2014-04-03 16:01:44 +00:00
|
|
|
; RUN: llc -mtriple=arm-eabi -mattr=+neon -fast-isel=0 -O0 %s -o - | FileCheck %s
|
2009-08-06 18:47:44 +00:00
|
|
|
|
|
|
|
define void @vst3i8(i8* %A, <8 x i8>* %B) nounwind {
|
2013-07-14 06:24:09 +00:00
|
|
|
;CHECK-LABEL: vst3i8:
|
2010-09-23 23:42:37 +00:00
|
|
|
;Check the alignment value. Max for this instruction is 64 bits:
|
|
|
|
;This test runs at -O0 so do not check for specific register numbers.
|
2013-02-22 10:01:33 +00:00
|
|
|
;CHECK: vst3.8 {d{{.*}}, d{{.*}}, d{{.*}}}, [r{{.*}}:64]
|
2009-08-06 18:47:44 +00:00
|
|
|
%tmp1 = load <8 x i8>* %B
|
2010-09-23 23:42:37 +00:00
|
|
|
call void @llvm.arm.neon.vst3.v8i8(i8* %A, <8 x i8> %tmp1, <8 x i8> %tmp1, <8 x i8> %tmp1, i32 32)
|
2009-08-06 18:47:44 +00:00
|
|
|
ret void
|
|
|
|
}
|
|
|
|
|
|
|
|
define void @vst3i16(i16* %A, <4 x i16>* %B) nounwind {
|
2013-07-14 06:24:09 +00:00
|
|
|
;CHECK-LABEL: vst3i16:
|
2009-08-06 18:47:44 +00:00
|
|
|
;CHECK: vst3.16
|
2010-04-20 00:17:16 +00:00
|
|
|
%tmp0 = bitcast i16* %A to i8*
|
2009-08-06 18:47:44 +00:00
|
|
|
%tmp1 = load <4 x i16>* %B
|
2010-08-27 17:13:24 +00:00
|
|
|
call void @llvm.arm.neon.vst3.v4i16(i8* %tmp0, <4 x i16> %tmp1, <4 x i16> %tmp1, <4 x i16> %tmp1, i32 1)
|
2009-08-06 18:47:44 +00:00
|
|
|
ret void
|
|
|
|
}
|
|
|
|
|
|
|
|
define void @vst3i32(i32* %A, <2 x i32>* %B) nounwind {
|
2013-07-14 06:24:09 +00:00
|
|
|
;CHECK-LABEL: vst3i32:
|
2009-08-06 18:47:44 +00:00
|
|
|
;CHECK: vst3.32
|
2010-04-20 00:17:16 +00:00
|
|
|
%tmp0 = bitcast i32* %A to i8*
|
2009-08-06 18:47:44 +00:00
|
|
|
%tmp1 = load <2 x i32>* %B
|
2010-08-27 17:13:24 +00:00
|
|
|
call void @llvm.arm.neon.vst3.v2i32(i8* %tmp0, <2 x i32> %tmp1, <2 x i32> %tmp1, <2 x i32> %tmp1, i32 1)
|
2009-08-06 18:47:44 +00:00
|
|
|
ret void
|
|
|
|
}
|
|
|
|
|
2011-02-07 17:43:21 +00:00
|
|
|
;Check for a post-increment updating store.
|
|
|
|
define void @vst3i32_update(i32** %ptr, <2 x i32>* %B) nounwind {
|
2013-07-14 06:24:09 +00:00
|
|
|
;CHECK-LABEL: vst3i32_update:
|
2011-02-07 17:43:21 +00:00
|
|
|
;CHECK: vst3.32 {d{{.*}}, d{{.*}}, d{{.*}}}, [r{{.*}}]!
|
|
|
|
%A = load i32** %ptr
|
|
|
|
%tmp0 = bitcast i32* %A to i8*
|
|
|
|
%tmp1 = load <2 x i32>* %B
|
|
|
|
call void @llvm.arm.neon.vst3.v2i32(i8* %tmp0, <2 x i32> %tmp1, <2 x i32> %tmp1, <2 x i32> %tmp1, i32 1)
|
|
|
|
%tmp2 = getelementptr i32* %A, i32 6
|
|
|
|
store i32* %tmp2, i32** %ptr
|
|
|
|
ret void
|
|
|
|
}
|
|
|
|
|
2009-08-06 18:47:44 +00:00
|
|
|
define void @vst3f(float* %A, <2 x float>* %B) nounwind {
|
2013-07-14 06:24:09 +00:00
|
|
|
;CHECK-LABEL: vst3f:
|
2009-08-06 18:47:44 +00:00
|
|
|
;CHECK: vst3.32
|
2010-04-20 00:17:16 +00:00
|
|
|
%tmp0 = bitcast float* %A to i8*
|
2009-08-06 18:47:44 +00:00
|
|
|
%tmp1 = load <2 x float>* %B
|
2010-08-27 17:13:24 +00:00
|
|
|
call void @llvm.arm.neon.vst3.v2f32(i8* %tmp0, <2 x float> %tmp1, <2 x float> %tmp1, <2 x float> %tmp1, i32 1)
|
2009-08-06 18:47:44 +00:00
|
|
|
ret void
|
|
|
|
}
|
|
|
|
|
2009-10-08 00:28:28 +00:00
|
|
|
define void @vst3i64(i64* %A, <1 x i64>* %B) nounwind {
|
2013-07-14 06:24:09 +00:00
|
|
|
;CHECK-LABEL: vst3i64:
|
2010-09-23 23:42:37 +00:00
|
|
|
;Check the alignment value. Max for this instruction is 64 bits:
|
|
|
|
;This test runs at -O0 so do not check for specific register numbers.
|
2013-02-22 10:01:33 +00:00
|
|
|
;CHECK: vst1.64 {d{{.*}}, d{{.*}}, d{{.*}}}, [r{{.*}}:64]
|
2010-04-20 00:17:16 +00:00
|
|
|
%tmp0 = bitcast i64* %A to i8*
|
2009-10-08 00:28:28 +00:00
|
|
|
%tmp1 = load <1 x i64>* %B
|
2010-09-23 23:42:37 +00:00
|
|
|
call void @llvm.arm.neon.vst3.v1i64(i8* %tmp0, <1 x i64> %tmp1, <1 x i64> %tmp1, <1 x i64> %tmp1, i32 16)
|
2009-10-08 00:28:28 +00:00
|
|
|
ret void
|
|
|
|
}
|
|
|
|
|
2014-01-16 09:16:13 +00:00
|
|
|
define void @vst3i64_update(i64** %ptr, <1 x i64>* %B) nounwind {
|
|
|
|
;CHECK-LABEL: vst3i64_update
|
|
|
|
;CHECK: vst1.64 {d{{.*}}, d{{.*}}, d{{.*}}}, [r{{.*}}]!
|
|
|
|
%A = load i64** %ptr
|
|
|
|
%tmp0 = bitcast i64* %A to i8*
|
|
|
|
%tmp1 = load <1 x i64>* %B
|
|
|
|
call void @llvm.arm.neon.vst3.v1i64(i8* %tmp0, <1 x i64> %tmp1, <1 x i64> %tmp1, <1 x i64> %tmp1, i32 1)
|
|
|
|
%tmp2 = getelementptr i64* %A, i32 3
|
|
|
|
store i64* %tmp2, i64** %ptr
|
|
|
|
ret void
|
|
|
|
}
|
|
|
|
|
2009-10-07 20:30:08 +00:00
|
|
|
define void @vst3Qi8(i8* %A, <16 x i8>* %B) nounwind {
|
2013-07-14 06:24:09 +00:00
|
|
|
;CHECK-LABEL: vst3Qi8:
|
2010-09-23 23:42:37 +00:00
|
|
|
;Check the alignment value. Max for this instruction is 64 bits:
|
|
|
|
;This test runs at -O0 so do not check for specific register numbers.
|
2013-02-22 10:01:33 +00:00
|
|
|
;CHECK: vst3.8 {d{{.*}}, d{{.*}}, d{{.*}}}, [r{{.*}}:64]!
|
|
|
|
;CHECK: vst3.8 {d{{.*}}, d{{.*}}, d{{.*}}}, [r{{.*}}:64]
|
2009-10-07 20:30:08 +00:00
|
|
|
%tmp1 = load <16 x i8>* %B
|
2010-09-23 23:42:37 +00:00
|
|
|
call void @llvm.arm.neon.vst3.v16i8(i8* %A, <16 x i8> %tmp1, <16 x i8> %tmp1, <16 x i8> %tmp1, i32 32)
|
2009-10-07 20:30:08 +00:00
|
|
|
ret void
|
|
|
|
}
|
|
|
|
|
|
|
|
define void @vst3Qi16(i16* %A, <8 x i16>* %B) nounwind {
|
2013-07-14 06:24:09 +00:00
|
|
|
;CHECK-LABEL: vst3Qi16:
|
2009-10-07 20:30:08 +00:00
|
|
|
;CHECK: vst3.16
|
|
|
|
;CHECK: vst3.16
|
2010-04-20 00:17:16 +00:00
|
|
|
%tmp0 = bitcast i16* %A to i8*
|
2009-10-07 20:30:08 +00:00
|
|
|
%tmp1 = load <8 x i16>* %B
|
2010-08-27 17:13:24 +00:00
|
|
|
call void @llvm.arm.neon.vst3.v8i16(i8* %tmp0, <8 x i16> %tmp1, <8 x i16> %tmp1, <8 x i16> %tmp1, i32 1)
|
2009-10-07 20:30:08 +00:00
|
|
|
ret void
|
|
|
|
}
|
|
|
|
|
2011-02-07 17:43:21 +00:00
|
|
|
;Check for a post-increment updating store.
|
|
|
|
define void @vst3Qi16_update(i16** %ptr, <8 x i16>* %B) nounwind {
|
2013-07-14 06:24:09 +00:00
|
|
|
;CHECK-LABEL: vst3Qi16_update:
|
2011-02-07 17:43:21 +00:00
|
|
|
;CHECK: vst3.16 {d{{.*}}, d{{.*}}, d{{.*}}}, [r{{.*}}]!
|
|
|
|
;CHECK: vst3.16 {d{{.*}}, d{{.*}}, d{{.*}}}, [r{{.*}}]!
|
|
|
|
%A = load i16** %ptr
|
|
|
|
%tmp0 = bitcast i16* %A to i8*
|
|
|
|
%tmp1 = load <8 x i16>* %B
|
|
|
|
call void @llvm.arm.neon.vst3.v8i16(i8* %tmp0, <8 x i16> %tmp1, <8 x i16> %tmp1, <8 x i16> %tmp1, i32 1)
|
|
|
|
%tmp2 = getelementptr i16* %A, i32 24
|
|
|
|
store i16* %tmp2, i16** %ptr
|
|
|
|
ret void
|
|
|
|
}
|
|
|
|
|
2009-10-07 20:30:08 +00:00
|
|
|
define void @vst3Qi32(i32* %A, <4 x i32>* %B) nounwind {
|
2013-07-14 06:24:09 +00:00
|
|
|
;CHECK-LABEL: vst3Qi32:
|
2009-10-07 20:30:08 +00:00
|
|
|
;CHECK: vst3.32
|
|
|
|
;CHECK: vst3.32
|
2010-04-20 00:17:16 +00:00
|
|
|
%tmp0 = bitcast i32* %A to i8*
|
2009-10-07 20:30:08 +00:00
|
|
|
%tmp1 = load <4 x i32>* %B
|
2010-08-27 17:13:24 +00:00
|
|
|
call void @llvm.arm.neon.vst3.v4i32(i8* %tmp0, <4 x i32> %tmp1, <4 x i32> %tmp1, <4 x i32> %tmp1, i32 1)
|
2009-10-07 20:30:08 +00:00
|
|
|
ret void
|
|
|
|
}
|
|
|
|
|
|
|
|
define void @vst3Qf(float* %A, <4 x float>* %B) nounwind {
|
2013-07-14 06:24:09 +00:00
|
|
|
;CHECK-LABEL: vst3Qf:
|
2009-10-07 20:30:08 +00:00
|
|
|
;CHECK: vst3.32
|
|
|
|
;CHECK: vst3.32
|
2010-04-20 00:17:16 +00:00
|
|
|
%tmp0 = bitcast float* %A to i8*
|
2009-10-07 20:30:08 +00:00
|
|
|
%tmp1 = load <4 x float>* %B
|
2010-08-27 17:13:24 +00:00
|
|
|
call void @llvm.arm.neon.vst3.v4f32(i8* %tmp0, <4 x float> %tmp1, <4 x float> %tmp1, <4 x float> %tmp1, i32 1)
|
2009-10-07 20:30:08 +00:00
|
|
|
ret void
|
|
|
|
}
|
|
|
|
|
2010-08-27 17:13:24 +00:00
|
|
|
declare void @llvm.arm.neon.vst3.v8i8(i8*, <8 x i8>, <8 x i8>, <8 x i8>, i32) nounwind
|
|
|
|
declare void @llvm.arm.neon.vst3.v4i16(i8*, <4 x i16>, <4 x i16>, <4 x i16>, i32) nounwind
|
|
|
|
declare void @llvm.arm.neon.vst3.v2i32(i8*, <2 x i32>, <2 x i32>, <2 x i32>, i32) nounwind
|
|
|
|
declare void @llvm.arm.neon.vst3.v2f32(i8*, <2 x float>, <2 x float>, <2 x float>, i32) nounwind
|
|
|
|
declare void @llvm.arm.neon.vst3.v1i64(i8*, <1 x i64>, <1 x i64>, <1 x i64>, i32) nounwind
|
2009-10-07 20:30:08 +00:00
|
|
|
|
2010-08-27 17:13:24 +00:00
|
|
|
declare void @llvm.arm.neon.vst3.v16i8(i8*, <16 x i8>, <16 x i8>, <16 x i8>, i32) nounwind
|
|
|
|
declare void @llvm.arm.neon.vst3.v8i16(i8*, <8 x i16>, <8 x i16>, <8 x i16>, i32) nounwind
|
|
|
|
declare void @llvm.arm.neon.vst3.v4i32(i8*, <4 x i32>, <4 x i32>, <4 x i32>, i32) nounwind
|
|
|
|
declare void @llvm.arm.neon.vst3.v4f32(i8*, <4 x float>, <4 x float>, <4 x float>, i32) nounwind
|