Emulator-Archive/ppsspp
1
0
Fork 0
mirror of https://github.com/hrydgard/ppsspp.git synced 2024-11-23 13:30:02 +00:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Interpreter: Optimize ReadVector/WriteVector by removing voffset lookups

Browse Source 
Drops these functions down the ranking of top functions by quite a bit in GTA,
speedup at most 0.5% though. But enough of these small ones and they
start adding up.

Not sure why GTA falls back to the interpreter for these so much though.
I guess some "uneaten" prefix..
This commit is contained in:
Henrik Rydgård 2023-10-05 19:07:01 +02:00
parent 60a304f29b
commit 0d06af87b6
1 changed files with 11 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
Core/MIPS/MIPSVFPUUtils.cpp
View File

@ -164,6 +164,7 @@ void GetMatrixRows(int matrixReg, MatrixSize msize, u8 vecs[4]) {
}
}
void ReadVector(float *rd, VectorSize size, int reg) {
int row;
int length;
@ -181,11 +182,11 @@ void ReadVector(float *rd, VectorSize size, int reg) {
if (transpose) {
const int base = mtx + col;
for (int i = 0; i < length; i++)
rd[i] = currentMIPS->v[base + ((row+i)&3) * 4];
rd[i] = currentMIPS->v[base + ((row + i) & 3) * 4];
} else {
const int base = mtx + col * 4;
for (int i = 0; i < length; i++)
rd[i] = currentMIPS->v[base + ((row+i)&3)];
rd[i] = currentMIPS->v[base + ((row + i) & 3)];
}
}
@ -201,30 +202,31 @@ void WriteVector(const float *rd, VectorSize size, int reg) {
default: length = 0; break;
}
const int mtx = reg & (7 << 2);
const int mtx = ((reg << 2) & 0x70);
const int col = reg & 3;
bool transpose = (reg >> 5) & 1;
// NOTE: We now skip the voffset lookups.
if (transpose) {
const int base = mtx + col * 32;
const int base = mtx + col;
if (currentMIPS->VfpuWriteMask() == 0) {
for (int i = 0; i < length; i++)
V(base + ((row+i)&3)) = rd[i];
currentMIPS->v[base + ((row+i) & 3) * 4] = rd[i];
} else {
for (int i = 0; i < length; i++) {
if (!currentMIPS->VfpuWriteMask(i)) {
V(base + ((row + i) & 3)) = rd[i];
currentMIPS->v[base + ((row+i) & 3) * 4] = rd[i];
}
}
}
} else {
const int base = mtx + col;
const int base = mtx + col * 4;
if (currentMIPS->VfpuWriteMask() == 0) {
for (int i = 0; i < length; i++)
V(base + ((row + i) & 3) * 32) = rd[i];
currentMIPS->v[base + ((row + i) & 3)] = rd[i];
} else {
for (int i = 0; i < length; i++) {
if (!currentMIPS->VfpuWriteMask(i)) {
V(base + ((row + i) & 3) * 32) = rd[i];
currentMIPS->v[base + ((row + i) & 3)] = rd[i];
}
}
}