repository of my complied code to go with notes:

• cython basics work-bench

cython

cython is python with statically typed python/C data types, so essentially a mash-up of performance benefits of C and high-level ease-of-programming of python

• cython is python – almost any piece of python code is also valid cython code

• cython is not a python to C translator, it is an alternative complier for python
  • more specifically, an optimizing static compiler
• the code runs within the python runtime environment; but rather than compiling to interpreted python code, it compiles to native machine code

cython benefits

• because parameters and variables can be declared to have C data types and explicitly declared python variable types (statically typed), code which manipulates python variable types and C variable types can be freely intermixed, with conversions occurring automatically wherever possible
• reference count maintenance and error checking of python operations is automatic
• python’s exception handling facilities, including the
  • try-except and
  • try-finally statements, is available to in the midst of manipulating C data

compiling code with cython

• one of the hardest part of getting started with cython: figuring out how to compile the code file
  • cython uses ‘.pyx’ file extension instead of the typical python’s ‘.py’ extension

• to compile with cython to ‘.c files, save python code to a '.pyx' extensioned file (fileToComplie.pyx` for example)

• then, create a setup.py in the same directory with following code:

    from distutils.core import setup
    from Cython.Build import cythonize
  
    setup(
        ext_modules = cythonize("fileToComplie.pyx")
    )
  

• then, run this setup.py file to build the ‘.pyx' file to C:

    $ python setup.py build_ext --inplace
  

  this will create a ‘.c’ and a ‘.so’ (shared object) in the working directory along with a build directory

• then, in the python CLI, do:

    import fileToCompile
  

  this will import the complied ‘.so’ into python as a package, and immediately run the script (displaying any print statement outputs)

• the functions defined within this import can be used like using functions from any other imported python package

cython data types

• cython data types have to be defined explicitly in the ‘.pyx’ file code to utilize the statically typed performance benefits
  • using cdef or cpdef along with explicitly setting variable type enables compilation to machine level code

  • statically typed python-style cython variables:

      cdef int x,y,z
      cdef char *s
      cdef float x = 5.2 (single precision)
      cdef double x = 40.5 (double precision)
      cdef list languages
      cdef dict abc_dict
      cdef object thing
    

• function definition prefixes
  • def: regular python function interpretation, returns python object
  • cdef: cython only functions, can’t access these from python-only code (.py files), must access within cython, since there will be no C translation to python for these
  • cpdef: for cython and python compilation , will create a cython function and a wrapper for python
    • in some cases when there is a C only pointer like a C array, then cpdef will throw errors

  • C arrays (in lieu of numpy array) can be used through cython by using

      from cython.view cimport array as cvarray
    

python 3 compiler directive

• add # cython: language_level=3 as the first line of the .pyx file to ensure complier knows python3 code is being complied

learnings

• cython is a compiler which compiles python-like code files to C code

• cython is not a python to C translator, the code runs within the python runtime environment; but rather than compiling to interpreted python code, it compiles to native machine code

• cython data types have to be defined explicitly in the ‘.pyx’ file code to utilize the statically typed performance benefits

• having compared a mean-squared-error function performance in cython and python, the cython function is 5 times faster on an average

fig: performance boost measurement with timeit

• run runtest.py in performance-tests folder in the work-bench repository to see performance results on your machine

---

References:

basics

• cython - Basic Tutorial
• cython type declarations
• cython for numpy users

performance comparisons

• performance gain from cython
• performance comparison of python, cython and numpy

limitatons

• cython - Limitations

---