psychopy.org | Reference | Downloads | Github

Recommended hardware


#1

Hi,

I built an experiment on Mac OS 10.14.1 with PsychoPy 1.85.4 and it runs well.

When I try to run it on the research assistant’s PC, it did work until I added audio files (.aiff). Now the experiment fails to start. PsychoPy only displays the background color of the image of the first routine.

Could it be because of the hardware? For example, I have 16 GB of RAM and he has 8 GB.

image%20(1) image%20(1)|581x220

Many thanks,
Geneviève


#2

Hi, can you specify that a little more? What exactly happens? Are there any error messages that you could post here? For example, it could be that the path of your audio files cannot be found on your assistant’s PC. If you use absolute paths, you could avoid that by using relative paths.
Most likely, it’s not at all related to your hardware. For example, his RAM is definitely more than enough.
In general, I would recommend to use a newer version (e.g., 1.90.3, or even the current beta of PsychoPy3) because many bugs have been fixed since 1.85.4. For example, there used to be an issue with audio loops that weren’t played completely, which would also explain your problem.


#3

Hi,

There is no error message. The screen freezes, the esc button doesn’t work and the only way out is to press ctrl+alt+delete to end the task.

How can I check if I use absolute or relative path? We also have images in the experiment and that is not a problem when we transfer the experiment folder from one computer to another.

Thanks a lot,
Geneviève


#4

Relative paths indicate the location of your file relative to the PsychoPy file (e.g., “/sounds/sound.aiff”). Absolute paths indicate the full path to the file on your computer (I don’t really know MacOS paths, but for example something like “/Users/YourName/Experiment/sounds/sound.aiff”), which would differ on your assistant’s PC. However, like you say, if the images work, you probably used relative paths for the sounds, too. You could check them in the conditions file which you’ve specified in the trial loop in PsychoPy.

Since there is no Python/PsychoPy error message, I’m afraid I can’t help you further because I’m not familiar with MacOS.

Does a very simple PsychoPy experiment (e.g., just playing one sound file) work on your assistant’s PC? If not, maybe something is wrong with his audio settings.


#5

I would also suggest not running your experiment full-screen while you are trying to sort things out like this. It can often be the case that the experiment has not actually crashed, but, for example, there could be some custom code that is just interfering with responding to keypresses. With the ability to click on other windows, you can check for messages and other feedback that would otherwise be hidden by your main window.


#6

Dear Micheal, Dear Torge,

Thanks for your help. Disabling the full screen mode solved our problem. It is not ideal, but the task runs smoothy this way.

Best,
Geneviève


#7

Wouldn’t recommend doing that for actually running your experiment (as performance suffers when PsychoPy doesn’t have full control of the screen), just for helping to diagnose what the actual issue is.


#8

Hi Michael,

This is a last resort option as we tried every fix we could think of.

Thank you for warning us about potential unintended consequences. Do you have any examples of difficulties we could experience and ways to potentially lower the risks? Would closing all other programs and turning off the wifi help?

Best,
Geneviève


#9

But running in a window rather than fullscreen wouldn’t itself fix any problem: it just allows you to click on to other windows to see log messages and so on, and not get hindered by PsychoPy not responding to a keypress. Can you describe exactly what you can do now that you can’t with the full screen window? i.e. we should be aiming to solve the actual problem rather than use a dodgy work-around.

You should reduce all competing processes whenever psychoPy is running an experiment, regardless of whether it is full screen or not. But when it is not running full screen, other processes have more ability to interrupt. This can lead to worse timing performance.


#10

Dear Micheal,

Thanks for helping me understand more deeply what might be the issue.

The experiment runs well in full-screen mode on two computers (MacOS and Windows), but when we try it on my RA’s laptop (Windows), the experiment starts, i.e. we can enter the participant number and session in the dialogue box, but it freezes on a grey screen. PsychoPy does not respond to keypress and we have to use Windows tasks manager to quit the program. However, when disabling full screen mode, it runs smoothly.

The problem started when I added audio files to the experiment. That’s the only clue we have so far as to why the bug appeared. The audio files are in .aiff format.

Best,
Geneviève


#11

OK, thanks, that makes it clear. I have to say I’m quite mystified how there would be an interaction between full-screen mode and use of sound that would make the experiment unresponsive. Might need to get the input of @jon or @dvbridges here…


#12

I’m also mystified as to why the audio files would make any difference, but maybe you could try a different audio backend (in the preferences set ‘pyo’ to be the first option?

When an experiment “freezes” in fullscreen mode it’s sometimes because another window has grabbed focus behind the scenes (e.g. the experimenter has clicked on something on a different monitor) so the experiment is no longer receiving the mouse/key events to carry on.

Beyond that I have no idea. Maybe you could share the experiment (preferably a minimal working example ) with us so we can have a look.