symerio / neurtu Goto Github PK

psutil includes a number of interesting metrics for any given process that could be worth supporting.

These include, for instance,

• io_counters
• memory_info etc..

Related to #18, some metrics in psutil are not cumulative (e.g. the number of open file descriptors), meaning that to evaluate some code with it, one needs to measure the quantity in question in parallel with some frequency.

This is, for instance, the case for memory usage and a similar approach is taken in memory_profiler to estimate the peak memory usage.

Currently if pandas is used, version 0.20+ is required because of this section,

which fail for earlier versions if I remember correctly.

There should be a way to achieve the same result with earlier versions of pandas though.

To test it, one should decrease the tested pandas version in Travis CI.

There have been a few updates to the code base, and it would be good to upload a new release on PyPi.

The timeit.default_timer() measures wall time, it would be worth adding another timer for CPU time, as IPython does,

-c: use time.clock to measure the time, which is the default on Windows and measures wall time. On Unix, resource.getrusage is used instead and returns the CPU user time.