Memory profiling

[Rune Froysa]

I have a fairly large program that uses a number of third-party
modules. After a few hours the program eats several 100M of memory.

Does Python provide any tools that can help in determining where the
memory is used? Ideally, I would like to see a tool that can list all
objects by memory-usage, but any tool that can return the number of
bytes used by an arbitrary object would be of great help. Even
something that could find all allocated dicts&lists and return their
size+name would help.

Regards,
--
Rune Frøysa
University of Oslo, Norway

[Michael Hudson]

Re: Memory profiling

Rune Froysa <rune.froysa@us it.uio.no> writes:
[color=blue]
> Does Python provide any tools that can help in determining where the
> memory is used?[/color]

*Memory*, no. Especially in a debug build, there are tools for
finding out where the objects are, though.
[color=blue]
> Ideally, I would like to see a tool that can list all objects by
> memory-usage, but any tool that can return the number of bytes used
> by an arbitrary object would be of great help. Even something that
> could find all allocated dicts&lists and return their size+name
> would help.[/color]

In a debug build, look at sys.getobjects. In a recent release build,
look at gc.get_objects.

Cheers,
mwh

--
<arigo> something happens, what I'm not exactly sure.
-- PyPy debugging fun