spacetelescope / jwst

Python library for science observations from the James Webb Space Telescope
https://jwst-pipeline.readthedocs.io/en/latest/
Other
571 stars 167 forks source link

Improve AmiAnalyze memory usage #8860

Open stscijgbot-jp opened 1 month ago

stscijgbot-jp commented 1 month ago

Issue JP-3776 was created on JIRA by Rachel Cooper:

Input files over a certain size (exposures with many integrations) seem to use too much memory and cause the process to fail with "Killed: 9" error. I haven't identified what part of the code is causing this failure yet but assume it is memory-related as inputs with ~100 integrations work, while those with ~1000 fail. Since nints in theory could be as large as 10000, this needs to be improved.

stscijgbot-jp commented 1 month ago

Comment by David Law on JIRA:

Rachel Cooper Can you provide an example of a file that fails processing?  This is likely another case for which the failure point will depend on the amount of RAM a given user has installed.

stscijgbot-jp commented 1 month ago

Comment by Rachel Cooper on JIRA:

Hi David Law, Yes, I've attached a file that fails both on my laptop and on the witserv machines. I have other files that can be processed on witserv, but not locally, so it definitely seems like a RAM issue. I just had it print the process memory usage while the code is looping over the integrations to do the model fit (here) to get an idea of what it was doing and it got here before crashing; see attached plot.