coinstac-pyprofiler 0.1.0

A wrapper for some existing python profilers

coinstac python profiler

Profile your python code (includes code running using coinstac-simulator). This primarily uses pyinstrument profiler, but can also be extended to include other python profilers

Prerequisites

Python 3.6+
Other packages listed in Requirements.txt

Usage

NOTE: The code currently implements profiling based on pyinstrument profiler. Other profilier (such as cprofile) can be included based on requirements.

First way:

Use the following line above the method definition which needs to be profiled.

from coinstac_pyprofiler import custom_profiler as cprof
@cprof.profile(type="pyinstrument", output_file_prefix=output_file_prefix)

Decorator class to profile any method. Note: 'output_file_prefix' should include its (absolute) directory path

Second way:

Create object of Profile class in custom_profiler.py and use start() and stop() methods to control profiling.

from coinstac_pyprofiler import custom_profiler as cprof
@cprof.profile(type="pyinstrument", output_file_prefix=output_file_prefix)
profile = cprof.Profile(type='pyinstrument',  output_file_prefix=<your_dir_path>/<some_file_prefix>")
profile.start()
<your code to profile>
profile.stop()
profile.persist_log()

Merging multiple profile output files

Merges the json profiler output files generated using pyinstrument profiling and saves merged output.

Use-case:

For a computation in coinstac-simulator, some computation has many iterations and every iteration of python call generates a separate profile json output file. All such json files can be merged separately for each client/remote using this call.

Example

An example usage is included in tests/examples/profiler_usage.py which demonstrates the usage of the above mentioned profiling methods and also provides examples to merge multiple profile output files.

Happy profiling!!