Program freezes during an experiment
Hi, we are trying to program an experiment in Open Sesame and exporting it onto our JATOS server. Currently, we are stuck with a strange problem: when running the experiment by using a JATOS link, it freezes around block 4 (out of 16 blocks of 32 trials each). It stalls, then flashes a black box instead of the display, and then hangs, fully crashed.
The design of this experiment is identical to a previous experiment that we ran successfully on JATOS. Here are similarities and differences between the experiment that worked (Exp 1) and the one that is freezing (Exp 2).
Common to both Exp 1 and Exp 2:
- used “Legacy” as the backend
- used a series of 23 .png files for instructions that ranged in size from 20 – 300 Kb
- left the default “disable collect garbage” clicked
- ran two practice blocks of 16 trials each followed by 16 blocks of 32 trials each
Different in Exp 2 from Exp 1
- Whereas in Exp 1, we drew our stimuli using shapes available in Open Sesame (e.g., rectangles and circles), in Exp 2, we import 12 .png files that range from 1 – 7 kb each and use subsets of them on different trials Exploring the forum, we found what sounds like a similar error described (https://forum.cogsci.nl/discussion/2245/images-appearing-as-black-boxes-after-many-trials ).
We thought perhaps it is a memory issue on our JATOS server. It has 2 Gb. We are in the process of upgrading it to 4 Gb. But we tried running our experiment on the free server (experiments.jatos.org), which we think has 4 Gb, and it crashed there too.
Experiment 2 ran successfully through to completion (multiple times) on Open Sesame itself (i.e., before importing it to JATOS) running on a MacBook Pro with 16 Gb of memory, but failed on a MacBook Pro with 8 Gb of memory.
The stimuli that we are using are not particularly large (1 – 7 Kb each). So it seems surprising that it would be running into a memory limit. But we thought perhaps as we reuse them trial after trial, they are getting reloaded into some buffer that isn’t being cleared. That led us to wonder about the “disable garbage collection” option. But sadly disabling that did not solve the problem.
Finally, we tried a version of our Exp. 2 with the PsychoPy backend, instead of Legacy backend, and the program still crashed.
Currently, we are unsure of what else to try, and would be very grateful if you have any ideas about what the problem might be or what further diagnostic steps we might take. Thank you!