1 ; RUN: llc -mtriple=arm-eabi -mattr=+neon %s -o - | FileCheck %s
3 define <8 x i8> @vqadds8(<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.arm.neon.vqadds.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
12 define <4 x i16> @vqadds16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
13 ;CHECK-LABEL: vqadds16:
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.arm.neon.vqadds.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
21 define <2 x i32> @vqadds32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
22 ;CHECK-LABEL: vqadds32:
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.arm.neon.vqadds.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
30 define <1 x i64> @vqadds64(<1 x i64>* %A, <1 x i64>* %B) nounwind {
31 ;CHECK-LABEL: vqadds64:
33 %tmp1 = load <1 x i64>, <1 x i64>* %A
34 %tmp2 = load <1 x i64>, <1 x i64>* %B
35 %tmp3 = call <1 x i64> @llvm.arm.neon.vqadds.v1i64(<1 x i64> %tmp1, <1 x i64> %tmp2)
39 define <8 x i8> @vqaddu8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
40 ;CHECK-LABEL: vqaddu8:
42 %tmp1 = load <8 x i8>, <8 x i8>* %A
43 %tmp2 = load <8 x i8>, <8 x i8>* %B
44 %tmp3 = call <8 x i8> @llvm.arm.neon.vqaddu.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
48 define <4 x i16> @vqaddu16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
49 ;CHECK-LABEL: vqaddu16:
51 %tmp1 = load <4 x i16>, <4 x i16>* %A
52 %tmp2 = load <4 x i16>, <4 x i16>* %B
53 %tmp3 = call <4 x i16> @llvm.arm.neon.vqaddu.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
57 define <2 x i32> @vqaddu32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
58 ;CHECK-LABEL: vqaddu32:
60 %tmp1 = load <2 x i32>, <2 x i32>* %A
61 %tmp2 = load <2 x i32>, <2 x i32>* %B
62 %tmp3 = call <2 x i32> @llvm.arm.neon.vqaddu.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
66 define <1 x i64> @vqaddu64(<1 x i64>* %A, <1 x i64>* %B) nounwind {
67 ;CHECK-LABEL: vqaddu64:
69 %tmp1 = load <1 x i64>, <1 x i64>* %A
70 %tmp2 = load <1 x i64>, <1 x i64>* %B
71 %tmp3 = call <1 x i64> @llvm.arm.neon.vqaddu.v1i64(<1 x i64> %tmp1, <1 x i64> %tmp2)
75 define <16 x i8> @vqaddQs8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
76 ;CHECK-LABEL: vqaddQs8:
78 %tmp1 = load <16 x i8>, <16 x i8>* %A
79 %tmp2 = load <16 x i8>, <16 x i8>* %B
80 %tmp3 = call <16 x i8> @llvm.arm.neon.vqadds.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
84 define <8 x i16> @vqaddQs16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
85 ;CHECK-LABEL: vqaddQs16:
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.arm.neon.vqadds.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
93 define <4 x i32> @vqaddQs32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
94 ;CHECK-LABEL: vqaddQs32:
96 %tmp1 = load <4 x i32>, <4 x i32>* %A
97 %tmp2 = load <4 x i32>, <4 x i32>* %B
98 %tmp3 = call <4 x i32> @llvm.arm.neon.vqadds.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
102 define <2 x i64> @vqaddQs64(<2 x i64>* %A, <2 x i64>* %B) nounwind {
103 ;CHECK-LABEL: vqaddQs64:
105 %tmp1 = load <2 x i64>, <2 x i64>* %A
106 %tmp2 = load <2 x i64>, <2 x i64>* %B
107 %tmp3 = call <2 x i64> @llvm.arm.neon.vqadds.v2i64(<2 x i64> %tmp1, <2 x i64> %tmp2)
111 define <16 x i8> @vqaddQu8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
112 ;CHECK-LABEL: vqaddQu8:
114 %tmp1 = load <16 x i8>, <16 x i8>* %A
115 %tmp2 = load <16 x i8>, <16 x i8>* %B
116 %tmp3 = call <16 x i8> @llvm.arm.neon.vqaddu.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
120 define <8 x i16> @vqaddQu16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
121 ;CHECK-LABEL: vqaddQu16:
123 %tmp1 = load <8 x i16>, <8 x i16>* %A
124 %tmp2 = load <8 x i16>, <8 x i16>* %B
125 %tmp3 = call <8 x i16> @llvm.arm.neon.vqaddu.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
129 define <4 x i32> @vqaddQu32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
130 ;CHECK-LABEL: vqaddQu32:
132 %tmp1 = load <4 x i32>, <4 x i32>* %A
133 %tmp2 = load <4 x i32>, <4 x i32>* %B
134 %tmp3 = call <4 x i32> @llvm.arm.neon.vqaddu.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
138 define <2 x i64> @vqaddQu64(<2 x i64>* %A, <2 x i64>* %B) nounwind {
139 ;CHECK-LABEL: vqaddQu64:
141 %tmp1 = load <2 x i64>, <2 x i64>* %A
142 %tmp2 = load <2 x i64>, <2 x i64>* %B
143 %tmp3 = call <2 x i64> @llvm.arm.neon.vqaddu.v2i64(<2 x i64> %tmp1, <2 x i64> %tmp2)
148 define <8 x i8> @vqsubs8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
149 ;CHECK-LABEL: vqsubs8:
151 %tmp1 = load <8 x i8>, <8 x i8>* %A
152 %tmp2 = load <8 x i8>, <8 x i8>* %B
153 %tmp3 = call <8 x i8> @llvm.arm.neon.vqsubs.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
157 define <4 x i16> @vqsubs16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
158 ;CHECK-LABEL: vqsubs16:
160 %tmp1 = load <4 x i16>, <4 x i16>* %A
161 %tmp2 = load <4 x i16>, <4 x i16>* %B
162 %tmp3 = call <4 x i16> @llvm.arm.neon.vqsubs.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
166 define <2 x i32> @vqsubs32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
167 ;CHECK-LABEL: vqsubs32:
169 %tmp1 = load <2 x i32>, <2 x i32>* %A
170 %tmp2 = load <2 x i32>, <2 x i32>* %B
171 %tmp3 = call <2 x i32> @llvm.arm.neon.vqsubs.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
175 define <1 x i64> @vqsubs64(<1 x i64>* %A, <1 x i64>* %B) nounwind {
176 ;CHECK-LABEL: vqsubs64:
178 %tmp1 = load <1 x i64>, <1 x i64>* %A
179 %tmp2 = load <1 x i64>, <1 x i64>* %B
180 %tmp3 = call <1 x i64> @llvm.arm.neon.vqsubs.v1i64(<1 x i64> %tmp1, <1 x i64> %tmp2)
184 define <8 x i8> @vqsubu8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
185 ;CHECK-LABEL: vqsubu8:
187 %tmp1 = load <8 x i8>, <8 x i8>* %A
188 %tmp2 = load <8 x i8>, <8 x i8>* %B
189 %tmp3 = call <8 x i8> @llvm.arm.neon.vqsubu.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
193 define <4 x i16> @vqsubu16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
194 ;CHECK-LABEL: vqsubu16:
196 %tmp1 = load <4 x i16>, <4 x i16>* %A
197 %tmp2 = load <4 x i16>, <4 x i16>* %B
198 %tmp3 = call <4 x i16> @llvm.arm.neon.vqsubu.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
202 define <2 x i32> @vqsubu32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
203 ;CHECK-LABEL: vqsubu32:
205 %tmp1 = load <2 x i32>, <2 x i32>* %A
206 %tmp2 = load <2 x i32>, <2 x i32>* %B
207 %tmp3 = call <2 x i32> @llvm.arm.neon.vqsubu.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
211 define <1 x i64> @vqsubu64(<1 x i64>* %A, <1 x i64>* %B) nounwind {
212 ;CHECK-LABEL: vqsubu64:
214 %tmp1 = load <1 x i64>, <1 x i64>* %A
215 %tmp2 = load <1 x i64>, <1 x i64>* %B
216 %tmp3 = call <1 x i64> @llvm.arm.neon.vqsubu.v1i64(<1 x i64> %tmp1, <1 x i64> %tmp2)
220 define <16 x i8> @vqsubQs8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
221 ;CHECK-LABEL: vqsubQs8:
223 %tmp1 = load <16 x i8>, <16 x i8>* %A
224 %tmp2 = load <16 x i8>, <16 x i8>* %B
225 %tmp3 = call <16 x i8> @llvm.arm.neon.vqsubs.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
229 define <8 x i16> @vqsubQs16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
230 ;CHECK-LABEL: vqsubQs16:
232 %tmp1 = load <8 x i16>, <8 x i16>* %A
233 %tmp2 = load <8 x i16>, <8 x i16>* %B
234 %tmp3 = call <8 x i16> @llvm.arm.neon.vqsubs.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
238 define <4 x i32> @vqsubQs32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
239 ;CHECK-LABEL: vqsubQs32:
241 %tmp1 = load <4 x i32>, <4 x i32>* %A
242 %tmp2 = load <4 x i32>, <4 x i32>* %B
243 %tmp3 = call <4 x i32> @llvm.arm.neon.vqsubs.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
247 define <2 x i64> @vqsubQs64(<2 x i64>* %A, <2 x i64>* %B) nounwind {
248 ;CHECK-LABEL: vqsubQs64:
250 %tmp1 = load <2 x i64>, <2 x i64>* %A
251 %tmp2 = load <2 x i64>, <2 x i64>* %B
252 %tmp3 = call <2 x i64> @llvm.arm.neon.vqsubs.v2i64(<2 x i64> %tmp1, <2 x i64> %tmp2)
256 define <16 x i8> @vqsubQu8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
257 ;CHECK-LABEL: vqsubQu8:
259 %tmp1 = load <16 x i8>, <16 x i8>* %A
260 %tmp2 = load <16 x i8>, <16 x i8>* %B
261 %tmp3 = call <16 x i8> @llvm.arm.neon.vqsubu.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
265 define <8 x i16> @vqsubQu16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
266 ;CHECK-LABEL: vqsubQu16:
268 %tmp1 = load <8 x i16>, <8 x i16>* %A
269 %tmp2 = load <8 x i16>, <8 x i16>* %B
270 %tmp3 = call <8 x i16> @llvm.arm.neon.vqsubu.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
274 define <4 x i32> @vqsubQu32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
275 ;CHECK-LABEL: vqsubQu32:
277 %tmp1 = load <4 x i32>, <4 x i32>* %A
278 %tmp2 = load <4 x i32>, <4 x i32>* %B
279 %tmp3 = call <4 x i32> @llvm.arm.neon.vqsubu.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
283 define <2 x i64> @vqsubQu64(<2 x i64>* %A, <2 x i64>* %B) nounwind {
284 ;CHECK-LABEL: vqsubQu64:
286 %tmp1 = load <2 x i64>, <2 x i64>* %A
287 %tmp2 = load <2 x i64>, <2 x i64>* %B
288 %tmp3 = call <2 x i64> @llvm.arm.neon.vqsubu.v2i64(<2 x i64> %tmp1, <2 x i64> %tmp2)
292 declare <8 x i8> @llvm.arm.neon.vqadds.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
293 declare <4 x i16> @llvm.arm.neon.vqadds.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
294 declare <2 x i32> @llvm.arm.neon.vqadds.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
295 declare <1 x i64> @llvm.arm.neon.vqadds.v1i64(<1 x i64>, <1 x i64>) nounwind readnone
297 declare <8 x i8> @llvm.arm.neon.vqaddu.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
298 declare <4 x i16> @llvm.arm.neon.vqaddu.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
299 declare <2 x i32> @llvm.arm.neon.vqaddu.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
300 declare <1 x i64> @llvm.arm.neon.vqaddu.v1i64(<1 x i64>, <1 x i64>) nounwind readnone
302 declare <16 x i8> @llvm.arm.neon.vqadds.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
303 declare <8 x i16> @llvm.arm.neon.vqadds.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
304 declare <4 x i32> @llvm.arm.neon.vqadds.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
305 declare <2 x i64> @llvm.arm.neon.vqadds.v2i64(<2 x i64>, <2 x i64>) nounwind readnone
307 declare <16 x i8> @llvm.arm.neon.vqaddu.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
308 declare <8 x i16> @llvm.arm.neon.vqaddu.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
309 declare <4 x i32> @llvm.arm.neon.vqaddu.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
310 declare <2 x i64> @llvm.arm.neon.vqaddu.v2i64(<2 x i64>, <2 x i64>) nounwind readnone
312 declare <8 x i8> @llvm.arm.neon.vqsubs.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
313 declare <4 x i16> @llvm.arm.neon.vqsubs.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
314 declare <2 x i32> @llvm.arm.neon.vqsubs.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
315 declare <1 x i64> @llvm.arm.neon.vqsubs.v1i64(<1 x i64>, <1 x i64>) nounwind readnone
317 declare <8 x i8> @llvm.arm.neon.vqsubu.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
318 declare <4 x i16> @llvm.arm.neon.vqsubu.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
319 declare <2 x i32> @llvm.arm.neon.vqsubu.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
320 declare <1 x i64> @llvm.arm.neon.vqsubu.v1i64(<1 x i64>, <1 x i64>) nounwind readnone
322 declare <16 x i8> @llvm.arm.neon.vqsubs.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
323 declare <8 x i16> @llvm.arm.neon.vqsubs.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
324 declare <4 x i32> @llvm.arm.neon.vqsubs.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
325 declare <2 x i64> @llvm.arm.neon.vqsubs.v2i64(<2 x i64>, <2 x i64>) nounwind readnone
327 declare <16 x i8> @llvm.arm.neon.vqsubu.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
328 declare <8 x i16> @llvm.arm.neon.vqsubu.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
329 declare <4 x i32> @llvm.arm.neon.vqsubu.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
330 declare <2 x i64> @llvm.arm.neon.vqsubu.v2i64(<2 x i64>, <2 x i64>) nounwind readnone