[AMDGPU] Run always inliner early in opt
ClosedPublic

Authored by kzhuravl on Mon, Mar 20, 10:59 AM.

Diff Detail

Repository
rL LLVM
kzhuravl created this revision.Mon, Mar 20, 10:59 AM

Not sure which bug you were trying to solve by original change, but moving if we move it "Internalize" then we would have to make function in early-inline.ll amdgpu_kernel. Would that be ok? I added a separate option for now.

rampitec added inline comments.Mon, Mar 20, 11:02 AM
lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
98 ↗(On Diff #92359)

No need in separate option. It falls under the same ideology as internalize. Having a separate option would mean we need pass it from RT as well.

Not sure which bug you were trying to solve by original change, but moving if we move it "Internalize" then we would have to make function in early-inline.ll amdgpu_kernel. Would that be ok? I added a separate option for now.

Not sure which bug you were trying to solve by original change, but moving if we move it "Internalize" then we would have to make function in early-inline.ll amdgpu_kernel. Would that be ok? I added a separate option for now.

Not sure which bug you were trying to solve by original change, but moving if we move it "Internalize" then we would have to make function in early-inline.ll amdgpu_kernel. Would that be ok? I added a separate option for now.

Yes, changing caller to be a kernel is good.

kzhuravl updated this revision to Diff 92363.Mon, Mar 20, 11:12 AM
kzhuravl marked an inline comment as done.

Address review feedback.

rampitec accepted this revision.Mon, Mar 20, 11:13 AM

Thanks!

This revision is now accepted and ready to land.Mon, Mar 20, 11:13 AM
This revision was automatically updated to reflect the committed changes.