This is an archive of the discontinued LLVM Phabricator instance.

[AMDGPU] Support horizontal vectorization of min/max.
ClosedPublic

Authored by FarhanaAleen on May 8 2018, 2:54 PM.

Diff Detail

Repository
rL LLVM

Event Timeline

FarhanaAleen created this revision.May 8 2018, 2:54 PM
FarhanaAleen created this object with visibility "No One".
FarhanaAleen created this object with edit policy "No One".
FarhanaAleen changed the visibility from "No One" to "Custom Policy".May 9 2018, 9:39 AM
FarhanaAleen changed the edit policy from "No One" to "Custom Policy".
FarhanaAleen retitled this revision from [AMDGPU] Support horizontal vectorization on min/max. to [AMDGPU] Support horizontal vectorization of min/max..May 9 2018, 10:11 AM
FarhanaAleen added a reviewer: rampitec.
FarhanaAleen removed a reviewer: rampitec.
FarhanaAleen changed the visibility from "Custom Policy" to "Custom Policy".
FarhanaAleen changed the visibility from "Custom Policy" to "Public (No Login Required)".
FarhanaAleen changed the edit policy from "Custom Policy" to "All Users".
FarhanaAleen edited subscribers, added: Restricted Project; removed: kzhuravl.
rampitec added inline comments.May 9 2018, 12:04 PM
test/CodeGen/AMDGPU/reduction.ll
132 ↗(On Diff #145792)

There is no guarantee RA will choose the same register as destination, neither it is needed.

153 ↗(On Diff #145792)

Same here.

225 ↗(On Diff #145792)

Ditto.

Removed the dependency from the destination register.

rampitec accepted this revision.May 9 2018, 1:13 PM

LGTM

This revision is now accepted and ready to land.May 9 2018, 1:13 PM
This revision was automatically updated to reflect the committed changes.