[llvm-commits] [llvm] r163968 - /llvm/trunk/lib/Transforms/IPO/PassManagerBuilder.cpp

[Chandler Carruth]

Thanks for disabling this after the bots had their fun.

An update on these issues:

On Sat, Sep 15, 2012 at 8:11 AM, Benjamin Kramer
<benny.kra at googlemail.com>wrote:

> What we have so far:
> - Some clang test failures (these were known already)
>
> - Perf results are mixed, some big regressions
>   http://llvm.org/perf/db_default/v4/nts/3844
>   http://llvm.org/perf/db_default/v4/nts/3845
>
>   bullet suffers a lot. matmul is interesting: slower scalar code, faster
> with -vectorize.
>

matmul is a RA deficiency in scalar code. the spills are poorly selected /
scheduled and the result is slower code. The IR is *significantly* better,
which is what allows the vectorizer to go to town on it.

bullet is a bug in the new SROA pass that I'm working on. we essentially
fail to do SROA on most of the most important functions. wow bad.


> - Some dragonegg selfhost bots crash in SROA during selfhost now
>
> http://lab.llvm.org:8011/builders/dragonegg-x86_64-linux-gcc-4.6-self-host-checks/builds/1632
>
> http://lab.llvm.org:8011/builders/dragonegg-x86_64-linux-gcc-4.5-self-host/builds/1891


Duncan has sent me some repro steps and I'm going to look into that next
too.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20120915/b9b5edbc/attachment.html>