GPU rendering, please

[schultzeworks]

I did a search of the forum and did not see any posts on this topic, which is hard to believe.

Anyways, I am noticing that new CUDA video cards from Nvidia have 500+ GPU processors, which would be a fantastically great boost for rendering, especially now that we have a realtime capabilites.

Please advise / scold me that this has been already posted / inform me when I can get it!

[v-cube]

yes I would be interested too

are there any plans for gpu support for RT ?

best regards

Andreas Walther

[wcmansp]

Related, in concept, is distributed rendering. Going forward the question is always how to make things faster, more real-time; one aspect is how to incorporate more compute power into the work. In both the GPU and CPU case this looks to be about how to bring more processors into the job. On the CPU side, the trend is being more cloud-like, supporting a model where the user devices (mobile, tablet) are light and the cloud is "heavy". It's possible this might end up with high-count multiple-CPU systems at the center but this doesn't seem to be the way these clouds are developing now -- they're designed more like extensible collections of medium weight machines, a model where adding processing power involves larger number of machines vs. more powerful machines. So this is one consideration -- as computing moves to a cloud model how can things be made to go faster? As one example, what is involved in uncapping the current limit on the number of distributed renderers? Is it just based on a Microsoft connection limit for workstations or is there V-Ray architecture that would need to change? GPU support is another way to add more processors to a job, just a different kind. The advantage is there have been tremendous increases in the power available in these solutions but these are, currently, still standalone offerings -- what do you need to prepare for in order to add more processing power if you pursue this path?