Skip to content

Commit 59cf9c8

Browse files
arsenmsearlmc1
authored andcommitted
AMDGPU: Set max supported div/rem size to 64 (llvm#80669)
This enables IR expansion for i128 divisions. The vector case is still broken because ExpandLargeDivRem doesn't try to handle them. Fixes: SWDEV-426193 Change-Id: Ie1bf0f3ae1d199a120a75dc663c39642308ecec9
1 parent 26428d5 commit 59cf9c8

File tree

3 files changed

+5552
-6
lines changed

3 files changed

+5552
-6
lines changed

llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -550,6 +550,7 @@ AMDGPUTargetLowering::AMDGPUTargetLowering(const TargetMachine &TM,
550550
ISD::FSUB, ISD::FNEG,
551551
ISD::FABS, ISD::AssertZext,
552552
ISD::AssertSext, ISD::INTRINSIC_WO_CHAIN});
553+
setMaxDivRemBitWidthSupported(64);
553554
}
554555

555556
bool AMDGPUTargetLowering::mayIgnoreSignedZero(SDValue Op) const {

0 commit comments

Comments
 (0)