| Bob Wilson | 2ed3346 | 2009-07-08 20:32:02 +0000 | [diff] [blame^] | 1 | ; RUN: llvm-as < %s | llc -march=arm -mattr=+neon > %t |
| 2 | ; RUN: grep {vst1\\.8} %t | count 2 |
| 3 | ; RUN: grep {vst1\\.16} %t | count 2 |
| 4 | ; RUN: grep {vst1\\.32} %t | count 4 |
| 5 | ; RUN: grep {vst1\\.64} %t | count 2 |
| 6 | |
| 7 | define void @vst1i8(i8* %A, <8 x i8>* %B) nounwind { |
| 8 | %tmp1 = load <8 x i8>* %B |
| 9 | call void @llvm.arm.neon.vsti.v8i8(i8* %A, <8 x i8> %tmp1, i32 1) |
| 10 | ret void |
| 11 | } |
| 12 | |
| 13 | define void @vst1i16(i16* %A, <4 x i16>* %B) nounwind { |
| 14 | %tmp1 = load <4 x i16>* %B |
| 15 | call void @llvm.arm.neon.vsti.v4i16(i16* %A, <4 x i16> %tmp1, i32 1) |
| 16 | ret void |
| 17 | } |
| 18 | |
| 19 | define void @vst1i32(i32* %A, <2 x i32>* %B) nounwind { |
| 20 | %tmp1 = load <2 x i32>* %B |
| 21 | call void @llvm.arm.neon.vsti.v2i32(i32* %A, <2 x i32> %tmp1, i32 1) |
| 22 | ret void |
| 23 | } |
| 24 | |
| 25 | define void @vst1f(float* %A, <2 x float>* %B) nounwind { |
| 26 | %tmp1 = load <2 x float>* %B |
| 27 | call void @llvm.arm.neon.vstf.v2f32(float* %A, <2 x float> %tmp1, i32 1) |
| 28 | ret void |
| 29 | } |
| 30 | |
| 31 | define void @vst1i64(i64* %A, <1 x i64>* %B) nounwind { |
| 32 | %tmp1 = load <1 x i64>* %B |
| 33 | call void @llvm.arm.neon.vsti.v1i64(i64* %A, <1 x i64> %tmp1, i32 1) |
| 34 | ret void |
| 35 | } |
| 36 | |
| 37 | define void @vst1Qi8(i8* %A, <16 x i8>* %B) nounwind { |
| 38 | %tmp1 = load <16 x i8>* %B |
| 39 | call void @llvm.arm.neon.vsti.v16i8(i8* %A, <16 x i8> %tmp1, i32 1) |
| 40 | ret void |
| 41 | } |
| 42 | |
| 43 | define void @vst1Qi16(i16* %A, <8 x i16>* %B) nounwind { |
| 44 | %tmp1 = load <8 x i16>* %B |
| 45 | call void @llvm.arm.neon.vsti.v8i16(i16* %A, <8 x i16> %tmp1, i32 1) |
| 46 | ret void |
| 47 | } |
| 48 | |
| 49 | define void @vst1Qi32(i32* %A, <4 x i32>* %B) nounwind { |
| 50 | %tmp1 = load <4 x i32>* %B |
| 51 | call void @llvm.arm.neon.vsti.v4i32(i32* %A, <4 x i32> %tmp1, i32 1) |
| 52 | ret void |
| 53 | } |
| 54 | |
| 55 | define void @vst1Qf(float* %A, <4 x float>* %B) nounwind { |
| 56 | %tmp1 = load <4 x float>* %B |
| 57 | call void @llvm.arm.neon.vstf.v4f32(float* %A, <4 x float> %tmp1, i32 1) |
| 58 | ret void |
| 59 | } |
| 60 | |
| 61 | define void @vst1Qi64(i64* %A, <2 x i64>* %B) nounwind { |
| 62 | %tmp1 = load <2 x i64>* %B |
| 63 | call void @llvm.arm.neon.vsti.v2i64(i64* %A, <2 x i64> %tmp1, i32 1) |
| 64 | ret void |
| 65 | } |
| 66 | |
| 67 | declare void @llvm.arm.neon.vsti.v8i8(i8*, <8 x i8>, i32) nounwind readnone |
| 68 | declare void @llvm.arm.neon.vsti.v4i16(i16*, <4 x i16>, i32) nounwind readnone |
| 69 | declare void @llvm.arm.neon.vsti.v2i32(i32*, <2 x i32>, i32) nounwind readnone |
| 70 | declare void @llvm.arm.neon.vstf.v2f32(float*, <2 x float>, i32) nounwind readnone |
| 71 | declare void @llvm.arm.neon.vsti.v1i64(i64*, <1 x i64>, i32) nounwind readnone |
| 72 | |
| 73 | declare void @llvm.arm.neon.vsti.v16i8(i8*, <16 x i8>, i32) nounwind readnone |
| 74 | declare void @llvm.arm.neon.vsti.v8i16(i16*, <8 x i16>, i32) nounwind readnone |
| 75 | declare void @llvm.arm.neon.vsti.v4i32(i32*, <4 x i32>, i32) nounwind readnone |
| 76 | declare void @llvm.arm.neon.vstf.v4f32(float*, <4 x float>, i32) nounwind readnone |
| 77 | declare void @llvm.arm.neon.vsti.v2i64(i64*, <2 x i64>, i32) nounwind readnone |