[Bf-cycles] Bf-cycles Digest, Vol 36, Issue 13
eibriel at yahoo.com.ar
Sun Apr 27 18:34:05 CEST 2014
I'm interested on testing on a Titan card, just let me know where
can I get a linux64 build!
El 27/04/14 07:00, bf-cycles-request at blender.org escribió:
> Send Bf-cycles mailing list submissions to
> bf-cycles at blender.org
> To subscribe or unsubscribe via the World Wide Web, visit
> or, via email, send a message with subject or body 'help' to
> bf-cycles-request at blender.org
> You can reach the person managing the list at
> bf-cycles-owner at blender.org
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Bf-cycles digest..."
> Today's Topics:
> 1. Re: CUDA performance tests (Brecht Van Lommel)
> Message: 1
> Date: Sat, 26 Apr 2014 15:33:59 +0200
> From: Brecht Van Lommel <brechtvanlommel at pandora.be>
> Subject: Re: [Bf-cycles] CUDA performance tests
> To: "bf-cycles at blender.org" <bf-cycles at blender.org>
> <CAKFUgC0qPs3DydkKrh_0ap5H5mkyAkdKjfA1FUBWeiwYf-HFzQ at mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
> For the two cards I have here, it seems that after tweaking the code,
> CUDA 6.0 performance is similar to previous releases with 5.0. There's
> no improvement and even some small slowdown, but we actually had to do
> quite a bit work since 5.5 was released to even get it to that point,
> and some features have been added in the meantime. Overall I think
> it's acceptable if other cards show similar results.
> The patches are from an earlier test I did with the CUDA 6.0 beta,
> based on feedback from NVidia. They help on sm_30 and sm_35, but not
> sm_20 and sm_21 (these changes were also tested on 480 GTX and Tesla
> K20m). So it seems we should enable them for sm_3x if we switch to
> CUDA 6.0.
> We need to test more cards and platforms, comparing latest master +
> CUDA 6.0 to the 2.69 and 2.70a releases, before we can make a
> On Fri, Apr 25, 2014 at 5:44 PM, Brecht Van Lommel
> <brechtvanlommel at pandora.be> wrote:
>> Hi all,
>> We're having some trouble with CUDA performance at the moment, with in
>> a performance regression on Titan cards compared to 2.69, and unclear
>> effects from recent changes. Further there's the new CUDA 6.0 toolkit,
>> which we ideally could move to, as sticking to 5.0 is a problem with
>> newer compilers.
>> We could use some help testing all this, especially for Titan cards as
>> I don't have one of those.
>> There's a google doc here:
>> For now I'm assuming someone with a Titan card can make own builds to
>> test these git revisions. If not we can build and upload binaries to
>> The test is:
>> * download: http://www.pasteall.org/blend/28679
>> * run: ./blender -b modified_bmw.blend -f 1
>> If you've got a different .blend file that is giving performance
>> regressions, or you find different behavior when rendering in the UI
>> rather than command line, then results from that are welcome too.
>> The information that I'm looking for from this is especially:
>> * Which revision caused the Titan performance regressions?
>> * Is CUDA 6.0 performance acceptable to us for the next release?
>> * Were there significant performance regressions due to recent changes?
>> * Which effect do the two provided patches have when building with CUDA 6.0?
> Bf-cycles mailing list
> Bf-cycles at blender.org
> End of Bf-cycles Digest, Vol 36, Issue 13
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Bf-cycles