[PATCH] D104751: Unpack the CostEstimate feature in ML inlining models.

Jacob Hegna via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Jun 28 12:56:14 PDT 2021


jacobhegna updated this revision to Diff 355000.
jacobhegna added a comment.
Herald added a subscriber: mgorny.

Switch to visitor design pattern

Additionally, we re-use the heuristic based cost estimate feature in the model,
which is then supplemented by the expanded cost features. This increases
stability in training.

In a later patch, we will revisit completely removing the heuristics from the ML.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D104751/new/

https://reviews.llvm.org/D104751

Files:
  llvm/include/llvm/Analysis/InlineAdvisor.h
  llvm/include/llvm/Analysis/InlineCost.h
  llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
  llvm/lib/Analysis/CMakeLists.txt
  llvm/lib/Analysis/InlineCost.cpp
  llvm/lib/Analysis/MLInlineAdvisor.cpp

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D104751.355000.patch
Type: text/x-patch
Size: 26186 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20210628/58a5b07a/attachment.bin>


More information about the llvm-commits mailing list