1 ; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s
3 %struct.__neon_int8x8x2_t = type { <8 x i8>, <8 x i8> }
4 %struct.__neon_int16x4x2_t = type { <4 x i16>, <4 x i16> }
5 %struct.__neon_int32x2x2_t = type { <2 x i32>, <2 x i32> }
6 %struct.__neon_float32x2x2_t = type { <2 x float>, <2 x float> }
7 %struct.__neon_int64x1x2_t = type { <1 x i64>, <1 x i64> }
9 %struct.__neon_int8x16x2_t = type { <16 x i8>, <16 x i8> }
10 %struct.__neon_int16x8x2_t = type { <8 x i16>, <8 x i16> }
11 %struct.__neon_int32x4x2_t = type { <4 x i32>, <4 x i32> }
12 %struct.__neon_float32x4x2_t = type { <4 x float>, <4 x float> }
14 define <8 x i8> @vld2i8(i8* %A) nounwind {
17 %tmp1 = call %struct.__neon_int8x8x2_t @llvm.arm.neon.vld2.v8i8(i8* %A)
18 %tmp2 = extractvalue %struct.__neon_int8x8x2_t %tmp1, 0
19 %tmp3 = extractvalue %struct.__neon_int8x8x2_t %tmp1, 1
20 %tmp4 = add <8 x i8> %tmp2, %tmp3
24 define <4 x i16> @vld2i16(i16* %A) nounwind {
27 %tmp1 = call %struct.__neon_int16x4x2_t @llvm.arm.neon.vld2.v4i16(i16* %A)
28 %tmp2 = extractvalue %struct.__neon_int16x4x2_t %tmp1, 0
29 %tmp3 = extractvalue %struct.__neon_int16x4x2_t %tmp1, 1
30 %tmp4 = add <4 x i16> %tmp2, %tmp3
34 define <2 x i32> @vld2i32(i32* %A) nounwind {
37 %tmp1 = call %struct.__neon_int32x2x2_t @llvm.arm.neon.vld2.v2i32(i32* %A)
38 %tmp2 = extractvalue %struct.__neon_int32x2x2_t %tmp1, 0
39 %tmp3 = extractvalue %struct.__neon_int32x2x2_t %tmp1, 1
40 %tmp4 = add <2 x i32> %tmp2, %tmp3
44 define <2 x float> @vld2f(float* %A) nounwind {
47 %tmp1 = call %struct.__neon_float32x2x2_t @llvm.arm.neon.vld2.v2f32(float* %A)
48 %tmp2 = extractvalue %struct.__neon_float32x2x2_t %tmp1, 0
49 %tmp3 = extractvalue %struct.__neon_float32x2x2_t %tmp1, 1
50 %tmp4 = add <2 x float> %tmp2, %tmp3
54 define <1 x i64> @vld2i64(i64* %A) nounwind {
57 %tmp1 = call %struct.__neon_int64x1x2_t @llvm.arm.neon.vld2.v1i64(i64* %A)
58 %tmp2 = extractvalue %struct.__neon_int64x1x2_t %tmp1, 0
59 %tmp3 = extractvalue %struct.__neon_int64x1x2_t %tmp1, 1
60 %tmp4 = add <1 x i64> %tmp2, %tmp3
64 define <16 x i8> @vld2Qi8(i8* %A) nounwind {
67 %tmp1 = call %struct.__neon_int8x16x2_t @llvm.arm.neon.vld2.v16i8(i8* %A)
68 %tmp2 = extractvalue %struct.__neon_int8x16x2_t %tmp1, 0
69 %tmp3 = extractvalue %struct.__neon_int8x16x2_t %tmp1, 1
70 %tmp4 = add <16 x i8> %tmp2, %tmp3
74 define <8 x i16> @vld2Qi16(i16* %A) nounwind {
77 %tmp1 = call %struct.__neon_int16x8x2_t @llvm.arm.neon.vld2.v8i16(i16* %A)
78 %tmp2 = extractvalue %struct.__neon_int16x8x2_t %tmp1, 0
79 %tmp3 = extractvalue %struct.__neon_int16x8x2_t %tmp1, 1
80 %tmp4 = add <8 x i16> %tmp2, %tmp3
84 define <4 x i32> @vld2Qi32(i32* %A) nounwind {
87 %tmp1 = call %struct.__neon_int32x4x2_t @llvm.arm.neon.vld2.v4i32(i32* %A)
88 %tmp2 = extractvalue %struct.__neon_int32x4x2_t %tmp1, 0
89 %tmp3 = extractvalue %struct.__neon_int32x4x2_t %tmp1, 1
90 %tmp4 = add <4 x i32> %tmp2, %tmp3
94 define <4 x float> @vld2Qf(float* %A) nounwind {
97 %tmp1 = call %struct.__neon_float32x4x2_t @llvm.arm.neon.vld2.v4f32(float* %A)
98 %tmp2 = extractvalue %struct.__neon_float32x4x2_t %tmp1, 0
99 %tmp3 = extractvalue %struct.__neon_float32x4x2_t %tmp1, 1
100 %tmp4 = add <4 x float> %tmp2, %tmp3
101 ret <4 x float> %tmp4
104 declare %struct.__neon_int8x8x2_t @llvm.arm.neon.vld2.v8i8(i8*) nounwind readonly
105 declare %struct.__neon_int16x4x2_t @llvm.arm.neon.vld2.v4i16(i8*) nounwind readonly
106 declare %struct.__neon_int32x2x2_t @llvm.arm.neon.vld2.v2i32(i8*) nounwind readonly
107 declare %struct.__neon_float32x2x2_t @llvm.arm.neon.vld2.v2f32(i8*) nounwind readonly
108 declare %struct.__neon_int64x1x2_t @llvm.arm.neon.vld2.v1i64(i8*) nounwind readonly
110 declare %struct.__neon_int8x16x2_t @llvm.arm.neon.vld2.v16i8(i8*) nounwind readonly
111 declare %struct.__neon_int16x8x2_t @llvm.arm.neon.vld2.v8i16(i8*) nounwind readonly
112 declare %struct.__neon_int32x4x2_t @llvm.arm.neon.vld2.v4i32(i8*) nounwind readonly
113 declare %struct.__neon_float32x4x2_t @llvm.arm.neon.vld2.v4f32(i8*) nounwind readonly