psychopy.org | Reference | Downloads | Github

Experiment Builder is super slow

MacBook Pro Mid-2014
MacOS BigSur 11.3
PsychoPy version 2020.2.10
Standard Standalone

Any change that I do in Psychopy takes about 3 minutes to be performed. The “busy” icon appears on screen an do not allow me to do anything else on PsychoPy. If I add a new stimuli… 3 min. If I change its properties… 3 min. If save the experiment… 3 min. If run the experiment… 3 min.

I open and closed the program and even restarted the computer but it is always like that. It is very annoying. It is making me lose a lot of time. Has anyone else experience tsimilar problems?

Miguel

1 Like

Hi Miguel,

I have the same issue (with version 2020.2.10 on Windows 10). I do however think I know what causes it in my situation: my experiment is too big/inefficient. I have too many routines in my flow.

My solution for this was to split my experiment in two experiments and link them together. Not ideal, but it works. I also tried to make my experiment more efficient, but I use a lot of different sliders and because there is a problem with the 2020.2.10 version and reading slider labels from excel files, there is a limit to how efficient it can be.

Is your experiment perhaps also very big? Maybe there is a way to make it more efficient/concise?

1 Like

When I’m working on other people’s experiments I often see way too many routines. If you have two routines that are doing almost the same thing you can probably use just one.

Multiples copies of routines is also a pain because a small change to one then has to be made to all the other copies.

@Anouk and @wakecarter Indeed. My experiment is very big. It is divided in three parts and the instructions take too many routines because each slide (routine) involve different components, not always the same ones. I thought about dividing the whole experiment in three files and linking them by means of completion urls but this will result in spending 3 times more credits in Pavlovia. That is not an ideal scenario when we need to collect data from 160 participants.

This certainly a performance issue to reflect about and which can maybe be solved in future versions of PsychoPy. Additionally, even though I have asked participants to use the same browser and the last version of it, 1/3 of participants have reported having problems when running the experiment at different stages (in both windows and macOS). That is odd and something that scapes my abilities to solve. Any suggestions on how to avoid this type of problems for future experiments?

Best,
Miguel

I’m sure I could significantly reduce the number of different routines, especially if you have several with similar contents. I sometimes use loops for instructions and in one case varied the text vs image and end condition for a set of around 10 pages which combined instructions and demo.

I’ve also written a scrollable text demo which means that a single routine can be used for more than one page of text and a study portal so that participant information can sit outside PsychoPy entirely.

Hi Miguel,

1/3 of participants have reported having problems when running the experiment at different stages (in both windows and macOS)

What kind of problems do you mean? There are 10000 things that could go wrong (or right! :slight_smile: ) and if a third of your participants experience problems, this sounds like something that will require a fix (and possibly there already is one). Also: do you know what browser these participants were using?

I would highly recommend applying this to your experiment to reduce the possibility of Chrome-participants opting out because of an unspecified javascript error.

I assume you mean the following method for solving Unspecified JavaScript error when it only occurs on some browsers.

If the experiment works on some browsers but not others try editing the index.html file directly to replace <script type="text/javascript" src="https://cdnjs.cloudflare.com/ajax/libs/pixi.js/5.3.3/pixi.min.js"></script>
with <script type="text/javascript" src="https://cdn.jsdelivr.net/npm/pixi.js-legacy@5.3.3/dist/pixi-legacy.min.js"></script>

1 Like

Exactly! Thank you

@Anouk and @wakecarter ,

Different people have reported different problems. However, one of the recurring issues that people have reported is that the experiment would not start or crash at the very beginning. Half of them reported that the error message below appeared onscreen:

I believe this is the error message you were taking about. I implemented the fix that you advised me to and test if things worked well. They seem to now work well in windows computers with Chrome, but now it seems the experiment does not start in Edge browsers after the resources are downloaded.

So far, Firefox seems to be the best browser to run experiments. I think it would be great if you could advice people using PsychoPy to always tell participants to use the latest versions of Firefox for running the experiment. It would be great if everyone knows that at least one browser is reliable enough for collecting data.

Best,
Miguel

As far as I can see if you can get your experiment working on Chrome and Safari then most participants will be fine.

1 Like