121
|
1 ; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=FUNC %s
|
|
2 ; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=FUNC %s
|
|
3
|
|
4 declare double @llvm.convert.from.fp16.f64(i16) nounwind readnone
|
|
5
|
|
6 ; FUNC-LABEL: {{^}}test_convert_fp16_to_fp64:
|
|
7 ; GCN: buffer_load_ushort [[VAL:v[0-9]+]]
|
|
8 ; GCN: v_cvt_f32_f16_e32 [[RESULT32:v[0-9]+]], [[VAL]]
|
|
9 ; GCN: v_cvt_f64_f32_e32 [[RESULT:v\[[0-9]+:[0-9]+\]]], [[RESULT32]]
|
|
10 ; GCN: buffer_store_dwordx2 [[RESULT]]
|
|
11 define amdgpu_kernel void @test_convert_fp16_to_fp64(double addrspace(1)* noalias %out, i16 addrspace(1)* noalias %in) nounwind {
|
|
12 %val = load i16, i16 addrspace(1)* %in, align 2
|
|
13 %cvt = call double @llvm.convert.from.fp16.f64(i16 %val) nounwind readnone
|
|
14 store double %cvt, double addrspace(1)* %out, align 4
|
|
15 ret void
|
|
16 }
|