search

jessieren / DeepVirFinder

Identifying viruses from metagenomic data by deep learning
Other
116 stars 32 forks source link

theano issue putatively related to the slowdown of the program #8

Open ilnamkang opened 4 years ago

ilnamkang commented 4 years ago

Hi,

When I run DeepVirFinder using multiIcores (e.g. 72 cores), I'm encountering warning messages like below. DeepVirFinder seems to be getting slower with these messages and use only <1 core in average.

Warning messages:
INFO (theano.gof.compilelock): Waiting for existing lock by process '52117' (I am process '52060') INFO (theano.gof.compilelock): To manually release the lock, delete /home/user/.theano/compiledir_Linux-5.3--generic-x86_64-with-Ubuntu-18.04-bionic-x86_64-3.6.9-64/lock_dir INFO (theano.gof.compilelock): Waiting for existing lock by process '52117' (I am process '52054') INFO (theano.gof.compilelock): To manually release the lock, delete /home/user/.theano/compiledir_Linux-5.3--generic-x86_64-with-Ubuntu-18.04-bionic-x86_64-3.6.9-64/lock_dir (continues maybe endlessly)

I'm running DeepVirFinder on Ubuntu 18.04. And, the messages don't disappear after deleting .theano directory.

Thanks.

soilmicrobiome commented 4 years ago

I also have the same messages when running the program;

  1. Encoding and Predicting Sequences. processing line 1 processing line 201 INFO (theano.gof.compilelock): Waiting for existing lock by process '2984' (I am process '2983') INFO (theano.gof.compilelock): To manually release the lock, delete /home/rachel/.theano/compiledir_Linux-4.4--generic-x86_64-with-debian-jessie-sid-x86_64-3.6.10-64/lock_dir processing line 401 INFO (theano.gof.compilelock): Waiting for existing lock by process '3735' (I am process '3734') INFO (theano.gof.compilelock): To manually release the lock, delete /home/rachel/.theano/compiledir_Linux-4.4--generic-x86_64-with-debian-jessie-sid-x86_64-3.6.10-64/lock_dir processing line 601

did you figure out the issue? Or if its a problem

ilnamkang commented 4 years ago

I didn't find a solution for this issue.

For my small data (a few hundred metagenomic contigs), turning off "-c" flag and using only 1 core (default value) gave me the best speed.