This is the mail archive of the
mailing list for the GCC project.
Re: IRA improvements 0/4
- From: Vladimir Makarov <vmakarov at redhat dot com>
- To: Mark Mitchell <mark at codesourcery dot com>
- Cc: gcc-patches <gcc-patches at gcc dot gnu dot org>, Jeffrey Law <law at redhat dot com>
- Date: Wed, 07 Jul 2010 11:44:13 -0400
- Subject: Re: IRA improvements 0/4
- References: <4C34818E.email@example.com> <4C349482.firstname.lastname@example.org>
Mark Mitchell wrote:
Vladimir Makarov wrote:For ppc64 (G5 machine) I've got more 1% improvement on SPECFP2000 and
0.3% degradation on SPECINT2000.
The branch was tested on x86/x86_64/ppc64/arm and benchmarked on
SPEC2000 and SPEC2006 on x86_64. For x86_64 SPEC20006 (Core i7) the
branch creates smaller size code (about 0.2%), needs less 1 sec for
compilation, and generates the same INT score and 1% better FP score.
Those are great results!
Would you please also post scores from at least one embedded RISC
architecture, e.g., ARM, MIPS, Power? Even x86_64 is a bit of a
different beast from these, and before we merge this in, I think we
should know what this does elsewhere.
I checked SPECInt2000 on ARM (Sheeva plug) about 3 weeks ago and as I
remember SPECInt2000 score was not worse or even better on the branch.
Unfortunately, this machine is not available right now
(gcc56.fsffrance.org). Otherwise, I could give more detail data.