AlxcBTWandClaude Opus 5.5 bc517d6058 fix(shader): give an exact zero mix f16 sum -0 when rounding toward -infinity ...

With f16 rounding toward -infinity, v_fma_mixlo_f16 and v_fma_mixhi_f16
gave +0 for an exact cancellation such as 1.0 * 1.0 + -1.0, because the
f64 sum is computed to nearest. RDNA2 gives -0 there, as IEEE 754 does
for an exact zero sum rounded toward -infinity, measured on gfx1036 in
#2631. An exact zero sum now gives -0 in that mode unless both the
product and the addend are +0. A nonzero result that rounds to zero
keeps its sign, so a tiny positive sum still gives +0.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

2026-10-11 14:59:56 +02:00

..

Decoder

…

libs

fix(shader): give an exact zero mix f16 sum -0 when rounding toward -infinity

2026-10-11 14:59:56 +02:00

relinker

Merge pull request #3495 from Lorkydey/fix/relinker-registry-input-preservation

2026-10-11 13:56:53 +02:00

shader/recompiler

fix(shader): give an exact zero mix f16 sum -0 when rounding toward -infinity

2026-10-11 14:59:56 +02:00