OTOY Forums

So I am just wondering why this happens. When I render with my 3 nodes (9 more cards) it breaks the render up into four parts then. Then I am assuming the top left corner is the host computer and the other 3 are the nodes. But the nodes take forever while the top left quadrant goes clear. It is very annoying. Any way to fix this so they all work on the image toghether? Or somehow speed it up. I just did a render of a room for 38 minutes and the top left quadrant was perfectly clean and the other 3 were not anywhere near acceptable.

Thanks a lot

LFedit wrote:So I am just wondering why this happens. When I render with my 3 nodes (9 more cards) it breaks the render up into four parts then. Then I am assuming the top left corner is the host computer and the other 3 are the nodes. But the nodes take forever while the top left quadrant goes clear. It is very annoying. Any way to fix this so they all work on the image toghether? Or somehow speed it up. I just did a render of a room for 38 minutes and the top left quadrant was perfectly clean and the other 3 were not anywhere near acceptable.

Thanks a lot

Please read the chapters "Moved film buffers to the host and tiled rendering" and "Overhauled work distribution in network rendering" of the v3 new features thread. There I tried to explain the changes in the kernels, film buffer and network rendering.

TL;DR: Disable the kernel option "minimize net traffic" to avoid that the net render slaves have to finish a tile before they continue rendering the next tile. Please be aware that this will increase the amount of data that is being sent over the network and the slaves may end up with a backlog of untransmitted results.

Thanks for the explanation. I will explore those options to see what works best for our situation.

Thanks