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
74
|
; RUN: llc -march=arm64 -aarch64-neon-syntax=apple < %s | FileCheck %s
define signext i8 @test_vminv_s8(<8 x i8> %a1) {
; CHECK: test_vminv_s8
; CHECK: sminv.8b b[[REGNUM:[0-9]+]], v0
; CHECK-NEXT: smov.b w0, v[[REGNUM]][0]
; CHECK-NEXT: ret
entry:
%vminv.i = tail call i32 @llvm.aarch64.neon.sminv.i32.v8i8(<8 x i8> %a1)
%0 = trunc i32 %vminv.i to i8
ret i8 %0
}
define signext i16 @test_vminv_s16(<4 x i16> %a1) {
; CHECK: test_vminv_s16
; CHECK: sminv.4h h[[REGNUM:[0-9]+]], v0
; CHECK-NEXT: smov.h w0, v[[REGNUM]][0]
; CHECK-NEXT: ret
entry:
%vminv.i = tail call i32 @llvm.aarch64.neon.sminv.i32.v4i16(<4 x i16> %a1)
%0 = trunc i32 %vminv.i to i16
ret i16 %0
}
define i32 @test_vminv_s32(<2 x i32> %a1) {
; CHECK: test_vminv_s32
; 2 x i32 is not supported by the ISA, thus, this is a special case
; CHECK: sminp.2s v[[REGNUM:[0-9]+]], v0, v0
; CHECK-NEXT: fmov w0, s[[REGNUM]]
; CHECK-NEXT: ret
entry:
%vminv.i = tail call i32 @llvm.aarch64.neon.sminv.i32.v2i32(<2 x i32> %a1)
ret i32 %vminv.i
}
define signext i8 @test_vminvq_s8(<16 x i8> %a1) {
; CHECK: test_vminvq_s8
; CHECK: sminv.16b b[[REGNUM:[0-9]+]], v0
; CHECK-NEXT: smov.b w0, v[[REGNUM]][0]
; CHECK-NEXT: ret
entry:
%vminv.i = tail call i32 @llvm.aarch64.neon.sminv.i32.v16i8(<16 x i8> %a1)
%0 = trunc i32 %vminv.i to i8
ret i8 %0
}
define signext i16 @test_vminvq_s16(<8 x i16> %a1) {
; CHECK: test_vminvq_s16
; CHECK: sminv.8h h[[REGNUM:[0-9]+]], v0
; CHECK-NEXT: smov.h w0, v[[REGNUM]][0]
; CHECK-NEXT: ret
entry:
%vminv.i = tail call i32 @llvm.aarch64.neon.sminv.i32.v8i16(<8 x i16> %a1)
%0 = trunc i32 %vminv.i to i16
ret i16 %0
}
define i32 @test_vminvq_s32(<4 x i32> %a1) {
; CHECK: test_vminvq_s32
; CHECK: sminv.4s [[REGNUM:s[0-9]+]], v0
; CHECK-NEXT: fmov w0, [[REGNUM]]
; CHECK-NEXT: ret
entry:
%vminv.i = tail call i32 @llvm.aarch64.neon.sminv.i32.v4i32(<4 x i32> %a1)
ret i32 %vminv.i
}
declare i32 @llvm.aarch64.neon.sminv.i32.v4i32(<4 x i32>)
declare i32 @llvm.aarch64.neon.sminv.i32.v8i16(<8 x i16>)
declare i32 @llvm.aarch64.neon.sminv.i32.v16i8(<16 x i8>)
declare i32 @llvm.aarch64.neon.sminv.i32.v2i32(<2 x i32>)
declare i32 @llvm.aarch64.neon.sminv.i32.v4i16(<4 x i16>)
declare i32 @llvm.aarch64.neon.sminv.i32.v8i8(<8 x i8>)
|
