SPU LLVM: Minor SUMB AVX-512 path optimization

- Tweak shuffle to allow LLVM to emit a cheap blend instruction instead of the expensive VPERMI2W instruction
This commit is contained in:
Malcolm Jestadt 2023-01-21 11:40:18 -05:00 committed by Ivan
parent 6532d2b650
commit 813f7b50c1

View file

@ -7709,7 +7709,7 @@ public:
const auto ax = vdbpsadbw(a, zeroes, 0);
const auto bx = vdbpsadbw(b, zeroes, 0);
set_vr(op.rt, shuffle2(ax, bx, 0, 8, 2, 10, 4, 12, 6, 14));
set_vr(op.rt, shuffle2(ax, bx, 0, 9, 2, 11, 4, 13, 6, 15));
return;
}