1 ; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s
3 define <2 x i32> @vrecpei32(<2 x i32>* %A) nounwind {
6 %tmp1 = load <2 x i32>* %A
7 %tmp2 = call <2 x i32> @llvm.arm.neon.vrecpe.v2i32(<2 x i32> %tmp1)
11 define <4 x i32> @vrecpeQi32(<4 x i32>* %A) nounwind {
14 %tmp1 = load <4 x i32>* %A
15 %tmp2 = call <4 x i32> @llvm.arm.neon.vrecpe.v4i32(<4 x i32> %tmp1)
19 define <2 x float> @vrecpef32(<2 x float>* %A) nounwind {
22 %tmp1 = load <2 x float>* %A
23 %tmp2 = call <2 x float> @llvm.arm.neon.vrecpe.v2f32(<2 x float> %tmp1)
27 define <4 x float> @vrecpeQf32(<4 x float>* %A) nounwind {
30 %tmp1 = load <4 x float>* %A
31 %tmp2 = call <4 x float> @llvm.arm.neon.vrecpe.v4f32(<4 x float> %tmp1)
35 declare <2 x i32> @llvm.arm.neon.vrecpe.v2i32(<2 x i32>) nounwind readnone
36 declare <4 x i32> @llvm.arm.neon.vrecpe.v4i32(<4 x i32>) nounwind readnone
38 declare <2 x float> @llvm.arm.neon.vrecpe.v2f32(<2 x float>) nounwind readnone
39 declare <4 x float> @llvm.arm.neon.vrecpe.v4f32(<4 x float>) nounwind readnone
41 define <2 x float> @vrecpsf32(<2 x float>* %A, <2 x float>* %B) nounwind {
44 %tmp1 = load <2 x float>* %A
45 %tmp2 = load <2 x float>* %B
46 %tmp3 = call <2 x float> @llvm.arm.neon.vrecps.v2f32(<2 x float> %tmp1, <2 x float> %tmp2)
50 define <4 x float> @vrecpsQf32(<4 x float>* %A, <4 x float>* %B) nounwind {
53 %tmp1 = load <4 x float>* %A
54 %tmp2 = load <4 x float>* %B
55 %tmp3 = call <4 x float> @llvm.arm.neon.vrecps.v4f32(<4 x float> %tmp1, <4 x float> %tmp2)
59 declare <2 x float> @llvm.arm.neon.vrecps.v2f32(<2 x float>, <2 x float>) nounwind readnone
60 declare <4 x float> @llvm.arm.neon.vrecps.v4f32(<4 x float>, <4 x float>) nounwind readnone
62 define <2 x i32> @vrsqrtei32(<2 x i32>* %A) nounwind {
65 %tmp1 = load <2 x i32>* %A
66 %tmp2 = call <2 x i32> @llvm.arm.neon.vrsqrte.v2i32(<2 x i32> %tmp1)
70 define <4 x i32> @vrsqrteQi32(<4 x i32>* %A) nounwind {
73 %tmp1 = load <4 x i32>* %A
74 %tmp2 = call <4 x i32> @llvm.arm.neon.vrsqrte.v4i32(<4 x i32> %tmp1)
78 define <2 x float> @vrsqrtef32(<2 x float>* %A) nounwind {
81 %tmp1 = load <2 x float>* %A
82 %tmp2 = call <2 x float> @llvm.arm.neon.vrsqrte.v2f32(<2 x float> %tmp1)
86 define <4 x float> @vrsqrteQf32(<4 x float>* %A) nounwind {
89 %tmp1 = load <4 x float>* %A
90 %tmp2 = call <4 x float> @llvm.arm.neon.vrsqrte.v4f32(<4 x float> %tmp1)
94 declare <2 x i32> @llvm.arm.neon.vrsqrte.v2i32(<2 x i32>) nounwind readnone
95 declare <4 x i32> @llvm.arm.neon.vrsqrte.v4i32(<4 x i32>) nounwind readnone
97 declare <2 x float> @llvm.arm.neon.vrsqrte.v2f32(<2 x float>) nounwind readnone
98 declare <4 x float> @llvm.arm.neon.vrsqrte.v4f32(<4 x float>) nounwind readnone
100 define <2 x float> @vrsqrtsf32(<2 x float>* %A, <2 x float>* %B) nounwind {
103 %tmp1 = load <2 x float>* %A
104 %tmp2 = load <2 x float>* %B
105 %tmp3 = call <2 x float> @llvm.arm.neon.vrsqrts.v2f32(<2 x float> %tmp1, <2 x float> %tmp2)
106 ret <2 x float> %tmp3
109 define <4 x float> @vrsqrtsQf32(<4 x float>* %A, <4 x float>* %B) nounwind {
112 %tmp1 = load <4 x float>* %A
113 %tmp2 = load <4 x float>* %B
114 %tmp3 = call <4 x float> @llvm.arm.neon.vrsqrts.v4f32(<4 x float> %tmp1, <4 x float> %tmp2)
115 ret <4 x float> %tmp3
118 declare <2 x float> @llvm.arm.neon.vrsqrts.v2f32(<2 x float>, <2 x float>) nounwind readnone
119 declare <4 x float> @llvm.arm.neon.vrsqrts.v4f32(<4 x float>, <4 x float>) nounwind readnone