This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] fix PR opt/2391 (mostly ARM)
- From: Richard Henderson <rth at redhat dot com>
- To: Adam Nemet <anemet at Lnxw dot com>
- Cc: gcc-patches at gcc dot gnu dot org, rearnsha at arm dot com
- Date: Wed, 5 Feb 2003 14:50:31 -0800
- Subject: Re: [PATCH] fix PR opt/2391 (mostly ARM)
- References: <email@example.com> <20030125213629.GY9253@redhat.com> <firstname.lastname@example.org> <20030127184047.GC18724@redhat.com> <email@example.com>
On Mon, Jan 27, 2003 at 11:51:43PM -0800, Adam Nemet wrote:
> Maybe I wasn't clear enough in my first email when I explained the
> result of my analysis. The reason for the exponential behavior lies
> in the way reg_last_set_value is set up and not in the functions
> traversing it.
Ah, I see.
>  96.5 1.39 0.00 92+10486199 update_table_tick 
>  0.0 0.00 0.00 92+1027 update_table_tick 
Well that's certainly convincing.
So, go ahead and update the patch such that you're comparing against
the mode as well, and it can go in for 3.3 and mainline.