Closed sthennin closed 8 years ago
@sthennin : What's the update on this one? Should this be looked on?
Is avoided by not writing out too many parameters into the RESULTS window, but rather updating the parameters via a Python script. Upgrade to an SQL database would eliminate this problem.
2dx_merge on OS X can grow to >10 GB in RAM volume, and also is extremely slow in reaction.
When running scripts that produce results in the LOGS/${scriptname}.results file, then 2dx_merge grows in size. This can be tested in the following way: Modify the "Custom Script 1" so that it outputs for every of the 500 images some 100 times the same line echo "set dummy = 1" >> LOGS/${scriptname}.results If you then launch 2dx_merge on this script, 2dx_merge explores in RAM memory. It is as if for every copy of a Parameter that it wants to save, it is reading an entire copy of the configuration file and then forgetting it in memory.
Other related issues: I operate 2dx_merge on a dataset with 550 images. Loading 2dx_merge takes about 15 minutes, even though parsing through 550 2dx_image.cfg files should only take one minute.
2dx_merge writes its own local 2dx_merge.cfg file many times per second. Is this needed?