llvm/test/Transforms/LoopStrengthReduce, branch release_27http://llvm.orghttps://git.amat.us/llvm/atom/test/Transforms/LoopStrengthReduce?h=release_272010-03-02T01:59:21ZNon-affine post-inc SCEV expansions have more code which must be2010-03-02T01:59:21ZDan Gohmangohman@apple.com2010-03-02T01:59:21Zurn:sha1:069d6f3396856655d5d4ba155ee16eb0209d38b0
emitted after the increment. Make sure the insert position
reflects this. This fixes PR6453.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97537 91177308-0d34-0410-b5e6-96231b3b80d8
Reapply the new LoopStrengthReduction code, with compile time and2010-02-12T10:34:29ZDan Gohmangohman@apple.com2010-02-12T10:34:29Zurn:sha1:572645cf84060c0fc25cb91d38cb9079918b3a88
bug fixes, and with improved heuristics for analyzing foreign-loop
addrecs.
This change also flattens IVUsers, eliminating the stride-oriented
groupings, which makes it easier to work with.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95975 91177308-0d34-0410-b5e6-96231b3b80d8
-disable-output is no longer needed with -analyze.2010-01-26T19:25:59ZDan Gohmangohman@apple.com2010-01-26T19:25:59Zurn:sha1:8a977e2edd2ca182851cea96262ed9ba0c854c52
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@94574 91177308-0d34-0410-b5e6-96231b3b80d8
Revert LoopStrengthReduce.cpp to pre-r94061 for now.2010-01-22T00:46:49ZDan Gohmangohman@apple.com2010-01-22T00:46:49Zurn:sha1:7979b72febb73f7bb1d1ed095a68f210822b2e7c
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@94123 91177308-0d34-0410-b5e6-96231b3b80d8
Re-implement the main strength-reduction portion of LoopStrengthReduction.2010-01-21T02:09:26ZDan Gohmangohman@apple.com2010-01-21T02:09:26Zurn:sha1:a10756ee657a4d43a48cca5c166919093930ed6b
This new version is much more aggressive about doing "full" reduction in
cases where it reduces register pressure, and also more aggressive about
rewriting induction variables to count down (or up) to zero when doing so
reduces register pressure.
It currently uses fairly simplistic algorithms for finding reuse
opportunities, but it introduces a new framework allows it to combine
multiple strategies at once to form hybrid solutions, instead of doing
all full-reduction or all base+index.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@94061 91177308-0d34-0410-b5e6-96231b3b80d8
Add nounwinds.2010-01-19T21:51:51ZDan Gohmangohman@apple.com2010-01-19T21:51:51Zurn:sha1:8e1aa8edec61d4361c66ad8483bb94d76e55dec8
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93919 91177308-0d34-0410-b5e6-96231b3b80d8
Make several tests less fragile.2010-01-12T04:52:47ZDan Gohmangohman@apple.com2010-01-12T04:52:47Zurn:sha1:ad4f7a6882b892abef8e08014de3305832d2c715
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93230 91177308-0d34-0410-b5e6-96231b3b80d8
Use WriteAsOperand instead of getName() to print loop header names,2010-01-09T18:17:45ZDan Gohmangohman@apple.com2010-01-09T18:17:45Zurn:sha1:3073329c532aab96b58a322f2ac8bf93abefc60e
so that unnamed blocks are handled.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93059 91177308-0d34-0410-b5e6-96231b3b80d8
Generalize OptimizeLoopTermCond to optimize more loop terminating icmp to use postinc iv.2009-11-17T18:10:11ZEvan Chengevan.cheng@apple.com2009-11-17T18:10:11Zurn:sha1:076e085698c484354c9e131f1bd8fd001581397b
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89116 91177308-0d34-0410-b5e6-96231b3b80d8
Block terminator may be a switch.2009-11-11T00:00:21ZEvan Chengevan.cheng@apple.com2009-11-11T00:00:21Zurn:sha1:f40888dc9b2b1a28fb4e7ba9a1fc930461e251aa
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86761 91177308-0d34-0410-b5e6-96231b3b80d8