memory benchmarks running/reporting
-----------------------------------------

``runbench.py`` runs all available and stores resulting information in a log file. 

``report.py`` takes a logfile and produces reporting information. 

In order to run benchmarks with multiple python interpreters::

    python runbench.py -e python2.5,pypy-c-opt3 

    
Benchmark types / files 
-------------------------

sizes.py contains explicitely checkpointed benchmarks used to measure sizes of several object types and constellations. 
    all 'bench_' functions take an NUMITER parameter which indicates how often an object
    structure should be put in a list.  This parameter can be set from ``runbench.py`` command line. 

XXX In order to run interpreter size benchmarks::

    XXX python runbench.py -e python2.5,pypy-c-opt3 --interpsize 

XXX todo:
app*.py contains time-checkpointed benchmarks used to measure memory usage of an application scenario. 
    --runwithparams=500,1000,1500 will run application benchmarks with with several paramat