1, <16 x i8> %tmp1)
52
define void @vst4Qi16(i16* %A, <8 x i16>* %B) nounwind {
56
%tmp1 = load <8 x i16>* %B
57
call void @llvm.arm.neon.vst4.v8i16(i16* %A, <8 x i16> %tmp1, <8 x i16> %tmp1, <8 x i16> %tmp1, <8 x i16> %tmp1)
61
define void @vst4Qi32(i32* %A, <4 x i32>* %B) nounwind {
65
%tmp1 = load <4 x i32>* %B
66
call void @llvm.arm.neon.vst4.v4i32(i32* %A, <4 x i32> %tmp1, <4 x i32> %tmp1, <4 x i32> %tmp1, <4 x i32> %tmp1)
70
define void @vst4Qf(float* %A, <4 x float>* %B) nounwind {
74
%tmp1 = load <4 x float>* %B
75
call void @llvm.arm.neon.vst4.v4f32(float* %A, <4 x float> %tmp1, <4 x float> %tmp1, <4 x float> %tmp1, <4 x float> %tmp1)
79
declare void @llvm.arm.neon.vst4.v8i8(i8*, <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8>) nounwind
80
declare void @llvm.arm.neon.vst4.v4i16(i8*, <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16>) nounwind
81
declare void @llvm.arm.neon.vst4.v2i32(i8*, <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32>) nounwind
82
declare void @llvm.arm.neon.vst4.v2f32(i8*, <2 x float>, <2 x float>, <2 x float>, <2 x float>) nounwind
83
declare void @llvm.arm.neon.vst4.v1i64(i8*, <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64>) nounwind
85
declare void @llvm.arm.neon.vst4.v16i8(i8*, <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8>) nounwind
86
declare void @llvm.arm.neon.vst4.v8i16(i8*, <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16>) nounwind
87
declare void @llvm.arm.neon.vst4.v4i32(i8*, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>) nounwind
88
declare void @llvm.arm.neon.vst4.v4f32(i8*, <4 x float>, <4 x float>, <4 x float>, <4 x float>) nounwind