I'm planning to use python to build a couple of programs to be used as services, run from PHP code later on. In terms of performance, which is faster, to compile the python code into a binary file using cx_freeze or to run the python interpreter each time I run the program?
Deployment environment:
OS: Arch Linux ARM
Hardware: Raspberry Pi [700MHz ARMv6 CPU, 256MB RAM, SD Card filesystem]
Python Interpreter: Python2.7
App calls frequency: High
You need to test it, because there's no single right answer. All cx_freeze does is wrap up the bytecode into an executable, vs. the interpreter reading from its cached .pyc on disk.
In theory the packaged executable could be quicker because it's reading fewer files, but on the other hand the interpreter could be quicker because it could already be in the disk cache.
There's likely to be little to choose, and whatever the difference is, it's not down to "compiled" vs. "interpreted".
cx_freeze (and the various competitors for "compiling" Python code to executables) don't actually compile anything. They're just a convenient way of packaging the app in such a way that it can be run directly. In other words, there's no performance difference.
Depending on what you need to do in your Python script, you could consider using Pypy to improve your performance.
