This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa] Alias analysis speedups
- From: law at redhat dot com
- To: Toon Moene <toon at moene dot indiv dot nluug dot nl>
- Cc: Diego Novillo <dnovillo at redhat dot com>, Graham Stott <graham dot stott at btinternet dot com>, "gcc-patches at gcc dot gnu dot org" <gcc-patches at gcc dot gnu dot org>
- Date: Fri, 21 Feb 2003 15:26:18 -0700
- Subject: Re: [tree-ssa] Alias analysis speedups
- Reply-to: law at redhat dot com
In message <3E5685F5 dot 8000300 at moene dot indiv dot nluug dot nl>, Toon Moene writes:
>law at redhat dot com wrote:
>
>> In message <1045844562 dot 8671 dot 18 dot camel at shadowfax>, Diego Novillo writes:
>
>> >Mixed blessing in this case. Better alias info may lead to longer
>> >compile times and memory problems. Hence our PHI pruning heuristics.
>
>> Yes, but those are very extreme cases.
>
>"Elk nadeel hep se foordeel" :-)
>
>What would this mean for an extremely non-aliasing language like Fortran ?
>
>Inquiring minds want to know ...
I don't expect it will be a problem for Fortran. Unless you create
something which is roughly equivalent to 20001226-1.c.
Specifically you'd need a boatload (read tends of thousands) of
array references, without any stores into the array to make PHI
node insertion go crazy. And remember go crazy in this sense means
that it takes a little while. It still completes thanks to the
heuristics to switch to fully-pruned PHI insertions.
jeff