This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: RFA: patch to fix IRA performance degradation
- From: Peter Bergner <bergner at vnet dot ibm dot com>
- To: Vladimir Makarov <vmakarov at redhat dot com>
- Cc: gcc-patches <gcc-patches at gcc dot gnu dot org>, Richard Sandiford <rdsandiford at googlemail dot com>
- Date: Thu, 04 Sep 2008 22:50:46 -0500
- Subject: Re: RFA: patch to fix IRA performance degradation
- References: <48C071E5.5030804@redhat.com>
On Thu, 2008-09-04 at 19:40 -0400, Vladimir Makarov wrote:
> The following patch fixes significant performance degradation
> introduced by a recent patch to process allocnos on reverse traverse
> of BB insns. Originally the patch was created by Richard Sandiford. I
> just added a comment to ALLOCNO_ORDER_COMPARE.
>
> It was succefully bootstrapped on x86_64.
>
> 2008-09-04 Richard Sandiford <rdsandiford@googlemail.com>
> Vladimir Makarov <vmakarov@redhat.com>
>
> * ira-int.h (ALLOCNO_ORDER_COMPARE): New.
>
> * ira-color.c (allocno_cost_compare_func,
> bucket_allocno_compare_func, bucket_allocno_compare_func,
> push_allocnos_to_stack, allocno_priority_compare_func): Use it.
>
> * ira-build.c (ira_swap_allocno_copy_ends_if_necessary): Ditto.
As mentioned in the other thread, let's wait on this patch until we
have determined why this patch helps performance. It the end this
may be what we apply, but my gut feeling is this is papering over
some real problem.
Peter