libretro/ppsspp
1
0
Fork 0
mirror of https://github.com/libretro/ppsspp.git synced 2025-02-02 14:52:20 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Minor x86 optimizations

Browse Source
This commit is contained in:
Henrik Rydgård 2013-11-25 16:13:52 +01:00
parent 13f979af8c
commit 9aa0c9a489
1 changed files with 11 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
GPU/GLES/VertexDecoderX86.cpp
View File

@ -354,10 +354,10 @@ void VertexDecoderJitCache::Jit_WeightsU8Skin() {
if (j == 0) {
MOVAPS(XMM4, MDisp(tempReg2, 0));
MOVAPS(XMM5, MDisp(tempReg2, 16));
MULPS(XMM4, R(XMM1));
MULPS(XMM5, R(XMM1));
MOVAPS(XMM6, MDisp(tempReg2, 32));
MOVAPS(XMM7, MDisp(tempReg2, 48));
MULPS(XMM4, R(XMM1));
MULPS(XMM5, R(XMM1));
MULPS(XMM6, R(XMM1));
MULPS(XMM7, R(XMM1));
} else {
@ -392,10 +392,10 @@ void VertexDecoderJitCache::Jit_WeightsU16Skin() {
if (j == 0) {
MOVAPS(XMM4, MDisp(tempReg2, 0));
MOVAPS(XMM5, MDisp(tempReg2, 16));
MULPS(XMM4, R(XMM1));
MULPS(XMM5, R(XMM1));
MOVAPS(XMM6, MDisp(tempReg2, 32));
MOVAPS(XMM7, MDisp(tempReg2, 48));
MULPS(XMM4, R(XMM1));
MULPS(XMM5, R(XMM1));
MULPS(XMM6, R(XMM1));
MULPS(XMM7, R(XMM1));
} else {
@ -428,10 +428,10 @@ void VertexDecoderJitCache::Jit_WeightsFloatSkin() {
if (j == 0) {
MOVAPS(XMM4, MDisp(tempReg2, 0));
MOVAPS(XMM5, MDisp(tempReg2, 16));
MULPS(XMM4, R(XMM1));
MULPS(XMM5, R(XMM1));
MOVAPS(XMM6, MDisp(tempReg2, 32));
MOVAPS(XMM7, MDisp(tempReg2, 48));
MULPS(XMM4, R(XMM1));
MULPS(XMM5, R(XMM1));
MULPS(XMM6, R(XMM1));
MULPS(XMM7, R(XMM1));
} else {
@ -709,16 +709,15 @@ void VertexDecoderJitCache::Jit_NormalFloat() {
// This could be a bit shorter with AVX 3-operand instructions and FMA.
void VertexDecoderJitCache::Jit_WriteMatrixMul(int outOff, bool pos) {
MOVAPS(XMM1, R(XMM3));
MOVAPS(XMM2, R(XMM3));
SHUFPS(XMM1, R(XMM1), _MM_SHUFFLE(0, 0, 0, 0));
MULPS(XMM1, R(XMM4));
MOVAPS(XMM2, R(XMM3));
SHUFPS(XMM2, R(XMM2), _MM_SHUFFLE(1, 1, 1, 1));
SHUFPS(XMM3, R(XMM3), _MM_SHUFFLE(2, 2, 2, 2));
MULPS(XMM1, R(XMM4));
MULPS(XMM2, R(XMM5));
MULPS(XMM3, R(XMM6));
ADDPS(XMM1, R(XMM2));
MOVAPS(XMM2, R(XMM3));
SHUFPS(XMM2, R(XMM2), _MM_SHUFFLE(2, 2, 2, 2));
MULPS(XMM2, R(XMM6));
ADDPS(XMM1, R(XMM2));
ADDPS(XMM1, R(XMM3));
if (pos) {
ADDPS(XMM1, R(XMM7));
}