Add tests for new NEON vld instructions.

llvm-svn: 78264
2009-08-06 00:38:31 +00:00 · 2009-08-06 00:38:31 +00:00 · 3389c2f7d0
parent 488db94e7b
commit 3389c2f7d0
3 changed files with 153 additions and 0 deletions
--- a/llvm/test/CodeGen/ARM/vld2.ll
+++ b/llvm/test/CodeGen/ARM/vld2.ll
@ -0,0 +1,51 @@
+; RUN: llvm-as < %s | llc -march=arm -mattr=+neon | FileCheck %s
+
+%struct.__builtin_neon_v8qi2 = type { <8 x i8>,  <8 x i8> }
+%struct.__builtin_neon_v4hi2 = type { <4 x i16>, <4 x i16> }
+%struct.__builtin_neon_v2si2 = type { <2 x i32>, <2 x i32> }
+%struct.__builtin_neon_v2sf2 = type { <2 x float>, <2 x float> }
+
+define <8 x i8> @vld2i8(i8* %A) nounwind {
+;CHECK: vld2i8:
+;CHECK: vld2.8
+	%tmp1 = call %struct.__builtin_neon_v8qi2 @llvm.arm.neon.vld2i.v8i8(i8* %A)
+        %tmp2 = extractvalue %struct.__builtin_neon_v8qi2 %tmp1, 0
+        %tmp3 = extractvalue %struct.__builtin_neon_v8qi2 %tmp1, 1
+        %tmp4 = add <8 x i8> %tmp2, %tmp3
+	ret <8 x i8> %tmp4
+}
+
+define <4 x i16> @vld2i16(i16* %A) nounwind {
+;CHECK: vld2i16:
+;CHECK: vld2.16
+	%tmp1 = call %struct.__builtin_neon_v4hi2 @llvm.arm.neon.vld2i.v4i16(i16* %A)
+        %tmp2 = extractvalue %struct.__builtin_neon_v4hi2 %tmp1, 0
+        %tmp3 = extractvalue %struct.__builtin_neon_v4hi2 %tmp1, 1
+        %tmp4 = add <4 x i16> %tmp2, %tmp3
+	ret <4 x i16> %tmp4
+}
+
+define <2 x i32> @vld2i32(i32* %A) nounwind {
+;CHECK: vld2i32:
+;CHECK: vld2.32
+	%tmp1 = call %struct.__builtin_neon_v2si2 @llvm.arm.neon.vld2i.v2i32(i32* %A)
+        %tmp2 = extractvalue %struct.__builtin_neon_v2si2 %tmp1, 0
+        %tmp3 = extractvalue %struct.__builtin_neon_v2si2 %tmp1, 1
+        %tmp4 = add <2 x i32> %tmp2, %tmp3
+	ret <2 x i32> %tmp4
+}
+
+define <2 x float> @vld2f(float* %A) nounwind {
+;CHECK: vld2f:
+;CHECK: vld2.32
+	%tmp1 = call %struct.__builtin_neon_v2sf2 @llvm.arm.neon.vld2f.v2f32(float* %A)
+        %tmp2 = extractvalue %struct.__builtin_neon_v2sf2 %tmp1, 0
+        %tmp3 = extractvalue %struct.__builtin_neon_v2sf2 %tmp1, 1
+        %tmp4 = add <2 x float> %tmp2, %tmp3
+	ret <2 x float> %tmp4
+}
+
+declare %struct.__builtin_neon_v8qi2 @llvm.arm.neon.vld2i.v8i8(i8*) nounwind readonly
+declare %struct.__builtin_neon_v4hi2 @llvm.arm.neon.vld2i.v4i16(i8*) nounwind readonly
+declare %struct.__builtin_neon_v2si2 @llvm.arm.neon.vld2i.v2i32(i8*) nounwind readonly
+declare %struct.__builtin_neon_v2sf2 @llvm.arm.neon.vld2f.v2f32(i8*) nounwind readonly
--- a/llvm/test/CodeGen/ARM/vld3.ll
+++ b/llvm/test/CodeGen/ARM/vld3.ll
@ -0,0 +1,51 @@
+; RUN: llvm-as < %s | llc -march=arm -mattr=+neon | FileCheck %s
+
+%struct.__builtin_neon_v8qi3 = type { <8 x i8>,  <8 x i8>,  <8 x i8> }
+%struct.__builtin_neon_v4hi3 = type { <4 x i16>, <4 x i16>, <4 x i16> }
+%struct.__builtin_neon_v2si3 = type { <2 x i32>, <2 x i32>, <2 x i32> }
+%struct.__builtin_neon_v2sf3 = type { <2 x float>, <2 x float>, <2 x float> }
+
+define <8 x i8> @vld3i8(i8* %A) nounwind {
+;CHECK: vld3i8:
+;CHECK: vld3.8
+	%tmp1 = call %struct.__builtin_neon_v8qi3 @llvm.arm.neon.vld3i.v8i8(i8* %A)
+        %tmp2 = extractvalue %struct.__builtin_neon_v8qi3 %tmp1, 0
+        %tmp3 = extractvalue %struct.__builtin_neon_v8qi3 %tmp1, 2
+        %tmp4 = add <8 x i8> %tmp2, %tmp3
+	ret <8 x i8> %tmp4
+}
+
+define <4 x i16> @vld3i16(i16* %A) nounwind {
+;CHECK: vld3i16:
+;CHECK: vld3.16
+	%tmp1 = call %struct.__builtin_neon_v4hi3 @llvm.arm.neon.vld3i.v4i16(i16* %A)
+        %tmp2 = extractvalue %struct.__builtin_neon_v4hi3 %tmp1, 0
+        %tmp3 = extractvalue %struct.__builtin_neon_v4hi3 %tmp1, 2
+        %tmp4 = add <4 x i16> %tmp2, %tmp3
+	ret <4 x i16> %tmp4
+}
+
+define <2 x i32> @vld3i32(i32* %A) nounwind {
+;CHECK: vld3i32:
+;CHECK: vld3.32
+	%tmp1 = call %struct.__builtin_neon_v2si3 @llvm.arm.neon.vld3i.v2i32(i32* %A)
+        %tmp2 = extractvalue %struct.__builtin_neon_v2si3 %tmp1, 0
+        %tmp3 = extractvalue %struct.__builtin_neon_v2si3 %tmp1, 2
+        %tmp4 = add <2 x i32> %tmp2, %tmp3
+	ret <2 x i32> %tmp4
+}
+
+define <2 x float> @vld3f(float* %A) nounwind {
+;CHECK: vld3f:
+;CHECK: vld3.32
+	%tmp1 = call %struct.__builtin_neon_v2sf3 @llvm.arm.neon.vld3f.v2f32(float* %A)
+        %tmp2 = extractvalue %struct.__builtin_neon_v2sf3 %tmp1, 0
+        %tmp3 = extractvalue %struct.__builtin_neon_v2sf3 %tmp1, 2
+        %tmp4 = add <2 x float> %tmp2, %tmp3
+	ret <2 x float> %tmp4
+}
+
+declare %struct.__builtin_neon_v8qi3 @llvm.arm.neon.vld3i.v8i8(i8*) nounwind readonly
+declare %struct.__builtin_neon_v4hi3 @llvm.arm.neon.vld3i.v4i16(i8*) nounwind readonly
+declare %struct.__builtin_neon_v2si3 @llvm.arm.neon.vld3i.v2i32(i8*) nounwind readonly
+declare %struct.__builtin_neon_v2sf3 @llvm.arm.neon.vld3f.v2f32(i8*) nounwind readonly
--- a/llvm/test/CodeGen/ARM/vld4.ll
+++ b/llvm/test/CodeGen/ARM/vld4.ll
@ -0,0 +1,51 @@
+; RUN: llvm-as < %s | llc -march=arm -mattr=+neon | FileCheck %s
+
+%struct.__builtin_neon_v8qi4 = type { <8 x i8>,  <8 x i8>,  <8 x i8>, <8 x i8> }
+%struct.__builtin_neon_v4hi4 = type { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> }
+%struct.__builtin_neon_v2si4 = type { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> }
+%struct.__builtin_neon_v2sf4 = type { <2 x float>, <2 x float>, <2 x float>, <2 x float> }
+
+define <8 x i8> @vld4i8(i8* %A) nounwind {
+;CHECK: vld4i8:
+;CHECK: vld4.8
+	%tmp1 = call %struct.__builtin_neon_v8qi4 @llvm.arm.neon.vld4i.v8i8(i8* %A)
+        %tmp2 = extractvalue %struct.__builtin_neon_v8qi4 %tmp1, 0
+        %tmp3 = extractvalue %struct.__builtin_neon_v8qi4 %tmp1, 2
+        %tmp4 = add <8 x i8> %tmp2, %tmp3
+	ret <8 x i8> %tmp4
+}
+
+define <4 x i16> @vld4i16(i16* %A) nounwind {
+;CHECK: vld4i16:
+;CHECK: vld4.16
+	%tmp1 = call %struct.__builtin_neon_v4hi4 @llvm.arm.neon.vld4i.v4i16(i16* %A)
+        %tmp2 = extractvalue %struct.__builtin_neon_v4hi4 %tmp1, 0
+        %tmp3 = extractvalue %struct.__builtin_neon_v4hi4 %tmp1, 2
+        %tmp4 = add <4 x i16> %tmp2, %tmp3
+	ret <4 x i16> %tmp4
+}
+
+define <2 x i32> @vld4i32(i32* %A) nounwind {
+;CHECK: vld4i32:
+;CHECK: vld4.32
+	%tmp1 = call %struct.__builtin_neon_v2si4 @llvm.arm.neon.vld4i.v2i32(i32* %A)
+        %tmp2 = extractvalue %struct.__builtin_neon_v2si4 %tmp1, 0
+        %tmp3 = extractvalue %struct.__builtin_neon_v2si4 %tmp1, 2
+        %tmp4 = add <2 x i32> %tmp2, %tmp3
+	ret <2 x i32> %tmp4
+}
+
+define <2 x float> @vld4f(float* %A) nounwind {
+;CHECK: vld4f:
+;CHECK: vld4.32
+	%tmp1 = call %struct.__builtin_neon_v2sf4 @llvm.arm.neon.vld4f.v2f32(float* %A)
+        %tmp2 = extractvalue %struct.__builtin_neon_v2sf4 %tmp1, 0
+        %tmp3 = extractvalue %struct.__builtin_neon_v2sf4 %tmp1, 2
+        %tmp4 = add <2 x float> %tmp2, %tmp3
+	ret <2 x float> %tmp4
+}
+
+declare %struct.__builtin_neon_v8qi4 @llvm.arm.neon.vld4i.v8i8(i8*) nounwind readonly
+declare %struct.__builtin_neon_v4hi4 @llvm.arm.neon.vld4i.v4i16(i8*) nounwind readonly
+declare %struct.__builtin_neon_v2si4 @llvm.arm.neon.vld4i.v2i32(i8*) nounwind readonly
+declare %struct.__builtin_neon_v2sf4 @llvm.arm.neon.vld4f.v2f32(i8*) nounwind readonly