That's a really good question - I am not aware of any rendering applications that utilize SLI, from what I've read it is almost exclusively used by games. However, since we have seen provable increase in gaming performance with say a 'high bandwidth' SLI bridge over a ribbon bridge it would stand to reason that an application coded in such a way as to be able to make use of NVLink's much greater bandwidth would likewise see an increase in performance, no?
Initially I thought the skinny tabs on nvlink were just SLI but the pinout is wrong for it. I would assume SLI has inherent additional bandwidth when run over nvlink. Whether or not that means anything for performance in SLI applications (games) is another question.
Those are strange numbers. GPUZ says 1200 and 1455 on my V.. Thinking about grabbing a Quadro for the extra memory but don't want to lose too much performance. Any idea what the maximum effective clock you are getting is? Some applications only use 1455 on my Titan, others go up to 1845.
Depends on GPU load. I can maintain nomal specified maximum boost even at full load and with a bit less load, I’ve seen it hit 1700-1800MHz.
Me too!!! Not just gaming benchmarks but workstation benchmarks. I really want to see the difference having two cards makes in everything from productive tasks to entertainment. Come on Lutjens ... be a pal.