Psychopy timing on M1/M2 Macs

There has been a significant amount of recent discussion around psychophysical timing on M1/M2 Macs, due to issues around the desktop compositor interfering with scheduled frame display.

Does the PsychoPy community have any recent updates on timing on M1/M2 Macs? It seems to be the opinion of the Psychtoolbox developers that these issues would interfere with any time-sensitive frame display operations, not just the methods employed by Psychtoolbox.

I am aware of the articles Measuring Software Timing Errors in the Presentation of Visual Stimuli in Cognitive Neuroscience Experiments (Garaizar et al) and The timing mega-study: comparing a range of experiment generators, both lab-based and online (Bridges et al).

Thank you for any additional information.

We haven’t run the timing megastudy tests on M1/M2 chip devices yet to see what the comparison is.

For online experiments, where the timing is not as precise as lab-based studies anyway, I don’t think the Apple silicon will be your main issue. In browser the JS scripts/libs don’t get access to the low-level info that the Python/Matlab scripts can access so it’s more guesswork about when a screen refresh has occurred. Mario at PTB would surely just be of the opinion that nothing can achieve reasonable timing in the browser so don’t bother. In my ownn opinion, it depends on your definition of reasonable timing. For a lot of people having roughly the precision of a frame (e.g. to request a stimulus for 3 frames, which is perfectly possible) is as good as they need and in that case I think M1 would also work.

If this matters to you though, get a photometer and check the timing on your system

For desktop performance: We have seen consistent timing on M1/M2 Macs working with EEG and PsychoPy3. Verified with photodiodes and other microswitches.

1 Like