1 ; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s
3 define <8 x i8> @v_movi8() nounwind {
5 ;CHECK: vmov.i8 d0, #0x8
6 ret <8 x i8> < i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8 >
9 define <4 x i16> @v_movi16a() nounwind {
11 ;CHECK: vmov.i16 d0, #0x10
12 ret <4 x i16> < i16 16, i16 16, i16 16, i16 16 >
15 define <4 x i16> @v_movi16b() nounwind {
17 ;CHECK: vmov.i16 d0, #0x1000
18 ret <4 x i16> < i16 4096, i16 4096, i16 4096, i16 4096 >
21 define <2 x i32> @v_movi32a() nounwind {
23 ;CHECK: vmov.i32 d0, #0x20
24 ret <2 x i32> < i32 32, i32 32 >
27 define <2 x i32> @v_movi32b() nounwind {
29 ;CHECK: vmov.i32 d0, #0x2000
30 ret <2 x i32> < i32 8192, i32 8192 >
33 define <2 x i32> @v_movi32c() nounwind {
35 ;CHECK: vmov.i32 d0, #0x200000
36 ret <2 x i32> < i32 2097152, i32 2097152 >
39 define <2 x i32> @v_movi32d() nounwind {
41 ;CHECK: vmov.i32 d0, #0x20000000
42 ret <2 x i32> < i32 536870912, i32 536870912 >
45 define <2 x i32> @v_movi32e() nounwind {
47 ;CHECK: vmov.i32 d0, #0x20FF
48 ret <2 x i32> < i32 8447, i32 8447 >
51 define <2 x i32> @v_movi32f() nounwind {
53 ;CHECK: vmov.i32 d0, #0x20FFFF
54 ret <2 x i32> < i32 2162687, i32 2162687 >
57 define <1 x i64> @v_movi64() nounwind {
59 ;CHECK: vmov.i64 d0, #0xFF0000FF0000FFFF
60 ret <1 x i64> < i64 18374687574888349695 >
63 define <16 x i8> @v_movQi8() nounwind {
65 ;CHECK: vmov.i8 q0, #0x8
66 ret <16 x i8> < i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8 >
69 define <8 x i16> @v_movQi16a() nounwind {
71 ;CHECK: vmov.i16 q0, #0x10
72 ret <8 x i16> < i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16 >
75 define <8 x i16> @v_movQi16b() nounwind {
77 ;CHECK: vmov.i16 q0, #0x1000
78 ret <8 x i16> < i16 4096, i16 4096, i16 4096, i16 4096, i16 4096, i16 4096, i16 4096, i16 4096 >
81 define <4 x i32> @v_movQi32a() nounwind {
83 ;CHECK: vmov.i32 q0, #0x20
84 ret <4 x i32> < i32 32, i32 32, i32 32, i32 32 >
87 define <4 x i32> @v_movQi32b() nounwind {
89 ;CHECK: vmov.i32 q0, #0x2000
90 ret <4 x i32> < i32 8192, i32 8192, i32 8192, i32 8192 >
93 define <4 x i32> @v_movQi32c() nounwind {
95 ;CHECK: vmov.i32 q0, #0x200000
96 ret <4 x i32> < i32 2097152, i32 2097152, i32 2097152, i32 2097152 >
99 define <4 x i32> @v_movQi32d() nounwind {
101 ;CHECK: vmov.i32 q0, #0x20000000
102 ret <4 x i32> < i32 536870912, i32 536870912, i32 536870912, i32 536870912 >
105 define <4 x i32> @v_movQi32e() nounwind {
107 ;CHECK: vmov.i32 q0, #0x20FF
108 ret <4 x i32> < i32 8447, i32 8447, i32 8447, i32 8447 >
111 define <4 x i32> @v_movQi32f() nounwind {
113 ;CHECK: vmov.i32 q0, #0x20FFFF
114 ret <4 x i32> < i32 2162687, i32 2162687, i32 2162687, i32 2162687 >
117 define <2 x i64> @v_movQi64() nounwind {
119 ;CHECK: vmov.i64 q0, #0xFF0000FF0000FFFF
120 ret <2 x i64> < i64 18374687574888349695, i64 18374687574888349695 >
123 ; Check for correct assembler printing for immediate values.
124 %struct.int8x8_t = type { <8 x i8> }
125 define void @vdupn128(%struct.int8x8_t* noalias nocapture sret %agg.result) nounwind {
128 ;CHECK: vmov.i8 d0, #0x80
129 %0 = getelementptr inbounds %struct.int8x8_t* %agg.result, i32 0, i32 0 ; <<8 x i8>*> [#uses=1]
130 store <8 x i8> <i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128>, <8 x i8>* %0, align 8
134 define void @vdupnneg75(%struct.int8x8_t* noalias nocapture sret %agg.result) nounwind {
137 ;CHECK: vmov.i8 d0, #0xB5
138 %0 = getelementptr inbounds %struct.int8x8_t* %agg.result, i32 0, i32 0 ; <<8 x i8>*> [#uses=1]
139 store <8 x i8> <i8 -75, i8 -75, i8 -75, i8 -75, i8 -75, i8 -75, i8 -75, i8 -75>, <8 x i8>* %0, align 8
143 define <8 x i16> @vmovls8(<8 x i8>* %A) nounwind {
146 %tmp1 = load <8 x i8>* %A
147 %tmp2 = call <8 x i16> @llvm.arm.neon.vmovls.v8i16(<8 x i8> %tmp1)
151 define <4 x i32> @vmovls16(<4 x i16>* %A) nounwind {
154 %tmp1 = load <4 x i16>* %A
155 %tmp2 = call <4 x i32> @llvm.arm.neon.vmovls.v4i32(<4 x i16> %tmp1)
159 define <2 x i64> @vmovls32(<2 x i32>* %A) nounwind {
162 %tmp1 = load <2 x i32>* %A
163 %tmp2 = call <2 x i64> @llvm.arm.neon.vmovls.v2i64(<2 x i32> %tmp1)
167 define <8 x i16> @vmovlu8(<8 x i8>* %A) nounwind {
170 %tmp1 = load <8 x i8>* %A
171 %tmp2 = call <8 x i16> @llvm.arm.neon.vmovlu.v8i16(<8 x i8> %tmp1)
175 define <4 x i32> @vmovlu16(<4 x i16>* %A) nounwind {
178 %tmp1 = load <4 x i16>* %A
179 %tmp2 = call <4 x i32> @llvm.arm.neon.vmovlu.v4i32(<4 x i16> %tmp1)
183 define <2 x i64> @vmovlu32(<2 x i32>* %A) nounwind {
186 %tmp1 = load <2 x i32>* %A
187 %tmp2 = call <2 x i64> @llvm.arm.neon.vmovlu.v2i64(<2 x i32> %tmp1)
191 declare <8 x i16> @llvm.arm.neon.vmovls.v8i16(<8 x i8>) nounwind readnone
192 declare <4 x i32> @llvm.arm.neon.vmovls.v4i32(<4 x i16>) nounwind readnone
193 declare <2 x i64> @llvm.arm.neon.vmovls.v2i64(<2 x i32>) nounwind readnone
195 declare <8 x i16> @llvm.arm.neon.vmovlu.v8i16(<8 x i8>) nounwind readnone
196 declare <4 x i32> @llvm.arm.neon.vmovlu.v4i32(<4 x i16>) nounwind readnone
197 declare <2 x i64> @llvm.arm.neon.vmovlu.v2i64(<2 x i32>) nounwind readnone
199 define <8 x i8> @vmovni16(<8 x i16>* %A) nounwind {
202 %tmp1 = load <8 x i16>* %A
203 %tmp2 = call <8 x i8> @llvm.arm.neon.vmovn.v8i8(<8 x i16> %tmp1)
207 define <4 x i16> @vmovni32(<4 x i32>* %A) nounwind {
210 %tmp1 = load <4 x i32>* %A
211 %tmp2 = call <4 x i16> @llvm.arm.neon.vmovn.v4i16(<4 x i32> %tmp1)
215 define <2 x i32> @vmovni64(<2 x i64>* %A) nounwind {
218 %tmp1 = load <2 x i64>* %A
219 %tmp2 = call <2 x i32> @llvm.arm.neon.vmovn.v2i32(<2 x i64> %tmp1)
223 declare <8 x i8> @llvm.arm.neon.vmovn.v8i8(<8 x i16>) nounwind readnone
224 declare <4 x i16> @llvm.arm.neon.vmovn.v4i16(<4 x i32>) nounwind readnone
225 declare <2 x i32> @llvm.arm.neon.vmovn.v2i32(<2 x i64>) nounwind readnone
227 define <8 x i8> @vqmovns16(<8 x i16>* %A) nounwind {
230 %tmp1 = load <8 x i16>* %A
231 %tmp2 = call <8 x i8> @llvm.arm.neon.vqmovns.v8i8(<8 x i16> %tmp1)
235 define <4 x i16> @vqmovns32(<4 x i32>* %A) nounwind {
238 %tmp1 = load <4 x i32>* %A
239 %tmp2 = call <4 x i16> @llvm.arm.neon.vqmovns.v4i16(<4 x i32> %tmp1)
243 define <2 x i32> @vqmovns64(<2 x i64>* %A) nounwind {
246 %tmp1 = load <2 x i64>* %A
247 %tmp2 = call <2 x i32> @llvm.arm.neon.vqmovns.v2i32(<2 x i64> %tmp1)
251 define <8 x i8> @vqmovnu16(<8 x i16>* %A) nounwind {
254 %tmp1 = load <8 x i16>* %A
255 %tmp2 = call <8 x i8> @llvm.arm.neon.vqmovnu.v8i8(<8 x i16> %tmp1)
259 define <4 x i16> @vqmovnu32(<4 x i32>* %A) nounwind {
262 %tmp1 = load <4 x i32>* %A
263 %tmp2 = call <4 x i16> @llvm.arm.neon.vqmovnu.v4i16(<4 x i32> %tmp1)
267 define <2 x i32> @vqmovnu64(<2 x i64>* %A) nounwind {
270 %tmp1 = load <2 x i64>* %A
271 %tmp2 = call <2 x i32> @llvm.arm.neon.vqmovnu.v2i32(<2 x i64> %tmp1)
275 define <8 x i8> @vqmovuns16(<8 x i16>* %A) nounwind {
278 %tmp1 = load <8 x i16>* %A
279 %tmp2 = call <8 x i8> @llvm.arm.neon.vqmovnsu.v8i8(<8 x i16> %tmp1)
283 define <4 x i16> @vqmovuns32(<4 x i32>* %A) nounwind {
286 %tmp1 = load <4 x i32>* %A
287 %tmp2 = call <4 x i16> @llvm.arm.neon.vqmovnsu.v4i16(<4 x i32> %tmp1)
291 define <2 x i32> @vqmovuns64(<2 x i64>* %A) nounwind {
294 %tmp1 = load <2 x i64>* %A
295 %tmp2 = call <2 x i32> @llvm.arm.neon.vqmovnsu.v2i32(<2 x i64> %tmp1)
299 declare <8 x i8> @llvm.arm.neon.vqmovns.v8i8(<8 x i16>) nounwind readnone
300 declare <4 x i16> @llvm.arm.neon.vqmovns.v4i16(<4 x i32>) nounwind readnone
301 declare <2 x i32> @llvm.arm.neon.vqmovns.v2i32(<2 x i64>) nounwind readnone
303 declare <8 x i8> @llvm.arm.neon.vqmovnu.v8i8(<8 x i16>) nounwind readnone
304 declare <4 x i16> @llvm.arm.neon.vqmovnu.v4i16(<4 x i32>) nounwind readnone
305 declare <2 x i32> @llvm.arm.neon.vqmovnu.v2i32(<2 x i64>) nounwind readnone
307 declare <8 x i8> @llvm.arm.neon.vqmovnsu.v8i8(<8 x i16>) nounwind readnone
308 declare <4 x i16> @llvm.arm.neon.vqmovnsu.v4i16(<4 x i32>) nounwind readnone
309 declare <2 x i32> @llvm.arm.neon.vqmovnsu.v2i32(<2 x i64>) nounwind readnone