mirror of
https://github.com/c64scene-ar/llvm-6502.git
synced 2024-08-26 23:29:22 +00:00
This eliminates extra extract instructions when loading an i8 vector to a float vector. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@210666 91177308-0d34-0410-b5e6-96231b3b80d8
43 lines
1.7 KiB
LLVM
43 lines
1.7 KiB
LLVM
; RUN: llc -march=r600 -mcpu=SI < %s | FileCheck -check-prefix=SI %s
|
|
|
|
declare float @llvm.AMDGPU.cvt.f32.ubyte0(i32) nounwind readnone
|
|
declare float @llvm.AMDGPU.cvt.f32.ubyte1(i32) nounwind readnone
|
|
declare float @llvm.AMDGPU.cvt.f32.ubyte2(i32) nounwind readnone
|
|
declare float @llvm.AMDGPU.cvt.f32.ubyte3(i32) nounwind readnone
|
|
|
|
; SI-LABEL: @test_unpack_byte0_to_float:
|
|
; SI: V_CVT_F32_UBYTE0
|
|
define void @test_unpack_byte0_to_float(float addrspace(1)* %out, i32 addrspace(1)* %in) nounwind {
|
|
%val = load i32 addrspace(1)* %in, align 4
|
|
%cvt = call float @llvm.AMDGPU.cvt.f32.ubyte0(i32 %val) nounwind readnone
|
|
store float %cvt, float addrspace(1)* %out, align 4
|
|
ret void
|
|
}
|
|
|
|
; SI-LABEL: @test_unpack_byte1_to_float:
|
|
; SI: V_CVT_F32_UBYTE1
|
|
define void @test_unpack_byte1_to_float(float addrspace(1)* %out, i32 addrspace(1)* %in) nounwind {
|
|
%val = load i32 addrspace(1)* %in, align 4
|
|
%cvt = call float @llvm.AMDGPU.cvt.f32.ubyte1(i32 %val) nounwind readnone
|
|
store float %cvt, float addrspace(1)* %out, align 4
|
|
ret void
|
|
}
|
|
|
|
; SI-LABEL: @test_unpack_byte2_to_float:
|
|
; SI: V_CVT_F32_UBYTE2
|
|
define void @test_unpack_byte2_to_float(float addrspace(1)* %out, i32 addrspace(1)* %in) nounwind {
|
|
%val = load i32 addrspace(1)* %in, align 4
|
|
%cvt = call float @llvm.AMDGPU.cvt.f32.ubyte2(i32 %val) nounwind readnone
|
|
store float %cvt, float addrspace(1)* %out, align 4
|
|
ret void
|
|
}
|
|
|
|
; SI-LABEL: @test_unpack_byte3_to_float:
|
|
; SI: V_CVT_F32_UBYTE3
|
|
define void @test_unpack_byte3_to_float(float addrspace(1)* %out, i32 addrspace(1)* %in) nounwind {
|
|
%val = load i32 addrspace(1)* %in, align 4
|
|
%cvt = call float @llvm.AMDGPU.cvt.f32.ubyte3(i32 %val) nounwind readnone
|
|
store float %cvt, float addrspace(1)* %out, align 4
|
|
ret void
|
|
}
|