klabhub / neurostim

Design and run visual neuroscience experiments using Matlab and the Psychophysics Toolbox.
MIT License
5 stars 4 forks source link

Unusual frame-drops #38

Closed adammorrissirrommada closed 7 years ago

adammorrissirrommada commented 8 years ago

A very specific and confusing issue with frame-drops:

If:

the Screen('Flip') line doesn't flip at the _first _available retrace. It waits one. So I lose every second frame. But worse, it doesn't happen every time I run my experiment. Maybe 2/3 of the time. The remaining times everything works fine. And when it works, it continues to work across trials indefinitely, but can be broken if I go into Eyelink setup mode.

It doesn't happen if I reduce the subject window so that it is not full screen, even if by only one pixel (1919 x 1199), or if I am using full-screen but NOT Eyelink.

The problem is NOT that the frame-loop code takes too long. It gets around to flip with plenty of time to spare - it's within the Screen('Flip') command that the delay happens - and the problem persists even with a much simpler paradigm that gets around the loop in 3ms or so (Display++ runs at 120Hz).

I haven't been able to try another display that is of comparable resolution and frame-rate, so I don't know if it is specific to the Display++.

bartkrekelberg commented 7 years ago

Is this still an issue @adammorrissirrommada ?

adammorrissirrommada commented 7 years ago

I'll test it tomorrow.

adammorrissirrommada commented 7 years ago

This appears to be solved.