Switching to 9070 XT felt like it had more input lag over NVIDIA. So I investigated the latency difference.
I switched from 5060 Ti 16GB to 9070 XT a few days ago, and when I hit up Overwatch 2, the AMD card felt sluggish compared to my RTX. I posted about it in this subreddit too: https://www.reddit.com/r/AMDHelp/s/3xyBI5GP15
So I finally decided to test the latency comparing 9070 XT vs 5060 Ti, matching as much as possible the same FPS, GPU, and CPU load.
Tested through 240fps slow mo on iphone recording mouse and monitor at the same time. I flick the mouse with finger and count the frames between first frame of mouse movement and first frame of monitor movement. I also used freesync and g-sync for both. My monitor is 160hz.
9070 XT:
90fps/hz (100% gpu util) - 10.16 frames/42.3ms
120fps/hz (100% gpu util) - 8.25 frames/34.4ms
153fps/hz (60% gpu util) - 5.3 frames/22.1ms
5060 Ti 16GB:
90fps/hz (100% gpu util) - 10.22 frames/42.6ms
120fps/hz (100% gpu util) - 7.77 frames/32.4ms
153fps/hz (60% gpu util) - 5.3 frames/22.1ms
I asked ChatGPT for a standard deviation (I don’t really know what it’s called) and it said 90 fps has a 5.6ms uncertainty, 120 fps has a 4.2ms uncertainty, and 153 fps has a 3.3ms uncertainty.
I also only took 10 data points each game scenario.
But all in all, I think I was just used to playing with Reflex + Boost, and when I started using the 9070 XT, it felt sluggish.
I tested Reflex also, which is basically 15-20% faster (60% gpu util only tho). But thats with an uncertainty of 3.3ms uncertainty.
TLDR; As far as I can test with my phone slow-mo, there are no differences in latency between AMD and NVIDIA without Anti-Lag or Reflex.