Its better to use LLM per game not Globally. Its used in games when Nvidia Reflex is not supported. They both help only when GPU Bound. If you play CPU Bound game its better to disable because can cause issues. Doesnt matter if the game has 60, 120, or 240+FPS. GSynch was originally created for low FPS games & gradually expanded. But not Reflex or Ultra. The difference is one is implemented in the game engine & the other in the drivers.imprecise wrote: ↑21 Mar 2023, 23:46In theory, LLM is meant to continue accepting and processing inputs when the GPU is under 100% load and can't keep up with any new processing. Nvidia claimed 23% latency improvement in 2019 when they introduced it and ran a 2060 at 4K in order to 100% load the GPU.
https://blurbusters.com/new-nvidia-grap ... s-latency/
60 fps = 16.67ms
120 fps = 8.33ms
240 fps = 4.17ms
480 fps = 2.08ms
At 120fps, even with LLM helping, 23% improvement is only going to be 1.92 ms. At 240fps, 0.96ms. Games running at over 120fps generally don't need a low latency mode intended for 100% GPU load scenarios, this technology is best when the GPU is running a low framerate high resolution single player game. Even then, it's better to cap at a more stable framerate than use the GPU at 100%.
https://www.nvidia.com/en-us/geforce/fo ... when-gpu-/
To answer your question... you might see a slight improvement in responsiveness if it works correctly on your system.
PS: BTW these are Frametimes. Reflex & LLM Ultra lower Input Lag. Looks like you confused Frametimes with System Latency & Input Lag..