1 ; RUN: llc < %s -mtriple=arm64-eabi -aarch64-neon-syntax=apple | FileCheck %s
3 define <8 x i8> @shadd8b(<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.shadd.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
12 define <16 x i8> @shadd16b(<16 x i8>* %A, <16 x i8>* %B) nounwind {
13 ;CHECK-LABEL: shadd16b:
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.shadd.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
21 define <4 x i16> @shadd4h(<4 x i16>* %A, <4 x i16>* %B) nounwind {
22 ;CHECK-LABEL: shadd4h:
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.shadd.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
30 define <8 x i16> @shadd8h(<8 x i16>* %A, <8 x i16>* %B) nounwind {
31 ;CHECK-LABEL: shadd8h:
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.shadd.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
39 define <2 x i32> @shadd2s(<2 x i32>* %A, <2 x i32>* %B) nounwind {
40 ;CHECK-LABEL: shadd2s:
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.shadd.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
48 define <4 x i32> @shadd4s(<4 x i32>* %A, <4 x i32>* %B) nounwind {
49 ;CHECK-LABEL: shadd4s:
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.shadd.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
57 define <8 x i8> @uhadd8b(<8 x i8>* %A, <8 x i8>* %B) nounwind {
58 ;CHECK-LABEL: uhadd8b:
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.uhadd.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
66 define <16 x i8> @uhadd16b(<16 x i8>* %A, <16 x i8>* %B) nounwind {
67 ;CHECK-LABEL: uhadd16b:
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.uhadd.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
75 define <4 x i16> @uhadd4h(<4 x i16>* %A, <4 x i16>* %B) nounwind {
76 ;CHECK-LABEL: uhadd4h:
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.uhadd.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
84 define <8 x i16> @uhadd8h(<8 x i16>* %A, <8 x i16>* %B) nounwind {
85 ;CHECK-LABEL: uhadd8h:
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.uhadd.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
93 define <2 x i32> @uhadd2s(<2 x i32>* %A, <2 x i32>* %B) nounwind {
94 ;CHECK-LABEL: uhadd2s:
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.uhadd.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
102 define <4 x i32> @uhadd4s(<4 x i32>* %A, <4 x i32>* %B) nounwind {
103 ;CHECK-LABEL: uhadd4s:
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.uhadd.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
111 declare <8 x i8> @llvm.aarch64.neon.shadd.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
112 declare <4 x i16> @llvm.aarch64.neon.shadd.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
113 declare <2 x i32> @llvm.aarch64.neon.shadd.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
115 declare <8 x i8> @llvm.aarch64.neon.uhadd.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
116 declare <4 x i16> @llvm.aarch64.neon.uhadd.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
117 declare <2 x i32> @llvm.aarch64.neon.uhadd.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
119 declare <16 x i8> @llvm.aarch64.neon.shadd.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
120 declare <8 x i16> @llvm.aarch64.neon.shadd.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
121 declare <4 x i32> @llvm.aarch64.neon.shadd.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
123 declare <16 x i8> @llvm.aarch64.neon.uhadd.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
124 declare <8 x i16> @llvm.aarch64.neon.uhadd.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
125 declare <4 x i32> @llvm.aarch64.neon.uhadd.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
127 define <8 x i8> @srhadd8b(<8 x i8>* %A, <8 x i8>* %B) nounwind {
128 ;CHECK-LABEL: srhadd8b:
130 %tmp1 = load <8 x i8>, <8 x i8>* %A
131 %tmp2 = load <8 x i8>, <8 x i8>* %B
132 %tmp3 = call <8 x i8> @llvm.aarch64.neon.srhadd.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
136 define <16 x i8> @srhadd16b(<16 x i8>* %A, <16 x i8>* %B) nounwind {
137 ;CHECK-LABEL: srhadd16b:
139 %tmp1 = load <16 x i8>, <16 x i8>* %A
140 %tmp2 = load <16 x i8>, <16 x i8>* %B
141 %tmp3 = call <16 x i8> @llvm.aarch64.neon.srhadd.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
145 define <4 x i16> @srhadd4h(<4 x i16>* %A, <4 x i16>* %B) nounwind {
146 ;CHECK-LABEL: srhadd4h:
148 %tmp1 = load <4 x i16>, <4 x i16>* %A
149 %tmp2 = load <4 x i16>, <4 x i16>* %B
150 %tmp3 = call <4 x i16> @llvm.aarch64.neon.srhadd.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
154 define <8 x i16> @srhadd8h(<8 x i16>* %A, <8 x i16>* %B) nounwind {
155 ;CHECK-LABEL: srhadd8h:
157 %tmp1 = load <8 x i16>, <8 x i16>* %A
158 %tmp2 = load <8 x i16>, <8 x i16>* %B
159 %tmp3 = call <8 x i16> @llvm.aarch64.neon.srhadd.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
163 define <2 x i32> @srhadd2s(<2 x i32>* %A, <2 x i32>* %B) nounwind {
164 ;CHECK-LABEL: srhadd2s:
166 %tmp1 = load <2 x i32>, <2 x i32>* %A
167 %tmp2 = load <2 x i32>, <2 x i32>* %B
168 %tmp3 = call <2 x i32> @llvm.aarch64.neon.srhadd.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
172 define <4 x i32> @srhadd4s(<4 x i32>* %A, <4 x i32>* %B) nounwind {
173 ;CHECK-LABEL: srhadd4s:
175 %tmp1 = load <4 x i32>, <4 x i32>* %A
176 %tmp2 = load <4 x i32>, <4 x i32>* %B
177 %tmp3 = call <4 x i32> @llvm.aarch64.neon.srhadd.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
181 define <8 x i8> @urhadd8b(<8 x i8>* %A, <8 x i8>* %B) nounwind {
182 ;CHECK-LABEL: urhadd8b:
184 %tmp1 = load <8 x i8>, <8 x i8>* %A
185 %tmp2 = load <8 x i8>, <8 x i8>* %B
186 %tmp3 = call <8 x i8> @llvm.aarch64.neon.urhadd.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
190 define <16 x i8> @urhadd16b(<16 x i8>* %A, <16 x i8>* %B) nounwind {
191 ;CHECK-LABEL: urhadd16b:
193 %tmp1 = load <16 x i8>, <16 x i8>* %A
194 %tmp2 = load <16 x i8>, <16 x i8>* %B
195 %tmp3 = call <16 x i8> @llvm.aarch64.neon.urhadd.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
199 define <4 x i16> @urhadd4h(<4 x i16>* %A, <4 x i16>* %B) nounwind {
200 ;CHECK-LABEL: urhadd4h:
202 %tmp1 = load <4 x i16>, <4 x i16>* %A
203 %tmp2 = load <4 x i16>, <4 x i16>* %B
204 %tmp3 = call <4 x i16> @llvm.aarch64.neon.urhadd.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
208 define <8 x i16> @urhadd8h(<8 x i16>* %A, <8 x i16>* %B) nounwind {
209 ;CHECK-LABEL: urhadd8h:
211 %tmp1 = load <8 x i16>, <8 x i16>* %A
212 %tmp2 = load <8 x i16>, <8 x i16>* %B
213 %tmp3 = call <8 x i16> @llvm.aarch64.neon.urhadd.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
217 define <2 x i32> @urhadd2s(<2 x i32>* %A, <2 x i32>* %B) nounwind {
218 ;CHECK-LABEL: urhadd2s:
220 %tmp1 = load <2 x i32>, <2 x i32>* %A
221 %tmp2 = load <2 x i32>, <2 x i32>* %B
222 %tmp3 = call <2 x i32> @llvm.aarch64.neon.urhadd.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
226 define <4 x i32> @urhadd4s(<4 x i32>* %A, <4 x i32>* %B) nounwind {
227 ;CHECK-LABEL: urhadd4s:
229 %tmp1 = load <4 x i32>, <4 x i32>* %A
230 %tmp2 = load <4 x i32>, <4 x i32>* %B
231 %tmp3 = call <4 x i32> @llvm.aarch64.neon.urhadd.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
235 declare <8 x i8> @llvm.aarch64.neon.srhadd.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
236 declare <4 x i16> @llvm.aarch64.neon.srhadd.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
237 declare <2 x i32> @llvm.aarch64.neon.srhadd.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
239 declare <8 x i8> @llvm.aarch64.neon.urhadd.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
240 declare <4 x i16> @llvm.aarch64.neon.urhadd.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
241 declare <2 x i32> @llvm.aarch64.neon.urhadd.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
243 declare <16 x i8> @llvm.aarch64.neon.srhadd.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
244 declare <8 x i16> @llvm.aarch64.neon.srhadd.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
245 declare <4 x i32> @llvm.aarch64.neon.srhadd.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
247 declare <16 x i8> @llvm.aarch64.neon.urhadd.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
248 declare <8 x i16> @llvm.aarch64.neon.urhadd.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
249 declare <4 x i32> @llvm.aarch64.neon.urhadd.v4i32(<4 x i32>, <4 x i32>) nounwind readnone