Emulation/ppsspp
1
0
Fork 0
mirror of https://github.com/hrydgard/ppsspp.git synced 2025-04-02 11:01:50 -04:00
Code Issues Projects Releases Packages Wiki Activity

vertexjit: Use SSE4 in tc u8/u16 code.

Browse source 
It's not much different for non-SSE 4, might as well be common.
This commit is contained in:
Unknown W. Brackets 2014-09-17 23:15:24 -07:00
parent 5276487611
commit 967df9c018
1 changed files with 4 additions and 14 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

18
GPU/Common/VertexDecoderX86.cpp
View file

@ -511,23 +511,13 @@ void VertexDecoderJitCache::Jit_TcU16() {
}
void VertexDecoderJitCache::Jit_TcU8ToFloat() {
// TODO: The first five instructions could be done in 1 or 2 in SSE4
MOVZX(32, 8, tempReg1, MDisp(srcReg, dec_->tcoff));
MOVZX(32, 8, tempReg2, MDisp(srcReg, dec_->tcoff + 1));
CVTSI2SS(fpScratchReg, R(tempReg1));
CVTSI2SS(fpScratchReg2, R(tempReg2));
UNPCKLPS(fpScratchReg, R(fpScratchReg2));
MULPS(fpScratchReg, M(&by128));
MOVQ_xmm(MDisp(dstReg, dec_->decFmt.uvoff), fpScratchReg);
Jit_AnyU8ToFloat(dec_->tcoff, 16);
MOVQ_xmm(MDisp(dstReg, dec_->decFmt.uvoff), XMM3);
}
void VertexDecoderJitCache::Jit_TcU16ToFloat() {
PXOR(fpScratchReg2, R(fpScratchReg2));
MOVD_xmm(fpScratchReg, MDisp(srcReg, dec_->tcoff));
PUNPCKLWD(fpScratchReg, R(fpScratchReg2));
CVTDQ2PS(fpScratchReg, R(fpScratchReg));
MULPS(fpScratchReg, M(&by32768));
MOVQ_xmm(MDisp(dstReg, dec_->decFmt.uvoff), fpScratchReg);
Jit_AnyU16ToFloat(dec_->tcoff, 32);
MOVQ_xmm(MDisp(dstReg, dec_->decFmt.uvoff), XMM3);
}
void VertexDecoderJitCache::Jit_TcU16Double() {