[PATCH] D63359: AMDGPU: Avoid most waitcnts before calls

[Matt Arsenault via Phabricator via llvm-commits]

arsenm created this revision.
arsenm added reviewers: rampitec, msearles, kerbowa.
Herald added subscribers: t-tye, tpr, dstuttard, yaxunl, nhaehnle, wdng, jvesely, kzhuravl.
arsenm updated this revision to Diff 204839.
arsenm added a comment.

Fix leftover junk in comment


Currently you get extra waits, because waits are inserted for the
register dependencies of the call, and the function prolog waits on
everything.

      

Currently waits are still inserted on returns. It may make sense to
not do this, and wait in the caller instead.


https://reviews.llvm.org/D63359

Files:
  lib/Target/AMDGPU/SIInsertWaitcnts.cpp
  test/CodeGen/AMDGPU/call-argument-types.ll
  test/CodeGen/AMDGPU/call-waitcnt.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D63359.204839.patch
Type: text/x-patch
Size: 16408 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20190614/7c1670c0/attachment.bin>