libretro/ppsspp
1
0
Fork 0
mirror of https://github.com/libretro/ppsspp.git synced 2025-02-12 20:40:49 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Small optimizations to the vertexjit.

Browse Source
This commit is contained in:
Unknown W. Brackets 2013-11-05 00:31:00 -08:00
parent f6662054bd
commit c7edf73cdb
1 changed files with 13 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

27
GPU/GLES/VertexDecoder.cpp
View File

@ -1034,16 +1034,13 @@ void VertexDecoderJitCache::Jit_NormalS16() {
}
void VertexDecoderJitCache::Jit_NormalFloat() {
//ADD(scratchReg, srcReg, dec_->nrmoff);
//LDMIA(scratchReg, false, 3, tempReg1, tempReg2, tempReg3);
//ADD(scratchReg, dstReg, dec_->decFmt.nrmoff);
//STMIA(scratchReg, false, 3, tempReg1, tempReg2, tempReg3);
// Might not be aligned to 4, so we can't use LDMIA.
LDR(tempReg1, srcReg, dec_->nrmoff);
LDR(tempReg2, srcReg, dec_->nrmoff + 4);
LDR(tempReg3, srcReg, dec_->nrmoff + 8);
STR(tempReg1, dstReg, dec_->decFmt.nrmoff);
STR(tempReg2, dstReg, dec_->decFmt.nrmoff + 4);
STR(tempReg3, dstReg, dec_->decFmt.nrmoff + 8);
// But this is always aligned to 4 so we're safe.
ADD(scratchReg, dstReg, dec_->decFmt.nrmoff);
STMIA(scratchReg, false, 3, tempReg1, tempReg2, tempReg3);
}
// Through expands into floats, always. Might want to look at changing this.
@ -1085,16 +1082,13 @@ void VertexDecoderJitCache::Jit_PosS16() {
// Just copy 12 bytes.
void VertexDecoderJitCache::Jit_PosFloat() {
//ADD(scratchReg, srcReg, dec_->posoff);
//LDMIA(scratchReg, false, 3, tempReg1, tempReg2, tempReg3);
//ADD(scratchReg, dstReg, dec_->decFmt.posoff);
//STMIA(scratchReg, false, 3, tempReg1, tempReg2, tempReg3);
// Might not be aligned to 4, so we can't use LDMIA.
LDR(tempReg1, srcReg, dec_->posoff);
LDR(tempReg2, srcReg, dec_->posoff + 4);
LDR(tempReg3, srcReg, dec_->posoff + 8);
STR(tempReg1, dstReg, dec_->decFmt.posoff);
STR(tempReg2, dstReg, dec_->decFmt.posoff + 4);
STR(tempReg3, dstReg, dec_->decFmt.posoff + 8);
// But this is always aligned to 4 so we're safe.
ADD(scratchReg, dstReg, dec_->decFmt.posoff);
STMIA(scratchReg, false, 3, tempReg1, tempReg2, tempReg3);
}
#elif defined(_M_X64) || defined(_M_IX86)
@ -1278,10 +1272,15 @@ void VertexDecoderJitCache::Jit_TcU16Through() {
}
void VertexDecoderJitCache::Jit_TcFloatThrough() {
#ifdef _M_X64
MOV(64, R(tempReg1), MDisp(srcReg, dec_->tcoff));
MOV(64, MDisp(dstReg, dec_->decFmt.uvoff), R(tempReg1));
#else
MOV(32, R(tempReg1), MDisp(srcReg, dec_->tcoff));
MOV(32, R(tempReg2), MDisp(srcReg, dec_->tcoff + 4));
MOV(32, MDisp(dstReg, dec_->decFmt.uvoff), R(tempReg1));
MOV(32, MDisp(dstReg, dec_->decFmt.uvoff + 4), R(tempReg2));
#endif
}
void VertexDecoderJitCache::Jit_Color8888() {