Python Processor

manish manish.paradkar at
Thu Aug 27 04:39:21 CEST 2009


    Contrary to the ancient and, I believe, obsolete text in the
     there is no 'python bytecode'. Dis is based on CPython bytecode.
     Jython uses Java bytecode. IronPython uses, I believe, Microsoft clr
bytecode. Object code compilers do not use bytecode.

         Before going through the one of the existing VM implementations, i
want to know if there is there any informaition
         available which elaborates on the structure of the bytecodes. Like
details regarding co_names, co_cellvars,
         co_freevars, the concept of frames, global, local, objects,
functions etc with respect to the bytecode.

     These are mostly version-specific Cpython implementation details.


I agree, i am looking at a version-specific CPython implementation, or can
be PyPy even . I need some information
on the implementation of these interpreters which would help me visualize
how they can be mapped to hardware.

      It probably wouldn't help much.  CPython's performance problems
     come from excessive dictionary lookups, not from instruction decode.

                               John Nagle

This is just for my personal interest so currently i am not looking at
performance - just be able to implement a basic core at first.
The idea is to then extend it and develop a self contained unit of
computation which can be reconfigured on the fly. So given enough
memory and I/Os the same module can behave differently depending on the
script loaded. One can either have a large monolithic cpu
running an OS of some sorts or smaller processors each executing their own
little scripts and communicating with each other.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the Python-list mailing list