1 ; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s
3 define <8 x i8> @vqsubs8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
6 %tmp1 = load <8 x i8>* %A
7 %tmp2 = load <8 x i8>* %B
8 %tmp3 = call <8 x i8> @llvm.arm.neon.vqsubs.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
12 define <4 x i16> @vqsubs16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
15 %tmp1 = load <4 x i16>* %A
16 %tmp2 = load <4 x i16>* %B
17 %tmp3 = call <4 x i16> @llvm.arm.neon.vqsubs.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
21 define <2 x i32> @vqsubs32(<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.vqsubs.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
30 define <1 x i64> @vqsubs64(<1 x i64>* %A, <1 x i64>* %B) nounwind {
33 %tmp1 = load <1 x i64>* %A
34 %tmp2 = load <1 x i64>* %B
35 %tmp3 = call <1 x i64> @llvm.arm.neon.vqsubs.v1i64(<1 x i64> %tmp1, <1 x i64> %tmp2)
39 define <8 x i8> @vqsubu8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
42 %tmp1 = load <8 x i8>* %A
43 %tmp2 = load <8 x i8>* %B
44 %tmp3 = call <8 x i8> @llvm.arm.neon.vqsubu.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
48 define <4 x i16> @vqsubu16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
51 %tmp1 = load <4 x i16>* %A
52 %tmp2 = load <4 x i16>* %B
53 %tmp3 = call <4 x i16> @llvm.arm.neon.vqsubu.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
57 define <2 x i32> @vqsubu32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
60 %tmp1 = load <2 x i32>* %A
61 %tmp2 = load <2 x i32>* %B
62 %tmp3 = call <2 x i32> @llvm.arm.neon.vqsubu.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
66 define <1 x i64> @vqsubu64(<1 x i64>* %A, <1 x i64>* %B) nounwind {
69 %tmp1 = load <1 x i64>* %A
70 %tmp2 = load <1 x i64>* %B
71 %tmp3 = call <1 x i64> @llvm.arm.neon.vqsubu.v1i64(<1 x i64> %tmp1, <1 x i64> %tmp2)
75 define <16 x i8> @vqsubQs8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
78 %tmp1 = load <16 x i8>* %A
79 %tmp2 = load <16 x i8>* %B
80 %tmp3 = call <16 x i8> @llvm.arm.neon.vqsubs.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
84 define <8 x i16> @vqsubQs16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
87 %tmp1 = load <8 x i16>* %A
88 %tmp2 = load <8 x i16>* %B
89 %tmp3 = call <8 x i16> @llvm.arm.neon.vqsubs.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
93 define <4 x i32> @vqsubQs32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
96 %tmp1 = load <4 x i32>* %A
97 %tmp2 = load <4 x i32>* %B
98 %tmp3 = call <4 x i32> @llvm.arm.neon.vqsubs.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
102 define <2 x i64> @vqsubQs64(<2 x i64>* %A, <2 x i64>* %B) nounwind {
105 %tmp1 = load <2 x i64>* %A
106 %tmp2 = load <2 x i64>* %B
107 %tmp3 = call <2 x i64> @llvm.arm.neon.vqsubs.v2i64(<2 x i64> %tmp1, <2 x i64> %tmp2)
111 define <16 x i8> @vqsubQu8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
114 %tmp1 = load <16 x i8>* %A
115 %tmp2 = load <16 x i8>* %B
116 %tmp3 = call <16 x i8> @llvm.arm.neon.vqsubu.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
120 define <8 x i16> @vqsubQu16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
123 %tmp1 = load <8 x i16>* %A
124 %tmp2 = load <8 x i16>* %B
125 %tmp3 = call <8 x i16> @llvm.arm.neon.vqsubu.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
129 define <4 x i32> @vqsubQu32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
132 %tmp1 = load <4 x i32>* %A
133 %tmp2 = load <4 x i32>* %B
134 %tmp3 = call <4 x i32> @llvm.arm.neon.vqsubu.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
138 define <2 x i64> @vqsubQu64(<2 x i64>* %A, <2 x i64>* %B) nounwind {
141 %tmp1 = load <2 x i64>* %A
142 %tmp2 = load <2 x i64>* %B
143 %tmp3 = call <2 x i64> @llvm.arm.neon.vqsubu.v2i64(<2 x i64> %tmp1, <2 x i64> %tmp2)
147 declare <8 x i8> @llvm.arm.neon.vqsubs.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
148 declare <4 x i16> @llvm.arm.neon.vqsubs.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
149 declare <2 x i32> @llvm.arm.neon.vqsubs.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
150 declare <1 x i64> @llvm.arm.neon.vqsubs.v1i64(<1 x i64>, <1 x i64>) nounwind readnone
152 declare <8 x i8> @llvm.arm.neon.vqsubu.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
153 declare <4 x i16> @llvm.arm.neon.vqsubu.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
154 declare <2 x i32> @llvm.arm.neon.vqsubu.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
155 declare <1 x i64> @llvm.arm.neon.vqsubu.v1i64(<1 x i64>, <1 x i64>) nounwind readnone
157 declare <16 x i8> @llvm.arm.neon.vqsubs.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
158 declare <8 x i16> @llvm.arm.neon.vqsubs.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
159 declare <4 x i32> @llvm.arm.neon.vqsubs.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
160 declare <2 x i64> @llvm.arm.neon.vqsubs.v2i64(<2 x i64>, <2 x i64>) nounwind readnone
162 declare <16 x i8> @llvm.arm.neon.vqsubu.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
163 declare <8 x i16> @llvm.arm.neon.vqsubu.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
164 declare <4 x i32> @llvm.arm.neon.vqsubu.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
165 declare <2 x i64> @llvm.arm.neon.vqsubu.v2i64(<2 x i64>, <2 x i64>) nounwind readnone