mirror of
https://github.com/hrydgard/ppsspp.git
synced 2026-11-16 02:59:49 +01:00
IR: Fold ClampToZero into the 31-bit packs
Vec4ClampToZero and Vec2ClampToZero only ever fed Vec4Pack31To8 and Vec2Pack31To16, for vi2uc and vi2us. The packs now clamp negative lanes to zero themselves, which saves an op and a vector temp, and lets x64 clamp with PACKUSWB's saturation after an arithmetic shift. While at it, RISC-V compiles Vec2Unpack16To31, Vec2Pack31To16 and Vec4Pack32To8, and LoongArch Vec2Unpack16To31, Vec2Pack31To16 and the non-LSX Vec4Pack32To8, all of which went to the IR interpreter. LoongArch's Vec2Pack32To16 and Vec2Unpack16To32 now take their scalar path with LSX too, instead of falling back. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This commit is contained in:
1 parent
3db7c65e86
commit
eaf55c467b
15 files changed
+162
-213
No files matched your search
@@ -833,8 +833,6 @@ bool PropagateConstants(const IRWriter &in, IRWriter &out, const IROptions &opts
|
||||
case IROp::Vec4Unpack8To32:
|
||||
case IROp::Vec2Unpack16To32:
|
||||
case IROp::Vec4DuplicateUpperBitsAndShift1:
|
||||
case IROp::Vec2ClampToZero:
|
||||
case IROp::Vec4ClampToZero:
|
||||
out.Write(inst);
|
||||
break;
|
||||
|
||||
|
||||
Reference in new issue
Block a user