Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Handle vrot overlap and vscl/vmscl prefixes more accurately #16302

Merged
merged 2 commits into from
Oct 30, 2022
Merged

Handle vrot overlap and vscl/vmscl prefixes more accurately #16302

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3f99751
irjit: Handle vrot overlap more correctly.
unknownbrackets Oct 30, 2022
bbdc8a8
interp: Correct vscl/vmscl t prefix handling.
unknownbrackets Oct 30, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
interp: Correct vscl/vmscl t prefix handling. 
This makes more sense.  Fixes Dissidia 012 issues.
  • Loading branch information
@unknownbrackets
unknownbrackets committed Oct 30, 2022
commit bbdc8a8f9847ba777056e8f11f9406cc1c8e18c0
4 changes: 2 additions & 2 deletions Core/MIPS/IR/IRCompVFPU.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1185,8 +1185,8 @@ namespace MIPSComp {
int vt = _VT;
u8 sregs[4], dregs[4], treg;
GetVectorRegsPrefixS(sregs, sz, vs);
// TODO: Prefixes seem strange...
GetVectorRegsPrefixT(&treg, V_Single, vt);
// T prefixes handled by interp.
GetVectorRegs(&treg, V_Single, vt);
GetVectorRegsPrefixD(dregs, sz, vd);

bool overlap = false;
Expand Down
9 changes: 6 additions & 3 deletions Core/MIPS/MIPSIntVFPU.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -540,8 +540,10 @@ namespace MIPSInt
ApplySwizzleS(&s[(n - 1) * 4], V_Quad);
// T prefix applies only for the last row, and is used per element.
// This is like vscl, but instead of zzzz it uses xxxx.
int tlane = (vt >> 5) & 3;
t[tlane] = t[0];
u32 tprefixRemove = VFPU_ANY_SWIZZLE();
u32 tprefixAdd = VFPU_SWIZZLE(0, 0, 0, 0);
u32 tprefixAdd = VFPU_SWIZZLE(tlane, tlane, tlane, tlane);
ApplyPrefixST(t, VFPURewritePrefix(VFPU_CTRL_TPREFIX, tprefixRemove, tprefixAdd), V_Quad);

for (int b = 0; b < n; b++) {
Expand Down Expand Up @@ -1518,9 +1520,10 @@ namespace MIPSInt

// T prefix forces swizzle (zzzz for some reason, so we force V_Quad.)
// That means negate still works, but constants are a bit weird.
t[2] = V(vt);
int tlane = (vt >> 5) & 3;
t[tlane] = V(vt);
u32 tprefixRemove = VFPU_ANY_SWIZZLE();
u32 tprefixAdd = VFPU_SWIZZLE(2, 2, 2, 2);
u32 tprefixAdd = VFPU_SWIZZLE(tlane, tlane, tlane, tlane);
ApplyPrefixST(t, VFPURewritePrefix(VFPU_CTRL_TPREFIX, tprefixRemove, tprefixAdd), V_Quad);

int n = GetNumVectorElements(sz);
Expand Down