JinghaoLu
commented
4 years ago It seems the original batch calculation results in the batch size that is on the boundary of the memory in your machine, because sometimes you can and sometimes you cannot pass the movement correction session. A simple but temporary solution is that you can go to "batch_compute.m" and replace in line 3 or 7 (based on which system you are using) with a higher denominator and see if it get the stuck points passed. I will need to see if there is a systematic fix for this, or this is something specific in your machine.
gviejo
Les0me
plodocus
marcoistasy
Mtkostecki
Hello,
Thanks for the effort in writing that package. I am testing it with 20 GB of data recorded with the UCLA miniscope. I keep stumbling onto memory errors that do not appear if I give only a few GB of data :
or
I tested it on a linux machine with matlab R2019b, 32GB memory and a window machine with matlab R2018b, 16 GB memory. Both have more than 100 GB free space on the disk. The linux machine has 31 GB of SWAP memory.
So, is there a way to adjust a batch size parameter somewhere instead of playing with the SWAP memory?
Apologies if the question has been asked before or if it's just trivial, I am not very familiar with matlab.
Thanks for your help
Guillaume