This is an archive of the discontinued LLVM Phabricator instance.

Move getIntrinsicIDForCall from LoopVectorize to VectorUtil
ClosedPublic

Authored by karthikthecool on Apr 29 2014, 10:55 PM.

Details

Summary

Part 1/2 of [ Vectorize intrinsic math function calls in SLPVectorizer ]

Splitting patch in http://reviews.llvm.org/D3535

Move getIntrinsicIDForCall and supporting calls to VectorUtil. The same function can be used by SLPVectorizer to decide if a CallInst can be treated as Intrinsic or not.
No Functionality change.

Once we move the relevent functions to identify intrinsics into VectorUtil we will use the same in second part of patch were we vectorize math intrinsics in SLPVectorizer.

Thanks
Karthik Bhat

Diff Detail

Event Timeline

karthikthecool retitled this revision from to Move getIntrinsicIDForCall from LoopVectorize to VectorUtil.
karthikthecool updated this object.
karthikthecool edited the test plan for this revision. (Show Details)
karthikthecool added subscribers: Unknown Object (MLST), zinovy.nis.
nadav edited edge metadata.Apr 29 2014, 10:57 PM

This part LGTM.

karthikthecool accepted this revision.May 5 2014, 8:54 PM
karthikthecool added a reviewer: karthikthecool.
This revision is now accepted and ready to land.May 5 2014, 8:54 PM
karthikthecool closed this revision.May 5 2014, 8:54 PM