1 ; RUN: llc < %s -march=arm -mattr=+neon > %t
2 ; RUN: grep {vmax\\.s8} %t | count 2
3 ; RUN: grep {vmax\\.s16} %t | count 2
4 ; RUN: grep {vmax\\.s32} %t | count 2
5 ; RUN: grep {vmax\\.u8} %t | count 2
6 ; RUN: grep {vmax\\.u16} %t | count 2
7 ; RUN: grep {vmax\\.u32} %t | count 2
8 ; RUN: grep {vmax\\.f32} %t | count 2
10 define <8 x i8> @vmaxs8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
11 %tmp1 = load <8 x i8>* %A
12 %tmp2 = load <8 x i8>* %B
13 %tmp3 = call <8 x i8> @llvm.arm.neon.vmaxs.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
17 define <4 x i16> @vmaxs16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
18 %tmp1 = load <4 x i16>* %A
19 %tmp2 = load <4 x i16>* %B
20 %tmp3 = call <4 x i16> @llvm.arm.neon.vmaxs.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
24 define <2 x i32> @vmaxs32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
25 %tmp1 = load <2 x i32>* %A
26 %tmp2 = load <2 x i32>* %B
27 %tmp3 = call <2 x i32> @llvm.arm.neon.vmaxs.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
31 define <8 x i8> @vmaxu8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
32 %tmp1 = load <8 x i8>* %A
33 %tmp2 = load <8 x i8>* %B
34 %tmp3 = call <8 x i8> @llvm.arm.neon.vmaxu.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
38 define <4 x i16> @vmaxu16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
39 %tmp1 = load <4 x i16>* %A
40 %tmp2 = load <4 x i16>* %B
41 %tmp3 = call <4 x i16> @llvm.arm.neon.vmaxu.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
45 define <2 x i32> @vmaxu32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
46 %tmp1 = load <2 x i32>* %A
47 %tmp2 = load <2 x i32>* %B
48 %tmp3 = call <2 x i32> @llvm.arm.neon.vmaxu.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
52 define <2 x float> @vmaxf32(<2 x float>* %A, <2 x float>* %B) nounwind {
53 %tmp1 = load <2 x float>* %A
54 %tmp2 = load <2 x float>* %B
55 %tmp3 = call <2 x float> @llvm.arm.neon.vmaxs.v2f32(<2 x float> %tmp1, <2 x float> %tmp2)
59 define <16 x i8> @vmaxQs8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
60 %tmp1 = load <16 x i8>* %A
61 %tmp2 = load <16 x i8>* %B
62 %tmp3 = call <16 x i8> @llvm.arm.neon.vmaxs.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
66 define <8 x i16> @vmaxQs16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
67 %tmp1 = load <8 x i16>* %A
68 %tmp2 = load <8 x i16>* %B
69 %tmp3 = call <8 x i16> @llvm.arm.neon.vmaxs.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
73 define <4 x i32> @vmaxQs32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
74 %tmp1 = load <4 x i32>* %A
75 %tmp2 = load <4 x i32>* %B
76 %tmp3 = call <4 x i32> @llvm.arm.neon.vmaxs.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
80 define <16 x i8> @vmaxQu8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
81 %tmp1 = load <16 x i8>* %A
82 %tmp2 = load <16 x i8>* %B
83 %tmp3 = call <16 x i8> @llvm.arm.neon.vmaxu.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
87 define <8 x i16> @vmaxQu16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
88 %tmp1 = load <8 x i16>* %A
89 %tmp2 = load <8 x i16>* %B
90 %tmp3 = call <8 x i16> @llvm.arm.neon.vmaxu.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
94 define <4 x i32> @vmaxQu32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
95 %tmp1 = load <4 x i32>* %A
96 %tmp2 = load <4 x i32>* %B
97 %tmp3 = call <4 x i32> @llvm.arm.neon.vmaxu.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
101 define <4 x float> @vmaxQf32(<4 x float>* %A, <4 x float>* %B) nounwind {
102 %tmp1 = load <4 x float>* %A
103 %tmp2 = load <4 x float>* %B
104 %tmp3 = call <4 x float> @llvm.arm.neon.vmaxs.v4f32(<4 x float> %tmp1, <4 x float> %tmp2)
105 ret <4 x float> %tmp3
108 declare <8 x i8> @llvm.arm.neon.vmaxs.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
109 declare <4 x i16> @llvm.arm.neon.vmaxs.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
110 declare <2 x i32> @llvm.arm.neon.vmaxs.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
112 declare <8 x i8> @llvm.arm.neon.vmaxu.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
113 declare <4 x i16> @llvm.arm.neon.vmaxu.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
114 declare <2 x i32> @llvm.arm.neon.vmaxu.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
116 declare <2 x float> @llvm.arm.neon.vmaxs.v2f32(<2 x float>, <2 x float>) nounwind readnone
118 declare <16 x i8> @llvm.arm.neon.vmaxs.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
119 declare <8 x i16> @llvm.arm.neon.vmaxs.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
120 declare <4 x i32> @llvm.arm.neon.vmaxs.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
122 declare <16 x i8> @llvm.arm.neon.vmaxu.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
123 declare <8 x i16> @llvm.arm.neon.vmaxu.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
124 declare <4 x i32> @llvm.arm.neon.vmaxu.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
126 declare <4 x float> @llvm.arm.neon.vmaxs.v4f32(<4 x float>, <4 x float>) nounwind readnone