[SciPy-user] shared memory machines

Gael Varoquaux gael.varoquaux@normalesup....
Wed Feb 11 05:46:20 CST 2009


I shouldn't be working on that, but this is way more fun :).

So I found a few more simple errors in the code and fixed them (code
attached). The garbage collector thread lock multiprocessing. I am not
sure why. I disabled it just to see what would happen. I added a few
print statement to try and debug eventual memory leaks. 

I think I have a memory leak, judging from the different prints. I am not
sure though, and I wonder if there is a good way of checking this, other
than running the test code in a big loop and checking if the test box
eventually dies. Valgrind does report some 'possibly lost' blocks that
increase with the size of the array. Anybody has a suggestion on how to
debug this?

Cheers,

Gaël
-------------- next part --------------
A non-text attachment was scrubbed...
Name: sharedmem.zip
Type: application/x-zip-compressed
Size: 11588 bytes
Desc: not available
Url : http://projects.scipy.org/pipermail/scipy-user/attachments/20090211/067d77d5/attachment-0001.bin 


More information about the SciPy-user mailing list