s-gupta / rcnn-depth

Learning Rich Features from RGB-D Images for Object Detection and Segmentation
BSD 2-Clause "Simplified" License
365 stars 150 forks source link

adapting jobParallel #31

Closed hojat-kaveh closed 7 years ago

hojat-kaveh commented 7 years ago

Hi I'm new to Caffe and deep learning. As stated here I'd like to use all the cores on my machine to achieve faster performance. I'm running the code on HP Envy 15t-j100 (4 real cores, 8 virtual cores). I want to know which part can be done in parallel? and also I want to know how I should go about this multi-threading?! Can you provide me with some resources?

hojat-kaveh commented 7 years ago

ok, I get a similar output like issue #15, and I don't know how to dapt jobParallel to fit my compute environment as @s-gupta has suggested. I tried editing script_edges.m in "Testing code" section like this: jobParam = struct('numThreads', 2, 'codeDir', pwd(), 'preamble', '', 'matlabpoolN', 1, 'globalVars', {{}}, 'fHandle', @empty, 'numOutputs', 1); resourceParam = struct('mem', 3, 'hh', 1, 'numJobs', 50, 'ppn', 2, 'nodes', 3, 'logDir', '/home/hojat/Documents/application/rcnn-depth/results/log/pbsBatchDir/', 'queue', 'psi', 'notif', false, 'username', 'hojat', 'headNode', 'psi');

is it incorrect? and help is greatly appreciated.

hojat-kaveh commented 7 years ago

jobParallel sends the parameters to simplePBS which should run shell script using system function. so the question is how can I make sure the shell script runs correctly?

hojat-kaveh commented 7 years ago

Since this is a duplicate issue of issue #15, I'm closing this issue.