arm64-vcvt_f32_su32.ll
2.22 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
; RUN: llc < %s -mtriple=arm64-eabi -aarch64-neon-syntax=apple | FileCheck %s
define <2 x float> @ucvt(<2 x i32> %a) nounwind readnone ssp {
; CHECK-LABEL: ucvt:
; CHECK: ucvtf.2s v0, v0
; CHECK: ret
%vcvt.i = uitofp <2 x i32> %a to <2 x float>
ret <2 x float> %vcvt.i
}
define <2 x float> @scvt(<2 x i32> %a) nounwind readnone ssp {
; CHECK-LABEL: scvt:
; CHECK: scvtf.2s v0, v0
; CHECK: ret
%vcvt.i = sitofp <2 x i32> %a to <2 x float>
ret <2 x float> %vcvt.i
}
define <4 x float> @ucvtq(<4 x i32> %a) nounwind readnone ssp {
; CHECK-LABEL: ucvtq:
; CHECK: ucvtf.4s v0, v0
; CHECK: ret
%vcvt.i = uitofp <4 x i32> %a to <4 x float>
ret <4 x float> %vcvt.i
}
define <4 x float> @scvtq(<4 x i32> %a) nounwind readnone ssp {
; CHECK-LABEL: scvtq:
; CHECK: scvtf.4s v0, v0
; CHECK: ret
%vcvt.i = sitofp <4 x i32> %a to <4 x float>
ret <4 x float> %vcvt.i
}
define <4 x float> @cvtf16(<4 x i16> %a) nounwind readnone ssp {
; CHECK-LABEL: cvtf16:
; CHECK: fcvtl v0.4s, v0.4h
; CHECK-NEXT: ret
%vcvt1.i = tail call <4 x float> @llvm.aarch64.neon.vcvthf2fp(<4 x i16> %a) nounwind
ret <4 x float> %vcvt1.i
}
define <4 x float> @cvtf16_high(<8 x i16> %a) nounwind readnone ssp {
; CHECK-LABEL: cvtf16_high:
; CHECK: fcvtl2 v0.4s, v0.8h
; CHECK-NEXT: ret
%in = shufflevector <8 x i16> %a, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
%vcvt1.i = tail call <4 x float> @llvm.aarch64.neon.vcvthf2fp(<4 x i16> %in) nounwind
ret <4 x float> %vcvt1.i
}
define <4 x i16> @cvtf16f32(<4 x float> %a) nounwind readnone ssp {
; CHECK-LABEL: cvtf16f32:
; CHECK: fcvtn v0.4h, v0.4s
; CHECK-NEXT: ret
%vcvt1.i = tail call <4 x i16> @llvm.aarch64.neon.vcvtfp2hf(<4 x float> %a) nounwind
ret <4 x i16> %vcvt1.i
}
define <8 x i16> @cvtf16f32_high(<4 x i16> %low, <4 x float> %high_big) {
; CHECK-LABEL: cvtf16f32_high:
; CHECK: fcvtn2 v0.8h, v1.4s
; CHECK-NEXT: ret
%high = call <4 x i16> @llvm.aarch64.neon.vcvtfp2hf(<4 x float> %high_big)
%res = shufflevector <4 x i16> %low, <4 x i16> %high, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
ret <8 x i16> %res
}
declare <4 x float> @llvm.aarch64.neon.vcvthf2fp(<4 x i16>) nounwind readnone
declare <4 x i16> @llvm.aarch64.neon.vcvtfp2hf(<4 x float>) nounwind readnone