[Piglit] [REGRESSION] runtime of gpu.py doubled by commit 26326d4032631cdb4eb6816aada6c81db6f186d7

[Dylan Baker]

On Monday, April 14, 2014 05:16:56 PM Michel Dänzer wrote:
> On 12.04.2014 04:35, Dylan Baker wrote:
> > On Tuesday, April 08, 2014 18:42:47 Michel Dänzer wrote:
> >  > I was away for the last two weeks, and when I came back I noticed that
> >  > gpu.py is now taking around 20 minutes to complete, whereas it only
> >  > took
> >  > around 10 minutes before. I was able to bisect this to:
> >  > 
> >  > 
> >  > 26326d4032631cdb4eb6816aada6c81db6f186d7 is the first bad commit
> >  > commit 26326d4032631cdb4eb6816aada6c81db6f186d7
> >  > Author: Dylan Baker <baker.dylan.c at gmail.com>
> >  > Date: Wed Feb 5 15:44:37 2014 -0800
> >  > 
> >  > framework/shader_test.py: simplify ShaderTest
> >  > 
> >  > 
> >  > This is on an A10-7850K (quad core Kaveri) APU FWIW.
> > 
> > I believe that I have found the bug. See if the patch labeled: [PATCH]
> > framework: Fix concurrency regression introduced by Test refactor
> > 
> > fixes this problem for you.
> 
> Looks like that was it, thanks.
> 
> Any idea why you didn't notice the regression on your machine(s)?

I always run with the -c switch (the always run everything concurrently 
whether they say they're safe or not) which is safe to use in a composited 
environment. Why I wasn't seeing it when I was running without the -c or -1 
switch I'm not sure.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 490 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.freedesktop.org/archives/piglit/attachments/20140414/4e700a2d/attachment.sig>