This is the mail archive of the
mailing list for the GCC project.
Re: DFA for PPro, P2, P3
- From: Jan Hubicka <jh at suse dot cz>
- To: law at redhat dot com
- Cc: gcc at gcc dot gnu dot org
- Date: Fri, 3 May 2002 10:38:26 +0200
- Subject: Re: DFA for PPro, P2, P3
- References: <firstname.lastname@example.org>
> This is mainly an informative message for those who may still be struggling
> to understand situations where the DFA pipeline description is superior
> to the current pipeline description model.
I planed to do that today. Thanks for letting me know about the progress!
I've brought today the pentium/Athlon description with me, so I will send you
The Athlon description is complette and it appears to generate better
schedules, but I do have some testcases where it loses and some where it wins.
I don't unserstand the purpose for loosing as the schedule looks better, so I
plan to install it to the cfg-branch for tonight so it will get benchmarked by
Andreas tester and see what happends.
The Pentium description is missing MMX pipe I plan to do today.
Concerning the p01/p0 yes, we finally have way to describe this in meaningfull
way. Definitly cool!
All of the CPUs are having same situation.