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

ARM Jit: Implement quaternion multiplication

Browse source
This commit is contained in:
Henrik Rydgard 2014-11-16 19:12:00 +01:00
parent bfcd3690b6
commit e43c7af32c
1 changed files with 36 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

39
Core/MIPS/ARM/ArmCompVFPU.cpp
View file

@ -1558,7 +1558,7 @@ namespace MIPSComp
fpr.MapRegsAndSpillLockV(tregs, sz, 0);
if (sz == V_Triple) {
int temp3 = fpr.GetTempV();
MIPSReg temp3 = fpr.GetTempV();
fpr.MapRegV(temp3, MAP_DIRTY | MAP_NOINIT);
// Cross product vcrsp.t
@ -1574,13 +1574,46 @@ namespace MIPSComp
VMUL(fpr.V(temp3), fpr.V(sregs[0]), fpr.V(tregs[1]));
VMLS(fpr.V(temp3), fpr.V(sregs[1]), fpr.V(tregs[0]));
fpr.MapRegsAndSpillLockV(dregs, V_Triple, MAP_DIRTY | MAP_NOINIT);
fpr.MapRegsAndSpillLockV(dregs, sz, MAP_DIRTY | MAP_NOINIT);
VMOV(fpr.V(dregs[0]), S0);
VMOV(fpr.V(dregs[1]), S1);
VMOV(fpr.V(dregs[2]), fpr.V(temp3));
} else if (sz == V_Quad) {
MIPSReg temp3 = fpr.GetTempV();
MIPSReg temp4 = fpr.GetTempV();
fpr.MapRegV(temp3, MAP_DIRTY | MAP_NOINIT);
fpr.MapRegV(temp4, MAP_DIRTY | MAP_NOINIT);
// Quaternion product vqmul.q untested
DISABLE;
// d[0] = s[0] * t[3] + s[1] * t[2] - s[2] * t[1] + s[3] * t[0];
VMUL(S0, fpr.V(sregs[0]), fpr.V(tregs[3]));
VMLA(S0, fpr.V(sregs[1]), fpr.V(tregs[2]));
VMLS(S0, fpr.V(sregs[2]), fpr.V(tregs[1]));
VMLA(S0, fpr.V(sregs[3]), fpr.V(tregs[0]));
//d[1] = -s[0] * t[2] + s[1] * t[3] + s[2] * t[0] + s[3] * t[1];
VNMUL(S1, fpr.V(sregs[0]), fpr.V(tregs[2]));
VMLA(S1, fpr.V(sregs[1]), fpr.V(tregs[3]));
VMLA(S1, fpr.V(sregs[2]), fpr.V(tregs[0]));
VMLA(S1, fpr.V(sregs[3]), fpr.V(tregs[1]));
//d[2] = s[0] * t[1] - s[1] * t[0] + s[2] * t[3] + s[3] * t[2];
VMUL(fpr.V(temp3), fpr.V(sregs[0]), fpr.V(tregs[1]));
VMLS(fpr.V(temp3), fpr.V(sregs[1]), fpr.V(tregs[0]));
VMLA(fpr.V(temp3), fpr.V(sregs[2]), fpr.V(tregs[3]));
VMLA(fpr.V(temp3), fpr.V(sregs[3]), fpr.V(tregs[2]));
//d[3] = -s[0] * t[0] - s[1] * t[1] - s[2] * t[2] + s[3] * t[3];
VNMUL(fpr.V(temp4), fpr.V(sregs[0]), fpr.V(tregs[0]));
VMLS(fpr.V(temp4), fpr.V(sregs[1]), fpr.V(tregs[1]));
VMLS(fpr.V(temp4), fpr.V(sregs[2]), fpr.V(tregs[2]));
VMLA(fpr.V(temp4), fpr.V(sregs[3]), fpr.V(tregs[3]));
fpr.MapRegsAndSpillLockV(dregs, sz, MAP_DIRTY | MAP_NOINIT);
VMOV(fpr.V(dregs[0]), S0);
VMOV(fpr.V(dregs[1]), S1);
VMOV(fpr.V(dregs[2]), fpr.V(temp3));
VMOV(fpr.V(dregs[3]), fpr.V(temp4));
}
fpr.ReleaseSpillLocksAndDiscardTemps();