1 ; RUN: llc < %s -march=arm64 -aarch64-neon-syntax=apple | FileCheck %s
3 define <8 x i8> @shsub8b(<8 x i8>* %A, <8 x i8>* %B) nounwind {
6 %tmp1 = load <8 x i8>, <8 x i8>* %A
7 %tmp2 = load <8 x i8>, <8 x i8>* %B
8 %tmp3 = call <8 x i8> @llvm.aarch64.neon.shsub.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
12 define <16 x i8> @shsub16b(<16 x i8>* %A, <16 x i8>* %B) nounwind {
13 ;CHECK-LABEL: shsub16b:
15 %tmp1 = load <16 x i8>, <16 x i8>* %A
16 %tmp2 = load <16 x i8>, <16 x i8>* %B
17 %tmp3 = call <16 x i8> @llvm.aarch64.neon.shsub.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
21 define <4 x i16> @shsub4h(<4 x i16>* %A, <4 x i16>* %B) nounwind {
22 ;CHECK-LABEL: shsub4h:
24 %tmp1 = load <4 x i16>, <4 x i16>* %A
25 %tmp2 = load <4 x i16>, <4 x i16>* %B
26 %tmp3 = call <4 x i16> @llvm.aarch64.neon.shsub.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
30 define <8 x i16> @shsub8h(<8 x i16>* %A, <8 x i16>* %B) nounwind {
31 ;CHECK-LABEL: shsub8h:
33 %tmp1 = load <8 x i16>, <8 x i16>* %A
34 %tmp2 = load <8 x i16>, <8 x i16>* %B
35 %tmp3 = call <8 x i16> @llvm.aarch64.neon.shsub.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
39 define <2 x i32> @shsub2s(<2 x i32>* %A, <2 x i32>* %B) nounwind {
40 ;CHECK-LABEL: shsub2s:
42 %tmp1 = load <2 x i32>, <2 x i32>* %A
43 %tmp2 = load <2 x i32>, <2 x i32>* %B
44 %tmp3 = call <2 x i32> @llvm.aarch64.neon.shsub.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
48 define <4 x i32> @shsub4s(<4 x i32>* %A, <4 x i32>* %B) nounwind {
49 ;CHECK-LABEL: shsub4s:
51 %tmp1 = load <4 x i32>, <4 x i32>* %A
52 %tmp2 = load <4 x i32>, <4 x i32>* %B
53 %tmp3 = call <4 x i32> @llvm.aarch64.neon.shsub.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
57 define <8 x i8> @uhsub8b(<8 x i8>* %A, <8 x i8>* %B) nounwind {
58 ;CHECK-LABEL: uhsub8b:
60 %tmp1 = load <8 x i8>, <8 x i8>* %A
61 %tmp2 = load <8 x i8>, <8 x i8>* %B
62 %tmp3 = call <8 x i8> @llvm.aarch64.neon.uhsub.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
66 define <16 x i8> @uhsub16b(<16 x i8>* %A, <16 x i8>* %B) nounwind {
67 ;CHECK-LABEL: uhsub16b:
69 %tmp1 = load <16 x i8>, <16 x i8>* %A
70 %tmp2 = load <16 x i8>, <16 x i8>* %B
71 %tmp3 = call <16 x i8> @llvm.aarch64.neon.uhsub.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
75 define <4 x i16> @uhsub4h(<4 x i16>* %A, <4 x i16>* %B) nounwind {
76 ;CHECK-LABEL: uhsub4h:
78 %tmp1 = load <4 x i16>, <4 x i16>* %A
79 %tmp2 = load <4 x i16>, <4 x i16>* %B
80 %tmp3 = call <4 x i16> @llvm.aarch64.neon.uhsub.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
84 define <8 x i16> @uhsub8h(<8 x i16>* %A, <8 x i16>* %B) nounwind {
85 ;CHECK-LABEL: uhsub8h:
87 %tmp1 = load <8 x i16>, <8 x i16>* %A
88 %tmp2 = load <8 x i16>, <8 x i16>* %B
89 %tmp3 = call <8 x i16> @llvm.aarch64.neon.uhsub.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
93 define <2 x i32> @uhsub2s(<2 x i32>* %A, <2 x i32>* %B) nounwind {
94 ;CHECK-LABEL: uhsub2s:
96 %tmp1 = load <2 x i32>, <2 x i32>* %A
97 %tmp2 = load <2 x i32>, <2 x i32>* %B
98 %tmp3 = call <2 x i32> @llvm.aarch64.neon.uhsub.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
102 define <4 x i32> @uhsub4s(<4 x i32>* %A, <4 x i32>* %B) nounwind {
103 ;CHECK-LABEL: uhsub4s:
105 %tmp1 = load <4 x i32>, <4 x i32>* %A
106 %tmp2 = load <4 x i32>, <4 x i32>* %B
107 %tmp3 = call <4 x i32> @llvm.aarch64.neon.uhsub.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
111 declare <8 x i8> @llvm.aarch64.neon.shsub.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
112 declare <4 x i16> @llvm.aarch64.neon.shsub.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
113 declare <2 x i32> @llvm.aarch64.neon.shsub.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
115 declare <8 x i8> @llvm.aarch64.neon.uhsub.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
116 declare <4 x i16> @llvm.aarch64.neon.uhsub.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
117 declare <2 x i32> @llvm.aarch64.neon.uhsub.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
119 declare <16 x i8> @llvm.aarch64.neon.shsub.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
120 declare <8 x i16> @llvm.aarch64.neon.shsub.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
121 declare <4 x i32> @llvm.aarch64.neon.shsub.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
123 declare <16 x i8> @llvm.aarch64.neon.uhsub.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
124 declare <8 x i16> @llvm.aarch64.neon.uhsub.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
125 declare <4 x i32> @llvm.aarch64.neon.uhsub.v4i32(<4 x i32>, <4 x i32>) nounwind readnone