[Bf-cycles] Issue with rendering using GPU's

Marco G marco.gzt at gmail.com
Wed Nov 19 17:40:52 CET 2014


Is Experimental feature kernel disabled? (switch to Supported).
Experimental needs quite a lot of memory.

MG

2014-11-19 17:38 GMT+01:00 Josh Cullum <josh.cullum at cfms.org.uk>:

> Hi Sergey,
>
> The blend file is here: http://www.pasteall.org/blend/32955
>
> In Blender, I'm getting the memory as 1516.46M, but both GPU's have 6GB
> memory each.
>
> Kind Regards,
>
> Josh Cullum
>
> On 19 November 2014 16:26, Sergey Sharybin <sergey.vfx at gmail.com> wrote:
>
>> My first suggestion would be to check GPU memory usage.
>>
>> But you referenced to attached .blend. Out ML doesn't allow to do big
>> attachments, so do you mind re-uploading it somewhere (like, pasteall) and
>> giving a link here?
>>
>> On Wed, Nov 19, 2014 at 5:19 PM, Josh Cullum <josh.cullum at cfms.org.uk>
>> wrote:
>>
>>> Ahh Sorry all, Will do in future Dan.
>>>
>>> Kind Regards,
>>>
>>> Josh Cullum
>>>
>>> On 19 November 2014 16:18, Dan McGrath <danmcgrath.ca at gmail.com> wrote:
>>>
>>>> Hi Josh,
>>>>
>>>> Oops, it looks like you sent a very large email to the list. For future
>>>> reference, please use pasteall.org to post large blender files and use
>>>> a link in the email instead of attaching it to the email itself.
>>>>
>>>> And apologies to the rest of the list for such a large email!
>>>>
>>>>
>>>> Dan
>>>>
>>>> On Wed, Nov 19, 2014 at 10:43 AM, Josh Cullum <josh.cullum at cfms.org.uk>
>>>> wrote:
>>>>
>>>>> Hi There,
>>>>>
>>>>> I have an issue with a particular scene that has been created on
>>>>> another PC (Windows 7, Nvidia GTX660) and sent to me to render using a HPC
>>>>> GPU node ( RHEL 6.4, 2x Nvidia Tesla M2070). We are seeing an error within
>>>>> blender when trying to render this specific scene on the HPC GPU node using
>>>>> Cycles render, *CUDA error: Invalid value in cuTexRefSetAddress(NULL,
>>>>> texref, cuda_device_ptr(mem.device_poi**nter), size)* which bombs out
>>>>> the render as soon as the BVH mesh has been calculated. The settings are
>>>>> 1000 samples @ 2160x3064 with 256x256 tiles, the .blend file is attached
>>>>> (zipped)
>>>>>
>>>>> However, we seem to be able to render in 4K @ 1000 samples with
>>>>> 256x256 tiles, Mike Pans BMW test case without any issues, as well as the
>>>>> latest test case, the cycles_bench_272.blend using the GPU's and having no
>>>>> issues at all.
>>>>>
>>>>> The version we are running on the GPU machine is Blender 2.72b, with
>>>>> the latest Nvidia drivers 340.58.
>>>>>
>>>>> Kind Regards,
>>>>>
>>>>> Josh Cullum
>>>>>
>>>>> --
>>>>> *Josh Cullum* // IT Systems Administrator
>>>>>
>>>>> *e: josh.cullum at cfms.org.uk <josh.cullum at cfms.org.uk>* // *t: *0117
>>>>> 906 1106 // *w: *www.cfms.org.uk <http://www.cfms.org.uk%22/> // [image:
>>>>> Linkedin grey icon scaled]
>>>>> <https://www.linkedin.com/pub/josh-cullum/35/b90/8b9>
>>>>> CFMS Services Ltd // Bristol & Bath Science Park // Dirac Crescent // Emersons
>>>>> Green // Bristol // BS16 7FR
>>>>>
>>>>> [image: 4.2 CFMS_Artwork_RGB] <http://www.cfms.org.uk/>
>>>>>
>>>>> _______________________________________________
>>>>> 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
>>>>
>>>>
>>>
>>>
>>> --
>>> *Josh Cullum* // IT Systems Administrator
>>>
>>> *e: josh.cullum at cfms.org.uk <josh.cullum at cfms.org.uk>* // *t: *0117 906
>>> 1106 // *w: *www.cfms.org.uk <http://www.cfms.org.uk%22/> // [image:
>>> Linkedin grey icon scaled]
>>> <https://www.linkedin.com/pub/josh-cullum/35/b90/8b9>
>>> CFMS Services Ltd // Bristol & Bath Science Park // Dirac Crescent // Emersons
>>> Green // Bristol // BS16 7FR
>>>
>>> [image: 4.2 CFMS_Artwork_RGB] <http://www.cfms.org.uk/>
>>>
>>> _______________________________________________
>>> Bf-cycles mailing list
>>> Bf-cycles at blender.org
>>> http://lists.blender.org/mailman/listinfo/bf-cycles
>>>
>>>
>>
>>
>> --
>> With best regards, Sergey Sharybin
>>
>> _______________________________________________
>> Bf-cycles mailing list
>> Bf-cycles at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-cycles
>>
>>
>
>
> --
> *Josh Cullum* // IT Systems Administrator
>
> *e: josh.cullum at cfms.org.uk <josh.cullum at cfms.org.uk>* // *t: *0117 906
> 1106 // *w: *www.cfms.org.uk <http://www.cfms.org.uk%22/> // [image:
> Linkedin grey icon scaled]
> <https://www.linkedin.com/pub/josh-cullum/35/b90/8b9>
> CFMS Services Ltd // Bristol & Bath Science Park // Dirac Crescent // Emersons
> Green // Bristol // BS16 7FR
>
> [image: 4.2 CFMS_Artwork_RGB] <http://www.cfms.org.uk/>
>
> _______________________________________________
> 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/20141119/300748b8/attachment.htm 


More information about the Bf-cycles mailing list