This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Suspected performance regressions
- To: Joe Buck <jbuck at Synopsys dot COM>
- Subject: Re: Suspected performance regressions
- From: Jeffrey A Law <law at hurl dot cygnus dot com>
- Date: Thu, 14 Jan 1999 19:27:13 -0700
- cc: janr at molienergy dot bc dot ca (Jan Reimers), egcs at cygnus dot com
- Reply-To: law at cygnus dot com
In message <199901150003.QAA02041@atrus.synopsys.com>you write:
> Since Spec95 costs and is not free software, it can't be used
> "officially". However, I know that a number of people on this list
> have licensed Spec95, so they could run the tests locally. There
> are some rules for publishing results, and I don't know if sending them
> out on egcs constitutes publication, but the reason for Spec95 numbers
> is that it's an industry standard.
Unfortunately I don't have local access to spec95. I do have the much
older spec92 which I do benchmark with semi-regularly for the PA and x86.
Results are what one would expect, general improvements, a small number
of regressions. For the both the pa & x86 floating point has improved
more than integer over time. Probably because we've had much more room
to improve.
jeff