module `benchmark.profiler_class`¶

Short summary¶

module pymlbenchmark.benchmark.profiler_class

Implements a benchmark about performance.

Classes¶

class	truncated documentation
`ProfilerCall`	Runs a profiler on a specific call. It can use either pyinstrument, either cProfile. The first module …

Methods¶

method	truncated documentation
`__init__`
`__iter__`	Iterates on stored profiled. Returns a couple `(profile, configuration)`.
`__len__`	Returns the number of stored profiles.
`match`	Tells if the profiler should be run on this set of parameters.
`profile`	Profiles function fct, calls it repeat times.
`to_txt`	Saves all profiles into one file.

Documentation¶

Implements a benchmark about performance.

source on GitHub

class pymlbenchmark.benchmark.profiler_class.ProfilerCall(fct_match=None, name='ProfilerCall', repeat=1000, interval=0.0001, module='pyinstrument')¶

Bases: object

Runs a profiler on a specific call. It can use either pyinstrument, either cProfile. The first module takes a snapshot every interval and stores the call stack. That explains why not all called functions appear in the summary but the overhead is smaller.

source on GitHub

Parameters:

fct_match – function which tells if the profiler should be run on a set of parameters, signature is match(**kwargs) -> boolean
repeat – number of times to repeat the function to profile
name – name of the profile
interval – see interval
module – 'pyinstrument' by default, 'cProfile' or 'profile' works too

source on GitHub

__init__(fct_match=None, name='ProfilerCall', repeat=1000, interval=0.0001, module='pyinstrument')¶

Parameters:

fct_match – function which tells if the profiler should be run on a set of parameters, signature is match(**kwargs) -> boolean
repeat – number of times to repeat the function to profile
name – name of the profile
interval –
see interval
module – 'pyinstrument' by default, 'cProfile' or 'profile' works too