This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: [patch] Vectorizer cost model implementation

From: "Andrew Pinski" <pinskia at gmail dot com>
To: "Jagasia, Harsha" <harsha dot jagasia at amd dot com>
Cc: "Dorit Nuzman" <DORIT at il dot ibm dot com>, "Linthicum, Tony" <tony dot linthicum at amd dot com>, "Meissner, Michael" <michael dot meissner at amd dot com>, "GCC Patches" <gcc-patches at gcc dot gnu dot org>
Date: Sat, 21 Apr 2007 23:07:28 -0700
Subject: Re: [patch] Vectorizer cost model implementation
References: <OF9F10D891.56E7C8DD-ONC22572B7.004FACB6-C22572B7.00508084@il.ibm.com> <D5B24B5251882048AD03DDFA431BB790012F00AE@SAUSEXMB3.amd.com>

On 4/21/07, Jagasia, Harsha <harsha.jagasia@amd.com> wrote:

For now, I have turned the cost model on in vect.exp for the vect-*
tests. For every vect-* test that failed, I duplicated it in
no-costmodel-vect-* test. For the duplicated test, I turned off the cost
model to allow it to pass.


I rather you do the opposite in that you create a costmodel-vect-* and
the tests that already exist just turn off the cost model.  This is
better to compare testresults between the trunk (and 4.3) and the
earlier releases.

Tommorrow (the 22nd) sometime I will run this patch on spu-elf and
report the results of the testsuite (for the vect-* tests).

Thanks,
Andrew Pinski

Follow-Ups:
- Re: [patch] Vectorizer cost model implementation
  - From: Eric Christopher

References:
- Re: [patch] Vectorizer cost model implementation
  - From: Dorit Nuzman
- RE: [patch] Vectorizer cost model implementation
  - From: Jagasia, Harsha

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]