blob: 4aeff09249556813545c9a880a2cffa924009339 (
plain)
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
|
; RUN: llc < %s -march=nvptx -mcpu=sm_20 | FileCheck %s
target triple = "nvptx-nvidia-cuda"
; Function Attrs: nounwind
; CHECK: .entry foo
define void @foo(float* nocapture %a) #0 {
%val = load float* %a
%tan = tail call fastcc float @__nv_fast_tanf(float %val)
store float %tan, float* %a
ret void
}
; Function Attrs: nounwind readnone
declare float @llvm.nvvm.sin.approx.ftz.f(float) #1
; Function Attrs: nounwind readnone
declare float @llvm.nvvm.cos.approx.ftz.f(float) #1
; Function Attrs: nounwind readnone
declare float @llvm.nvvm.div.approx.ftz.f(float, float) #1
; Function Attrs: alwaysinline inlinehint nounwind readnone
; CHECK: .func (.param .b32 func_retval0) __nv_fast_tanf
define internal fastcc float @__nv_fast_tanf(float %a) #2 {
entry:
%0 = tail call float @llvm.nvvm.sin.approx.ftz.f(float %a)
%1 = tail call float @llvm.nvvm.cos.approx.ftz.f(float %a)
%2 = tail call float @llvm.nvvm.div.approx.ftz.f(float %0, float %1)
ret float %2
}
attributes #0 = { nounwind }
attributes #1 = { nounwind readnone }
attributes #2 = { alwaysinline inlinehint nounwind readnone }
!nvvm.annotations = !{!0}
!0 = metadata !{void (float*)* @foo, metadata !"kernel", i32 1}
|
