meanwhile the core still has 1 additional frame of lag compared to standalone…
I moved this post because the other thread has nothing to do with MAME cores.
As for latency, AFAIK no one has actually verified this with a high-speed camera, only with the frame-advance method, which isn’t always reliable for comparing across different programs. Also, IIRC, @Tatsuya79 said the additional frame isn’t there if you use MAME-libretro’s internal pause+frame-advance instead of RetroArch’s.
Measuring absolute lag requires a high FPS camera. However, comparing input lag between two applications does not, so you can try and verify this yourself. Run MAME stand-alone next to the core. Then record your desktop to a video while both MAME and the core run the same ROM. Your controller should be able to control both at the same time.
Go through the recorded video frame by frame and see if the core reacts 1 frame later than stand-alone. I did for bsnes here:
Yikes, I guess I wasn’t up to speed on the latest. I regret my comment and my apologies to the maintainers
no biggie, don’t worry about it