1 ; RUN: llvm-as < %s | llc -march=arm -mattr=+neon > %t
2 ; RUN: grep {vhadd\\.s8} %t | count 2
3 ; RUN: grep {vhadd\\.s16} %t | count 2
4 ; RUN: grep {vhadd\\.s32} %t | count 2
5 ; RUN: grep {vhadd\\.u8} %t | count 2
6 ; RUN: grep {vhadd\\.u16} %t | count 2
7 ; RUN: grep {vhadd\\.u32} %t | count 2
9 define <8 x i8> @vhadds8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
10 %tmp1 = load <8 x i8>* %A
11 %tmp2 = load <8 x i8>* %B
12 %tmp3 = call <8 x i8> @llvm.arm.neon.vhadds.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
16 define <4 x i16> @vhadds16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
17 %tmp1 = load <4 x i16>* %A
18 %tmp2 = load <4 x i16>* %B
19 %tmp3 = call <4 x i16> @llvm.arm.neon.vhadds.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
23 define <2 x i32> @vhadds32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
24 %tmp1 = load <2 x i32>* %A
25 %tmp2 = load <2 x i32>* %B
26 %tmp3 = call <2 x i32> @llvm.arm.neon.vhadds.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
30 define <8 x i8> @vhaddu8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
31 %tmp1 = load <8 x i8>* %A
32 %tmp2 = load <8 x i8>* %B
33 %tmp3 = call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
37 define <4 x i16> @vhaddu16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
38 %tmp1 = load <4 x i16>* %A
39 %tmp2 = load <4 x i16>* %B
40 %tmp3 = call <4 x i16> @llvm.arm.neon.vhaddu.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
44 define <2 x i32> @vhaddu32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
45 %tmp1 = load <2 x i32>* %A
46 %tmp2 = load <2 x i32>* %B
47 %tmp3 = call <2 x i32> @llvm.arm.neon.vhaddu.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
51 define <16 x i8> @vhaddQs8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
52 %tmp1 = load <16 x i8>* %A
53 %tmp2 = load <16 x i8>* %B
54 %tmp3 = call <16 x i8> @llvm.arm.neon.vhadds.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
58 define <8 x i16> @vhaddQs16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
59 %tmp1 = load <8 x i16>* %A
60 %tmp2 = load <8 x i16>* %B
61 %tmp3 = call <8 x i16> @llvm.arm.neon.vhadds.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
65 define <4 x i32> @vhaddQs32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
66 %tmp1 = load <4 x i32>* %A
67 %tmp2 = load <4 x i32>* %B
68 %tmp3 = call <4 x i32> @llvm.arm.neon.vhadds.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
72 define <16 x i8> @vhaddQu8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
73 %tmp1 = load <16 x i8>* %A
74 %tmp2 = load <16 x i8>* %B
75 %tmp3 = call <16 x i8> @llvm.arm.neon.vhaddu.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
79 define <8 x i16> @vhaddQu16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
80 %tmp1 = load <8 x i16>* %A
81 %tmp2 = load <8 x i16>* %B
82 %tmp3 = call <8 x i16> @llvm.arm.neon.vhaddu.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
86 define <4 x i32> @vhaddQu32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
87 %tmp1 = load <4 x i32>* %A
88 %tmp2 = load <4 x i32>* %B
89 %tmp3 = call <4 x i32> @llvm.arm.neon.vhaddu.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
93 declare <8 x i8> @llvm.arm.neon.vhadds.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
94 declare <4 x i16> @llvm.arm.neon.vhadds.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
95 declare <2 x i32> @llvm.arm.neon.vhadds.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
97 declare <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
98 declare <4 x i16> @llvm.arm.neon.vhaddu.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
99 declare <2 x i32> @llvm.arm.neon.vhaddu.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
101 declare <16 x i8> @llvm.arm.neon.vhadds.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
102 declare <8 x i16> @llvm.arm.neon.vhadds.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
103 declare <4 x i32> @llvm.arm.neon.vhadds.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
105 declare <16 x i8> @llvm.arm.neon.vhaddu.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
106 declare <8 x i16> @llvm.arm.neon.vhaddu.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
107 declare <4 x i32> @llvm.arm.neon.vhaddu.v4i32(<4 x i32>, <4 x i32>) nounwind readnone