1 ; RUN: llc -mtriple=arm-eabi -mattr=+neon %s -o - | FileCheck %s 2 3 define void @vst1i8(i8* %A, <8 x i8>* %B) nounwind { 4 ;CHECK-LABEL: vst1i8: 5 ;Check the alignment value. Max for this instruction is 64 bits: 6 ;CHECK: vst1.8 {d16}, [r0:64] 7 %tmp1 = load <8 x i8>* %B 8 call void @llvm.arm.neon.vst1.v8i8(i8* %A, <8 x i8> %tmp1, i32 16) 9 ret void 10 } 11 12 define void @vst1i16(i16* %A, <4 x i16>* %B) nounwind { 13 ;CHECK-LABEL: vst1i16: 14 ;CHECK: vst1.16 15 %tmp0 = bitcast i16* %A to i8* 16 %tmp1 = load <4 x i16>* %B 17 call void @llvm.arm.neon.vst1.v4i16(i8* %tmp0, <4 x i16> %tmp1, i32 1) 18 ret void 19 } 20 21 define void @vst1i32(i32* %A, <2 x i32>* %B) nounwind { 22 ;CHECK-LABEL: vst1i32: 23 ;CHECK: vst1.32 24 %tmp0 = bitcast i32* %A to i8* 25 %tmp1 = load <2 x i32>* %B 26 call void @llvm.arm.neon.vst1.v2i32(i8* %tmp0, <2 x i32> %tmp1, i32 1) 27 ret void 28 } 29 30 define void @vst1f(float* %A, <2 x float>* %B) nounwind { 31 ;CHECK-LABEL: vst1f: 32 ;CHECK: vst1.32 33 %tmp0 = bitcast float* %A to i8* 34 %tmp1 = load <2 x float>* %B 35 call void @llvm.arm.neon.vst1.v2f32(i8* %tmp0, <2 x float> %tmp1, i32 1) 36 ret void 37 } 38 39 ;Check for a post-increment updating store. 40 define void @vst1f_update(float** %ptr, <2 x float>* %B) nounwind { 41 ;CHECK-LABEL: vst1f_update: 42 ;CHECK: vst1.32 {d16}, [r1]! 43 %A = load float** %ptr 44 %tmp0 = bitcast float* %A to i8* 45 %tmp1 = load <2 x float>* %B 46 call void @llvm.arm.neon.vst1.v2f32(i8* %tmp0, <2 x float> %tmp1, i32 1) 47 %tmp2 = getelementptr float* %A, i32 2 48 store float* %tmp2, float** %ptr 49 ret void 50 } 51 52 define void @vst1i64(i64* %A, <1 x i64>* %B) nounwind { 53 ;CHECK-LABEL: vst1i64: 54 ;CHECK: vst1.64 55 %tmp0 = bitcast i64* %A to i8* 56 %tmp1 = load <1 x i64>* %B 57 call void @llvm.arm.neon.vst1.v1i64(i8* %tmp0, <1 x i64> %tmp1, i32 1) 58 ret void 59 } 60 61 define void @vst1Qi8(i8* %A, <16 x i8>* %B) nounwind { 62 ;CHECK-LABEL: vst1Qi8: 63 ;Check the alignment value. Max for this instruction is 128 bits: 64 ;CHECK: vst1.8 {d16, d17}, [r0:64] 65 %tmp1 = load <16 x i8>* %B 66 call void @llvm.arm.neon.vst1.v16i8(i8* %A, <16 x i8> %tmp1, i32 8) 67 ret void 68 } 69 70 define void @vst1Qi16(i16* %A, <8 x i16>* %B) nounwind { 71 ;CHECK-LABEL: vst1Qi16: 72 ;Check the alignment value. Max for this instruction is 128 bits: 73 ;CHECK: vst1.16 {d16, d17}, [r0:128] 74 %tmp0 = bitcast i16* %A to i8* 75 %tmp1 = load <8 x i16>* %B 76 call void @llvm.arm.neon.vst1.v8i16(i8* %tmp0, <8 x i16> %tmp1, i32 32) 77 ret void 78 } 79 80 ;Check for a post-increment updating store with register increment. 81 define void @vst1Qi16_update(i16** %ptr, <8 x i16>* %B, i32 %inc) nounwind { 82 ;CHECK-LABEL: vst1Qi16_update: 83 ;CHECK: vst1.16 {d16, d17}, [r1:64], r2 84 %A = load i16** %ptr 85 %tmp0 = bitcast i16* %A to i8* 86 %tmp1 = load <8 x i16>* %B 87 call void @llvm.arm.neon.vst1.v8i16(i8* %tmp0, <8 x i16> %tmp1, i32 8) 88 %tmp2 = getelementptr i16* %A, i32 %inc 89 store i16* %tmp2, i16** %ptr 90 ret void 91 } 92 93 define void @vst1Qi32(i32* %A, <4 x i32>* %B) nounwind { 94 ;CHECK-LABEL: vst1Qi32: 95 ;CHECK: vst1.32 96 %tmp0 = bitcast i32* %A to i8* 97 %tmp1 = load <4 x i32>* %B 98 call void @llvm.arm.neon.vst1.v4i32(i8* %tmp0, <4 x i32> %tmp1, i32 1) 99 ret void 100 } 101 102 define void @vst1Qf(float* %A, <4 x float>* %B) nounwind { 103 ;CHECK-LABEL: vst1Qf: 104 ;CHECK: vst1.32 105 %tmp0 = bitcast float* %A to i8* 106 %tmp1 = load <4 x float>* %B 107 call void @llvm.arm.neon.vst1.v4f32(i8* %tmp0, <4 x float> %tmp1, i32 1) 108 ret void 109 } 110 111 define void @vst1Qi64(i64* %A, <2 x i64>* %B) nounwind { 112 ;CHECK-LABEL: vst1Qi64: 113 ;CHECK: vst1.64 114 %tmp0 = bitcast i64* %A to i8* 115 %tmp1 = load <2 x i64>* %B 116 call void @llvm.arm.neon.vst1.v2i64(i8* %tmp0, <2 x i64> %tmp1, i32 1) 117 ret void 118 } 119 120 declare void @llvm.arm.neon.vst1.v8i8(i8*, <8 x i8>, i32) nounwind 121 declare void @llvm.arm.neon.vst1.v4i16(i8*, <4 x i16>, i32) nounwind 122 declare void @llvm.arm.neon.vst1.v2i32(i8*, <2 x i32>, i32) nounwind 123 declare void @llvm.arm.neon.vst1.v2f32(i8*, <2 x float>, i32) nounwind 124 declare void @llvm.arm.neon.vst1.v1i64(i8*, <1 x i64>, i32) nounwind 125 126 declare void @llvm.arm.neon.vst1.v16i8(i8*, <16 x i8>, i32) nounwind 127 declare void @llvm.arm.neon.vst1.v8i16(i8*, <8 x i16>, i32) nounwind 128 declare void @llvm.arm.neon.vst1.v4i32(i8*, <4 x i32>, i32) nounwind 129 declare void @llvm.arm.neon.vst1.v4f32(i8*, <4 x float>, i32) nounwind 130 declare void @llvm.arm.neon.vst1.v2i64(i8*, <2 x i64>, i32) nounwind 131
