mirrors/rpcs3
1
0
Fork 0
mirror of https://github.com/RPCS3/rpcs3.git synced 2025-04-19 19:15:26 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

SPU LLVM: Don't use vperm2b256to128 outside of CPUTranslator.h

Browse source 
- The vperm2b function uses vperm2b256to128 when it's ideal
This commit is contained in:
Malcolm Jestadt 2025-03-24 07:56:17 -04:00
parent ff135384ef
commit 37718e568f
1 changed files with 6 additions and 6 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

12
rpcs3/Emu/Cell/SPULLVMRecompiler.cpp
View file

@ -5870,13 +5870,13 @@ public:
{
if (perm_only)
{
set_vr(op.rt4, vperm2b256to128(as, b, c));
set_vr(op.rt4, vperm2b(as, b, c));
return;
}
const auto m = gf2p8affineqb(c, build<u8[16]>(0x40, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x40, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20), 0x7f);
const auto mm = select(noncast<s8[16]>(m) >= 0, splat<u8[16]>(0), m);
const auto ab = vperm2b256to128(as, b, c);
const auto ab = vperm2b(as, b, c);
set_vr(op.rt4, select(noncast<s8[16]>(c) >= 0, ab, mm));
return;
}
@ -5920,13 +5920,13 @@ public:
{
if (perm_only)
{
set_vr(op.rt4, vperm2b256to128(a, b, eval(c ^ 0xf)));
set_vr(op.rt4, vperm2b(a, b, eval(c ^ 0xf)));
return;
}
const auto m = gf2p8affineqb(c, build<u8[16]>(0x40, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x40, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20), 0x7f);
const auto mm = select(noncast<s8[16]>(m) >= 0, splat<u8[16]>(0), m);
const auto ab = vperm2b256to128(a, b, eval(c ^ 0xf));
const auto ab = vperm2b(a, b, eval(c ^ 0xf));
set_vr(op.rt4, select(noncast<s8[16]>(c) >= 0, ab, mm));
return;
}
@ -5938,13 +5938,13 @@ public:
{
if (perm_only)
{
set_vr(op.rt4, vperm2b256to128(b, a, eval(c ^ 0x1f)));
set_vr(op.rt4, vperm2b(b, a, eval(c ^ 0x1f)));
return;
}
const auto m = gf2p8affineqb(c, build<u8[16]>(0x40, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x40, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20), 0x7f);
const auto mm = select(noncast<s8[16]>(m) >= 0, splat<u8[16]>(0), m);
const auto ab = vperm2b256to128(b, a, eval(c ^ 0x1f));
const auto ab = vperm2b(b, a, eval(c ^ 0x1f));
set_vr(op.rt4, select(noncast<s8[16]>(c) >= 0, ab, mm));
return;
}