[PATCH] Reimplement heuristic for estimating complete-unroll optimization effects.

Michael Zolotukhin mzolotukhin at apple.com
Mon Feb 23 21:08:13 PST 2015


Addressing Chandler's remarks, continued:

- Inline findConstantPointers into GEP-instructions visitor.
- Inline simulateLoop into analyzeLoop + some other simplifications.
- Don't try to handle early exits.


http://reviews.llvm.org/D7837

Files:
  lib/Transforms/Scalar/LoopUnrollPass.cpp
  test/Transforms/LoopUnroll/full-unroll-no-taken-branch.ll

EMAIL PREFERENCES
  http://reviews.llvm.org/settings/panel/emailpreferences/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: D7837.20569.patch
Type: text/x-patch
Size: 29966 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150224/1ea1ccfe/attachment.bin>


More information about the llvm-commits mailing list