This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Modulo-scheduling improvements. Patch 2 of 2.
- From: "Andrew Pinski" <pinskia at gmail dot com>
- To: "David Edelsohn" <dje at watson dot ibm dot com>, Trevor_Smigiel at playstation dot sony dot com
- Cc: "Ayal Zaks" <ZAKS at il dot ibm dot com>, "Andrew Pinski" <andrew_pinski at playstation dot sony dot com>, "Vladimir Yanovsky" <volodyan at gmail dot com>, "Andrey Belevantsev" <abel at ispras dot ru>, "Dorit Nuzman" <dorit at il dot ibm dot com>, gcc-patches at gcc dot gnu dot org, "Mircea Namolaru" <NAMOLARU at il dot ibm dot com>, "Vladimir Yanovsky" <yanov at il dot ibm dot com>, vmakarov at toronto dot redhat dot com, "Zdenek Dvorak" <rakdver at atrey dot karlin dot mff dot cuni dot cz>
- Date: Fri, 22 Jun 2007 16:00:04 -0700
- Subject: Re: [PATCH] Modulo-scheduling improvements. Patch 2 of 2.
- References: <ZAKS@il.ibm.com> <OFA422CFCB.1CE8002F-ONC2257300.0073C7DD-C2257300.0076BBB5@il.ibm.com> <200706221722.l5MHMmQ30736@makai.watson.ibm.com>
On 6/22/07, David Edelsohn <dje@watson.ibm.com> wrote:
>>>>> Ayal Zaks writes:
> Note that you need an spu maintainer to approve the spu.md part, and a loop
> optimizer maintainer to approve the doloop part. David, Zdenek, can you
> look into these parts, attached below, and see if they are acceptable?
Andrew Pinski would be the best person to review the SPU part of
the patch.
Note the patch can be found at
http://gcc.gnu.org/ml/gcc-patches/2007-01/msg01470.html
Actually Trevor would be a better person in this case. This is
because I remember that Trevor found the doloop_end pattern actually
caused performance regressions but I don't have further information
than just that, sorry.
I also will note that we really don't need a doloop pattern to find
loops or to look at the iteration of loops at the RTL level. cfgloop.h
contains all the functions we could use to find the loops and their
iterations. In fact this is how loop-doloop.c (which is 640 lines
long) finds those loops.
Please read loop-doloop.c (specifically doloop_optimize) for a way to
find the loops and iterations.
If we do add a do_loop_end pattern to the spu back-end in the end, can
we then make it dependent it on sms being enabled?
Thanks,
Andrew Pinski