Home Creators Posts Import Register Favorites Logout
haven't archived this post yet. have a subscription? use the importer!

Content

With the release of 3.0.09 Redshift added support for both NVLINK on Volumes (Geometry NVLINK support is in progress.)  and memory VRAM optimizations. This opened the foor for MUCH larger volume data sets to be renderable by Redshift. One of the limitations with volume data is there is no out of core support because it isn't really feasible. This limited volume data to the available VRAM per GPU. With the addition of NVLINK this essentially doubled the available VRAM resources and the new memory optimizations when dealing with volume data also improved the VRAM usage.

So I decided to do some NVLINK testing, so far even with nearly full VRAM across the two pooled GPU's the scene upload time performance hit for pooling VRAM is nearly negligible for how much data is moved across and it doesn't appear to make much if any difference to the render time itself once the data is uploaded at least on the full speed 100gb/s NVLINK found on the 2080ti.

In short, NVLINK is working great and allows for nearly double the amount of volume data to be used where before the scene would not be able to render.

For a more detailed post check out my forum post linked below.
https://www.redshift3d.com/forums/viewthread/28149/

If any Houdini users have 2070 SUPERs or 2080's or 2080 SUPERs + NVLINK bridge want to test out my example scene to see how half-speed 50gb/s NVLINK works please post your log results in the forum post.

Files

Previews only

Comments

No comments found for this post.