Cuda 4.0

[Dariusz Makowski (Dadal)]

Ah yea, I usuall down alot the samples for GPU. But since its not supporting the blend materials and a lot of others I don't use it at all recently... Hopping for better future ^^

[vlado]

V-Ray RT in general has some overhead because of the DR support and the client/server architecture - everything goes through a TCP connection between two separate processes, even when rendering on a single machine. This slows is down a bit compared to renderers, which keep everything in the same application. So we had to trade some performance for features... Also, the GPU renderer by default traces 16 samples per pixel - this might give it the impression that it updates slower. Reducing this value increases the interactive response quite a bit.

Best regards,
Vlado

[Dariusz Makowski (Dadal)]

Huh interesting... I kinda looked at other (arion etc etc) renders and they seem a tad more quicker when rerendering... but well good to know openCL is close !

[vlado]

Isn't it a lot more responsive(refresh speed) then opencl btw?

No, not really. In some of the last tests, OpenCL was a tad bit faster.

Best regards,
Vlado

[Dariusz Makowski (Dadal)]

Wow... Release it ploxxxxxxxxx, isn't it a lot more responsive(refresh speed) then opencl btw?

[vlado]

We do have a CUDA version of V-Ray RT that we use internally, so if we see that there are significant benefits of going this way, it is certainly something that we would do without too much hesitation.

Best regards,
Vlado

[Dariusz Makowski (Dadal)]

I had a feeling CUDA is gonna kick OpenCL in the ass and it is ! I hope that OpenCL will be able to share memory too at some point...

[Bpositive]

Hi

Just saw this announcement on cgsociety: http://features.cgsociety.org/story.php?story_id=6110


What seems very interesting is the shared memory ability - "Unified Virtual Addressing (UVA) – Provides a single merged-memory address space for the main system memory and the GPU memories, enabling quicker and easier parallel programming."


I know the RT GPU dosn't use Cuda, but perhaps such features would come to opencl once!!?