profiling.txt

.. _tut_profiling:

=========================
Profiling Theano function
=========================

.. note::

    This method replace the old ProfileMode. Do not use ProfileMode
    anymore.

Besides checking for errors, another important task is to profile your
code. For this, you can use Theano flags and/or parameters which are
to be passed as an argument to :func:`theano.function <function.function>`.

The simplest way to profile Theano functions is to use the Theano
flags described below. When the process exits, they will cause the
information to be printed on stdout.


Using the ProfileMode is a three-step process.

Enabling the profiler is pretty easy. Just use the Theano flag
:attr:`config.profile`.

To enable the memory profiler use the Theano flag:
:attr:`config.profile_memory` in addition to :attr:`config.profile`.

To enable the profiling of Theano optimization phase, use the Theano
flag: :attr:`config.profile_optimizer` in addition to
:attr:`config.profile`.

You can use the Theano flags :attr:`profiling.n_apply`,
:attr:`profiling.n_ops` and :attr:`profiling.min_memory_size` to
modify the quantify of information printed.

The profiler will output one profile per Theano function and profile
that is the sum of the printed profile. Each profile contains 4
sections: global info, class info, Ops info and Apply node info.

In the global section, the "Message" is the name of the Theano
function. theano.function() has an optional parameter ``name`` that
defaults to None. Change it to something else to help you profile many
Theano functions. In that section, we also see the number of time the
function was called (1) and the total time spent in all those
calls. The time spent in Function.fn.__call__ and in thunks is useful
to help understand Theano overhead.

Also, we see the time spent in the two parts of the compilation
process: optimization(modify the graph to make it more stable/faster)
and the linking (compile c code and make the Python callable returned
by function).

The class, Ops and Apply nodes sections are the same information:
information about the Apply node that ran. The Ops section takes the
information from the Apply section and merge the Apply nodes that have
exactly the same op. If two Apply nodes in the graph have two Ops that
compare equal, they will be merged. Some Ops like Elemwise, will not
compare equal, if their parameters differ (the scalar being
executed). So the class section will merge more Apply nodes then the
Ops section.

Here is an example output when we disable some Theano optimizations to
give you a better idea of the difference between sections. With all
optimizations enabled, there would be only one op left in the graph.

.. note::

    To profile the peak memory usage on the GPU you need to do::

        * In the file theano/sandbox/cuda/cuda_ndarray.cu, set the macro
          COMPUTE_GPU_MEM_USED to 1.
        * Then call theano.sandbox.cuda.theano_allocated()
          It return a tuple with two ints. The first is the current GPU
          memory allocated by Theano. The second is the peak  GPU memory
          that was allocated by Theano.

    Do not always enable this, as this slowdown memory allocation and
    free. As this slowdown the computation, this will affect speed
    profiling. So don't use both at the same time.

to run the example:

  THEANO_FLAGS=optimizer_excluding=fusion:inplace,profile=True python doc/tutorial/profiling_example.py

The output:

.. literalinclude:: profiling_example_out.prof