Files
clang-p2996/llvm/test/CodeGen/AMDGPU/amdgpu-codegenprepare-idiv.ll
choikwa 463e93b95f Reapply [AMDGPU] prevent shrinking udiv/urem if either operand exceeds signed max (#119325)
This reverts commit 254d206ee2.

+Added a fix in ExpandDivRem24 to disqualify if DivNumBits exceed 24.

Original commit & msg:
ce6e955ac374f2b86cbbb73b2f32174dffd85f25.
Handle signed and unsigned path differently in getDivNumBits. Using
computeKnownBits, this rejects shrinking unsigned div/rem if operands
exceed signed max since we know NumSignBits will be always 0.
2024-12-12 15:24:34 -05:00

452 KiB