1 ; RUN: llc < %s -mtriple=arm64-eabi -aarch64-neon-syntax=apple | FileCheck %s
3 define <8 x i8> @sqsub8b(<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.sqsub.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
12 define <4 x i16> @sqsub4h(<4 x i16>* %A, <4 x i16>* %B) nounwind {
13 ;CHECK-LABEL: sqsub4h:
15 %tmp1 = load <4 x i16>, <4 x i16>* %A
16 %tmp2 = load <4 x i16>, <4 x i16>* %B
17 %tmp3 = call <4 x i16> @llvm.aarch64.neon.sqsub.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
21 define <2 x i32> @sqsub2s(<2 x i32>* %A, <2 x i32>* %B) nounwind {
22 ;CHECK-LABEL: sqsub2s:
24 %tmp1 = load <2 x i32>, <2 x i32>* %A
25 %tmp2 = load <2 x i32>, <2 x i32>* %B
26 %tmp3 = call <2 x i32> @llvm.aarch64.neon.sqsub.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
30 define <8 x i8> @uqsub8b(<8 x i8>* %A, <8 x i8>* %B) nounwind {
31 ;CHECK-LABEL: uqsub8b:
33 %tmp1 = load <8 x i8>, <8 x i8>* %A
34 %tmp2 = load <8 x i8>, <8 x i8>* %B
35 %tmp3 = call <8 x i8> @llvm.aarch64.neon.uqsub.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
39 define <4 x i16> @uqsub4h(<4 x i16>* %A, <4 x i16>* %B) nounwind {
40 ;CHECK-LABEL: uqsub4h:
42 %tmp1 = load <4 x i16>, <4 x i16>* %A
43 %tmp2 = load <4 x i16>, <4 x i16>* %B
44 %tmp3 = call <4 x i16> @llvm.aarch64.neon.uqsub.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
48 define <2 x i32> @uqsub2s(<2 x i32>* %A, <2 x i32>* %B) nounwind {
49 ;CHECK-LABEL: uqsub2s:
51 %tmp1 = load <2 x i32>, <2 x i32>* %A
52 %tmp2 = load <2 x i32>, <2 x i32>* %B
53 %tmp3 = call <2 x i32> @llvm.aarch64.neon.uqsub.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
57 define <16 x i8> @sqsub16b(<16 x i8>* %A, <16 x i8>* %B) nounwind {
58 ;CHECK-LABEL: sqsub16b:
60 %tmp1 = load <16 x i8>, <16 x i8>* %A
61 %tmp2 = load <16 x i8>, <16 x i8>* %B
62 %tmp3 = call <16 x i8> @llvm.aarch64.neon.sqsub.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
66 define <8 x i16> @sqsub8h(<8 x i16>* %A, <8 x i16>* %B) nounwind {
67 ;CHECK-LABEL: sqsub8h:
69 %tmp1 = load <8 x i16>, <8 x i16>* %A
70 %tmp2 = load <8 x i16>, <8 x i16>* %B
71 %tmp3 = call <8 x i16> @llvm.aarch64.neon.sqsub.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
75 define <4 x i32> @sqsub4s(<4 x i32>* %A, <4 x i32>* %B) nounwind {
76 ;CHECK-LABEL: sqsub4s:
78 %tmp1 = load <4 x i32>, <4 x i32>* %A
79 %tmp2 = load <4 x i32>, <4 x i32>* %B
80 %tmp3 = call <4 x i32> @llvm.aarch64.neon.sqsub.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
84 define <2 x i64> @sqsub2d(<2 x i64>* %A, <2 x i64>* %B) nounwind {
85 ;CHECK-LABEL: sqsub2d:
87 %tmp1 = load <2 x i64>, <2 x i64>* %A
88 %tmp2 = load <2 x i64>, <2 x i64>* %B
89 %tmp3 = call <2 x i64> @llvm.aarch64.neon.sqsub.v2i64(<2 x i64> %tmp1, <2 x i64> %tmp2)
93 define <16 x i8> @uqsub16b(<16 x i8>* %A, <16 x i8>* %B) nounwind {
94 ;CHECK-LABEL: uqsub16b:
96 %tmp1 = load <16 x i8>, <16 x i8>* %A
97 %tmp2 = load <16 x i8>, <16 x i8>* %B
98 %tmp3 = call <16 x i8> @llvm.aarch64.neon.uqsub.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
102 define <8 x i16> @uqsub8h(<8 x i16>* %A, <8 x i16>* %B) nounwind {
103 ;CHECK-LABEL: uqsub8h:
105 %tmp1 = load <8 x i16>, <8 x i16>* %A
106 %tmp2 = load <8 x i16>, <8 x i16>* %B
107 %tmp3 = call <8 x i16> @llvm.aarch64.neon.uqsub.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
111 define <4 x i32> @uqsub4s(<4 x i32>* %A, <4 x i32>* %B) nounwind {
112 ;CHECK-LABEL: uqsub4s:
114 %tmp1 = load <4 x i32>, <4 x i32>* %A
115 %tmp2 = load <4 x i32>, <4 x i32>* %B
116 %tmp3 = call <4 x i32> @llvm.aarch64.neon.uqsub.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
120 define <2 x i64> @uqsub2d(<2 x i64>* %A, <2 x i64>* %B) nounwind {
121 ;CHECK-LABEL: uqsub2d:
123 %tmp1 = load <2 x i64>, <2 x i64>* %A
124 %tmp2 = load <2 x i64>, <2 x i64>* %B
125 %tmp3 = call <2 x i64> @llvm.aarch64.neon.uqsub.v2i64(<2 x i64> %tmp1, <2 x i64> %tmp2)
129 declare <8 x i8> @llvm.aarch64.neon.sqsub.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
130 declare <4 x i16> @llvm.aarch64.neon.sqsub.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
131 declare <2 x i32> @llvm.aarch64.neon.sqsub.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
132 declare <1 x i64> @llvm.aarch64.neon.sqsub.v1i64(<1 x i64>, <1 x i64>) nounwind readnone
134 declare <8 x i8> @llvm.aarch64.neon.uqsub.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
135 declare <4 x i16> @llvm.aarch64.neon.uqsub.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
136 declare <2 x i32> @llvm.aarch64.neon.uqsub.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
137 declare <1 x i64> @llvm.aarch64.neon.uqsub.v1i64(<1 x i64>, <1 x i64>) nounwind readnone
139 declare <16 x i8> @llvm.aarch64.neon.sqsub.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
140 declare <8 x i16> @llvm.aarch64.neon.sqsub.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
141 declare <4 x i32> @llvm.aarch64.neon.sqsub.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
142 declare <2 x i64> @llvm.aarch64.neon.sqsub.v2i64(<2 x i64>, <2 x i64>) nounwind readnone
144 declare <16 x i8> @llvm.aarch64.neon.uqsub.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
145 declare <8 x i16> @llvm.aarch64.neon.uqsub.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
146 declare <4 x i32> @llvm.aarch64.neon.uqsub.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
147 declare <2 x i64> @llvm.aarch64.neon.uqsub.v2i64(<2 x i64>, <2 x i64>) nounwind readnone