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
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc < %s -mtriple=nvptx64 -mcpu=sm_80 -mattr=+ptx71 --enable-unsafe-fp-math | FileCheck --check-prefixes=CHECK %s
; RUN: %if ptxas-11.8 %{ llc < %s -mtriple=nvptx64 -mcpu=sm_80 -mattr=+ptx71 --enable-unsafe-fp-math | %ptxas-verify -arch=sm_80 %}
target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
declare <2 x bfloat> @llvm.sin.f16(<2 x bfloat> %a) #0
declare <2 x bfloat> @llvm.cos.f16(<2 x bfloat> %a) #0
define <2 x bfloat> @test_sin(<2 x bfloat> %a) #0 #1 {
; CHECK-LABEL: test_sin(
; CHECK: {
; CHECK-NEXT: .reg .b16 %rs<3>;
; CHECK-NEXT: .reg .b32 %r<3>;
; CHECK-NEXT: .reg .f32 %f<5>;
; CHECK-EMPTY:
; CHECK-NEXT: // %bb.0:
; CHECK-NEXT: ld.param.b32 %r1, [test_sin_param_0];
; CHECK-NEXT: mov.b32 {%rs1, %rs2}, %r1;
; CHECK-NEXT: cvt.f32.bf16 %f1, %rs1;
; CHECK-NEXT: sin.approx.f32 %f2, %f1;
; CHECK-NEXT: cvt.f32.bf16 %f3, %rs2;
; CHECK-NEXT: sin.approx.f32 %f4, %f3;
; CHECK-NEXT: cvt.rn.bf16x2.f32 %r2, %f4, %f2;
; CHECK-NEXT: st.param.b32 [func_retval0], %r2;
; CHECK-NEXT: ret;
%r = call <2 x bfloat> @llvm.sin.f16(<2 x bfloat> %a)
ret <2 x bfloat> %r
}
define <2 x bfloat> @test_cos(<2 x bfloat> %a) #0 #1 {
; CHECK-LABEL: test_cos(
; CHECK: {
; CHECK-NEXT: .reg .b16 %rs<3>;
; CHECK-NEXT: .reg .b32 %r<3>;
; CHECK-NEXT: .reg .f32 %f<5>;
; CHECK-EMPTY:
; CHECK-NEXT: // %bb.0:
; CHECK-NEXT: ld.param.b32 %r1, [test_cos_param_0];
; CHECK-NEXT: mov.b32 {%rs1, %rs2}, %r1;
; CHECK-NEXT: cvt.f32.bf16 %f1, %rs1;
; CHECK-NEXT: cos.approx.f32 %f2, %f1;
; CHECK-NEXT: cvt.f32.bf16 %f3, %rs2;
; CHECK-NEXT: cos.approx.f32 %f4, %f3;
; CHECK-NEXT: cvt.rn.bf16x2.f32 %r2, %f4, %f2;
; CHECK-NEXT: st.param.b32 [func_retval0], %r2;
; CHECK-NEXT: ret;
%r = call <2 x bfloat> @llvm.cos.f16(<2 x bfloat> %a)
ret <2 x bfloat> %r
}
|
