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
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=arm -mattr=+armv8.2-a,+fp-armv8,+fullfp16,+bf16,-neon %s -o - | FileCheck %s
; RUN: llc -mtriple=thumb -mattr=+armv8.2-a,+fp-armv8,+fullfp16,+bf16,-neon %s -o - | FileCheck %s
define arm_aapcscc half @f_t(half %x) nounwind {
; CHECK-LABEL: f_t:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vmov.f16 s0, r0
; CHECK-NEXT: @APP
; CHECK-NEXT: vsqrt.f16 s0, s0
; CHECK-NEXT: @NO_APP
; CHECK-NEXT: vmov r0, s0
; CHECK-NEXT: bx lr
entry:
%0 = tail call half asm "vsqrt.f16 $0, $1", "=t,t"(half %x)
ret half %0
}
define arm_aapcscc half @f_w(half %x) nounwind {
; CHECK-LABEL: f_w:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vmov.f16 s0, r0
; CHECK-NEXT: @APP
; CHECK-NEXT: vsqrt.f16 s0, s0
; CHECK-NEXT: @NO_APP
; CHECK-NEXT: vmov r0, s0
; CHECK-NEXT: bx lr
entry:
%0 = tail call half asm "vsqrt.f16 $0, $1", "=w,w"(half %x)
ret half %0
}
define arm_aapcscc bfloat @h_t(bfloat %x) nounwind {
; CHECK-LABEL: h_t:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vmov.f16 s0, r0
; CHECK-NEXT: @APP
; CHECK-NEXT: vmov.f32 s0, s0
; CHECK-NEXT: @NO_APP
; CHECK-NEXT: vmov.f16 r0, s0
; CHECK-NEXT: bx lr
entry:
%0 = tail call bfloat asm "vmov.f32 $0, $1", "=t,t"(bfloat %x)
ret bfloat %0
}
define arm_aapcscc bfloat @h_w(bfloat %x) nounwind {
; CHECK-LABEL: h_w:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vmov.f16 s0, r0
; CHECK-NEXT: @APP
; CHECK-NEXT: vmov.f32 s0, s0
; CHECK-NEXT: @NO_APP
; CHECK-NEXT: vmov.f16 r0, s0
; CHECK-NEXT: bx lr
entry:
%0 = tail call bfloat asm "vmov.f32 $0, $1", "=w,w"(bfloat %x)
ret bfloat %0
}
|
