This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: IRA for GCC 4.4
Mark Mitchell wrote:
Chris Lattner wrote:
On Apr 24, 2008, at 12:53 AM, Mark Mitchell wrote:
* Ensure that IRA performance on SPEC is no worse than old-RA
performance on primary targets. Again, I hope and expect that the
target maintainers will work with you to gather data and investigate
any issues.
Are there any compile-time performance goals/requirements for this?
It would be nice to not take a 40% slowdown at -O0 for example (which
is apparently the case currently). What is the acceptable
compile-time hit at -O0 and -O3?
We did not explicitly state such goals, but, yes, I would consider a
40% slowdown a pretty serious issue. I'm personally less concerned
about -O3, but I would hope that we were in single digits in both cases.
Vladimir, what do you think about that?
I am going to put it definitely less than 10% in 3 weeks and to work
more to achieve the same time for -O0 on this extreme test.
By the way, there is no such big difference in -O2. I've just checked
this test for -O2 and the compiler with IRA is 3-4% slow.