[Bf-cycles] CUDA Memory usage: Call for testing

Dan McGrath danmcgrath.ca at gmail.com
Sun Jan 10 21:59:34 CET 2016


Hey,

Some quick results (mostly shared on irc already)

  http://www.pasteall.org/63328


Dan


On Sun, Jan 10, 2016 at 3:44 PM Nazim Mer <nazim.mer at gmail.com> wrote:

> Would this be added into Opencl Split Kernel
>
> On 10 January 2016 at 20:06, Zauber Paracelsus <zauber at gridmail.org>
> wrote:
>
>> Here's my results, using a character model that uses the SSS node in its
>> skin shader.
>>
>> Resolution: 512x512
>> Samples: 16 Branched PT
>> Tile Size: 128x128
>> Graphics Card: GeForce GTX 650 Ti 2gb, factory-overclocked
>>
>> Supported Feature Set: (No SSS, unsupported by feature set):
>>  > Render Time: 0:42.63
>>  > Peak Memory 187.97mb
>>
>> Experimental Feature Set: (no SSS):
>>  > Render Time: 0:48.36
>>  > Peak Memory: 187.97mb
>>
>> Experimental Feature Set: (with SSS):
>>  > Render Time: 1:25.55
>>  > Peak Memory 187.97mb
>>
>> So, with no SSS, there is a few seconds increase in render time, and no
>> change in memory usage.  With SSS enabled, there is (naturally) an
>> increase in rendering time, but still no increase in memory usage.
>>
>>
>> Note: There appears to be a regression of some kind with combed hair.  I
>> will be filing a bug report shortly.
>> _______________________________________________
>> Bf-cycles mailing list
>> Bf-cycles at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-cycles
>>
>
> _______________________________________________
> Bf-cycles mailing list
> Bf-cycles at blender.org
> http://lists.blender.org/mailman/listinfo/bf-cycles
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/bf-cycles/attachments/20160110/7b5bb24c/attachment.htm 


More information about the Bf-cycles mailing list