On 09/26/2014 01:31 AM, Joseph Martinot-Lagarde wrote:
I don't know exactly why, but here are some hints:
- A 7000*7000 array of float64 takes 7000 * 7000 * 8 / 1024 / 1024 = 373
Mo of memory (headers are negligible)
- Each line where the arrays are created creates an additional temporary
array (because of the addition), so in spyder the problem arises at 373
* 3 = 1119 Mo of memory. This has to be added to spyder's own memory
which I guess is higher than IDLE. Does the variable explorer make a
copy of the data, since it pickles then unpickles the arrays ? That
might be considered a bug.
- In IDLE, the 5 arrays + 1 temporary represent 2238Mo. I don't
understand why it didn't break. Is is certain that it is from the same
python install ?
That is what I am trying to get a handle on. If it is just a straight
memory access problem due to memory restrictions in 32 bit Windows then
the second case should fail if the first does. I can understand moving
up to 64bit, but the question is whether that is really a solution or a
way of hiding a problem? I do not have a Windows installation to play
with, so I am not able to investigate on this end, which is why I have
been asking the questions.
According to the OP there is only one version of Python installed:
" ...I only have Python 2.7.6 installed on my computer. Thus it is
definitely the same Python version and numpy module that I use in IDLE
and Spyder. "
Le jeudi 25 septembre 2014 19:45:55 UTC+2, Adrian Klaver a écrit :
--
Adrian Klaver
[email protected]
--
You received this message because you are subscribed to the Google Groups
"spyder" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/spyderlib.
For more options, visit https://groups.google.com/d/optout.
