OptiX 3.8 final release is out!

NVIDIA is excited to announce the final release of OptiX 3.8, with support for Quadro VCA. With very minimal changes to your OptiX-based application you will be able to render on a VCA-powered cloud-based cluster. Each VCA contains 8 NVIDIA Maxwell-based high-end GPUs. Read about the VCA here: http://www.nvidia.com/object/visual-computing-appliance.html . You can see Dave’s GTC 2015 talk with a live demo of OptiX on the VCA at http://bit.ly/1BJ7TsM .

The new progressive API allows applications that render a final image out of many subframes to launch them all in a single API call, thus encapsulating even more of the rendering algorithm within OptiX. Progressive updates are streamed to your standard output buffer as the frame resolves. Progressive mode enables performant communication with the VCA as well as simplifying rendering on local GPUs.

OptiX 3.8 also includes CUDA 7.0 support and two new samples: Instancing and SphereTessellate. We have also cleaned up the samples and set nearly all of them to use Trbvh by default, as it’s our fastest and highest quality acceleration structure builder in most cases. We’ve fixed several bugs throughout OptiX, including several stability bugs in Trbvh and OptiX Prime, so it’s quite likely to be the best builder for your application. See the full release notes for a full list of enhancements and fixes.

Registered developers can access OptiX 3.8 on the ftp site using your registered developer account. OptiX Commercial customers can access this release from the Google Drive corporate site. If you have forgotten the username or password please email OptiX-Help@nvidia.com . And please remember our customer forum at https://devtalk.nvidia.com/default/board/90/ for most of your questions.

David McAllister
OptiX Manager | NVIDIA Corporation

FYI, for anyone who downloaded the Linux build of the final release already, it was actually a copy of the beta. I just corrected it, so you should download it again to get the actual final file. The correct file has an md5 hash of ecf63cf32b87defa36d3cd283d085314.