Nvidia’s new ultra-low latency mode is the big new feature, and it’s designed for competitive gamers in mind. It’s a setting that can reduce latency by up to 33 percent, and uses a “just in time” technique that submits “frames to be rendered just before the GPU needs them.” Nvidia used to have a maximum pre-rendered frames option in its control panel, but Eurogamer explains that this new latency mode alters the pacing of frames so they’re generated just before they’re needed.
This new feature will be more noticeable on games that are GPU bound and running between 60fps and 100fps, according to Nvidia. The Ultra-low latency mode will work on DirectX 9 and DirectX 11 titles, but DirectX 12 and Vulkan games decide when to queue the frame.
Nvidia is also introducing a new Sharpen Freestyle filter, which seems to be a response to AMD’s Radeon Image Sharpening feature. Nvidia’s filter is designed to improve the image quality of the previous “detail” filter. “The performance impact is roughly half that of the prior filter,” claims Nvidia. You can customize the level of sharpness, and it supports more than 600 games that use DirectX 9, DirectX 11, DirectX 12, and Vulkan APIs.
This new driver have been pulled for now due to bug.
"NVIDIA has found a bug in our recent 436.02 driver posting, causing it to install GeForce Experience even if the user selects not to install it.
We are pausing the driver download from the NVIDIA website while we fix the issue. Users attempting to download the driver from the NVIDIA website will receive a “404 – Not Found” message when attempting to download.
If you have installed the driver and wish to uninstall GeForce Experience, you can do so from the ‘Window System Settings: Add or Remove programs’.
We apologize for the error and hope to have the fixed driver re-posted soon."
Meanwhile, perhaps the oddest part of all of this isn’t the first time that NVIDIA has offered Ultra mode. Until earlier this decade, NVIDIA’s drivers also supported a queue size of 0, which is why I’m not sure this entirely counts as a new feature. However given the tricky nature of queuing and the evolution of OSes, it’s also entirely possible that NVIDIA has implemented a newer algorithm for pacing frame submission.
I was speaking in past tense - there was no queue size of 0. I was dabbling around with the Nvidia drivers / Inspector and other tools quite extensively. Now, it goes with a queue size of 0. Before this patch, it didn’t.