NVIDIA is making a tile-based multi-GPU rendering technology

Posted on Thursday, November 21 2019 @ 12:47 CET by Thomas De Maesschalck
NVDA
Multi-GPU systems never really caught on, not just because of the high cost but also because of the many issues with software support and performance scaling. In recent years, NVIDIA and AMD both abandoned multi-GPU rendering but now there's word that NVIDIA is working on a new tile-based technique.

It's not exactly clear how this differs from what the company previously used, but TechPowerUp speculates NVIDIA may have altered the tile-based rendering method that it's using for single-GPU rendering since its Maxwell GPU family. The latest NVIDIA drivers contain references to "CFR", which could be short for "checkerboard frame rendering" or "checkered frame rendering." The method is present in the drivers but there's no developer documentation yet.
In CFR, the frame is divided into tiny square tiles, like a checkerboard. Odd-numbered tiles are rendered by one GPU, and even-numbered ones by the other. Unlike AFR (alternate frame rendering), in which each GPU's dedicated memory has a copy of all of the resources needed to render the frame, methods like CFR and SFR (split frame rendering) optimize resource allocation. CFR also purportedly offers lesser micro-stutter than AFR.
All we know about CFR so far is that it's a DirectX exclusive and that it will only work on Turing because you need NVLink support.

Tile based rendering

Via: TechPowerUp


About the Author

Thomas De Maesschalck

Thomas has been messing with computer since early childhood and firmly believes the Internet is the best thing since sliced bread. Enjoys playing with new tech, is fascinated by science, and passionate about financial markets. When not behind a computer, he can be found with running shoes on or lifting heavy weights in the weight room.



Loading Comments