kragen@inexorable:~$ python
    Python 2.6.6 (r266:84292, Sep 15 2010, 15:52:39) 
    [GCC 4.4.5] on linux2
    Type "help", "copyright", "credits" or "license" for more information.
    >>> import os, commands
    >>> print commands.getoutput('ps u %s' % os.getpid())
    USER       PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
    kragen   20823  0.3  0.1   9636  3780 pts/4    S+   14:00   0:00 python
    >>> x = [{} for ii in range(1000*1000)]
    >>> print commands.getoutput('ps u %s' % os.getpid())
    USER       PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
    kragen   20823 10.8  7.7 168176 159756 pts/4   S+   14:00   0:02 python
    >>> (159756-3780)*1024/(1000.0*1000)
    159.719424

So the baseline is about 160 bytes per dict. Objects, by default, are built on dicts. Objects with __slots__ can be a little more efficient.

You can look at the memory usage here: http://shootout.alioth.debian.org/u64/benchmark.php?test=all...