A
Al Franz
I believe there is a memory leak in cPickle. I am using python2.2. I
have a parallel code which uses array() and indices() from Numeric to
massage data buffers before being sent and received by Pypar. Pypar
subsequently uses cPickle to pickle the data. After many hours of
execution, my code crashes with one of the following error messages
(depending upon the run):
a = zeros(shape, typecode, savespace)
MemoryError: can't allocate memory for array
or:
s = dumps(x, 1)
MemoryError: out of memory
I have since modified my code to use a different data format so cPickle is
no longer used from PyPar and now the code runs fine.
-- Al Franz
Computer Scientist
Lawrence Livermore National Laboratory
have a parallel code which uses array() and indices() from Numeric to
massage data buffers before being sent and received by Pypar. Pypar
subsequently uses cPickle to pickle the data. After many hours of
execution, my code crashes with one of the following error messages
(depending upon the run):
a = zeros(shape, typecode, savespace)
MemoryError: can't allocate memory for array
or:
s = dumps(x, 1)
MemoryError: out of memory
I have since modified my code to use a different data format so cPickle is
no longer used from PyPar and now the code runs fine.
-- Al Franz
Computer Scientist
Lawrence Livermore National Laboratory