1 ; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s
3 define <4 x i16> @vpadals8(<4 x i16>* %A, <8 x i8>* %B) nounwind {
6 %tmp1 = load <4 x i16>* %A
7 %tmp2 = load <8 x i8>* %B
8 %tmp3 = call <4 x i16> @llvm.arm.neon.vpadals.v4i16.v8i8(<4 x i16> %tmp1, <8 x i8> %tmp2)
12 define <2 x i32> @vpadals16(<2 x i32>* %A, <4 x i16>* %B) nounwind {
15 %tmp1 = load <2 x i32>* %A
16 %tmp2 = load <4 x i16>* %B
17 %tmp3 = call <2 x i32> @llvm.arm.neon.vpadals.v2i32.v4i16(<2 x i32> %tmp1, <4 x i16> %tmp2)
21 define <1 x i64> @vpadals32(<1 x i64>* %A, <2 x i32>* %B) nounwind {
24 %tmp1 = load <1 x i64>* %A
25 %tmp2 = load <2 x i32>* %B
26 %tmp3 = call <1 x i64> @llvm.arm.neon.vpadals.v1i64.v2i32(<1 x i64> %tmp1, <2 x i32> %tmp2)
30 define <4 x i16> @vpadalu8(<4 x i16>* %A, <8 x i8>* %B) nounwind {
33 %tmp1 = load <4 x i16>* %A
34 %tmp2 = load <8 x i8>* %B
35 %tmp3 = call <4 x i16> @llvm.arm.neon.vpadalu.v4i16.v8i8(<4 x i16> %tmp1, <8 x i8> %tmp2)
39 define <2 x i32> @vpadalu16(<2 x i32>* %A, <4 x i16>* %B) nounwind {
42 %tmp1 = load <2 x i32>* %A
43 %tmp2 = load <4 x i16>* %B
44 %tmp3 = call <2 x i32> @llvm.arm.neon.vpadalu.v2i32.v4i16(<2 x i32> %tmp1, <4 x i16> %tmp2)
48 define <1 x i64> @vpadalu32(<1 x i64>* %A, <2 x i32>* %B) nounwind {
51 %tmp1 = load <1 x i64>* %A
52 %tmp2 = load <2 x i32>* %B
53 %tmp3 = call <1 x i64> @llvm.arm.neon.vpadalu.v1i64.v2i32(<1 x i64> %tmp1, <2 x i32> %tmp2)
57 define <8 x i16> @vpadalQs8(<8 x i16>* %A, <16 x i8>* %B) nounwind {
60 %tmp1 = load <8 x i16>* %A
61 %tmp2 = load <16 x i8>* %B
62 %tmp3 = call <8 x i16> @llvm.arm.neon.vpadals.v8i16.v16i8(<8 x i16> %tmp1, <16 x i8> %tmp2)
66 define <4 x i32> @vpadalQs16(<4 x i32>* %A, <8 x i16>* %B) nounwind {
69 %tmp1 = load <4 x i32>* %A
70 %tmp2 = load <8 x i16>* %B
71 %tmp3 = call <4 x i32> @llvm.arm.neon.vpadals.v4i32.v8i16(<4 x i32> %tmp1, <8 x i16> %tmp2)
75 define <2 x i64> @vpadalQs32(<2 x i64>* %A, <4 x i32>* %B) nounwind {
78 %tmp1 = load <2 x i64>* %A
79 %tmp2 = load <4 x i32>* %B
80 %tmp3 = call <2 x i64> @llvm.arm.neon.vpadals.v2i64.v4i32(<2 x i64> %tmp1, <4 x i32> %tmp2)
84 define <8 x i16> @vpadalQu8(<8 x i16>* %A, <16 x i8>* %B) nounwind {
87 %tmp1 = load <8 x i16>* %A
88 %tmp2 = load <16 x i8>* %B
89 %tmp3 = call <8 x i16> @llvm.arm.neon.vpadalu.v8i16.v16i8(<8 x i16> %tmp1, <16 x i8> %tmp2)
93 define <4 x i32> @vpadalQu16(<4 x i32>* %A, <8 x i16>* %B) nounwind {
96 %tmp1 = load <4 x i32>* %A
97 %tmp2 = load <8 x i16>* %B
98 %tmp3 = call <4 x i32> @llvm.arm.neon.vpadalu.v4i32.v8i16(<4 x i32> %tmp1, <8 x i16> %tmp2)
102 define <2 x i64> @vpadalQu32(<2 x i64>* %A, <4 x i32>* %B) nounwind {
105 %tmp1 = load <2 x i64>* %A
106 %tmp2 = load <4 x i32>* %B
107 %tmp3 = call <2 x i64> @llvm.arm.neon.vpadalu.v2i64.v4i32(<2 x i64> %tmp1, <4 x i32> %tmp2)
111 declare <4 x i16> @llvm.arm.neon.vpadals.v4i16.v8i8(<4 x i16>, <8 x i8>) nounwind readnone
112 declare <2 x i32> @llvm.arm.neon.vpadals.v2i32.v4i16(<2 x i32>, <4 x i16>) nounwind readnone
113 declare <1 x i64> @llvm.arm.neon.vpadals.v1i64.v2i32(<1 x i64>, <2 x i32>) nounwind readnone
115 declare <4 x i16> @llvm.arm.neon.vpadalu.v4i16.v8i8(<4 x i16>, <8 x i8>) nounwind readnone
116 declare <2 x i32> @llvm.arm.neon.vpadalu.v2i32.v4i16(<2 x i32>, <4 x i16>) nounwind readnone
117 declare <1 x i64> @llvm.arm.neon.vpadalu.v1i64.v2i32(<1 x i64>, <2 x i32>) nounwind readnone
119 declare <8 x i16> @llvm.arm.neon.vpadals.v8i16.v16i8(<8 x i16>, <16 x i8>) nounwind readnone
120 declare <4 x i32> @llvm.arm.neon.vpadals.v4i32.v8i16(<4 x i32>, <8 x i16>) nounwind readnone
121 declare <2 x i64> @llvm.arm.neon.vpadals.v2i64.v4i32(<2 x i64>, <4 x i32>) nounwind readnone
123 declare <8 x i16> @llvm.arm.neon.vpadalu.v8i16.v16i8(<8 x i16>, <16 x i8>) nounwind readnone
124 declare <4 x i32> @llvm.arm.neon.vpadalu.v4i32.v8i16(<4 x i32>, <8 x i16>) nounwind readnone
125 declare <2 x i64> @llvm.arm.neon.vpadalu.v2i64.v4i32(<2 x i64>, <4 x i32>) nounwind readnone