Announcement

Collapse
No announcement yet.

Pyston 0.3 Released For High-Performance Python

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Originally posted by coder111 View Post
    Now what about the countless CPython extensions written in C? Numpy? Others? Will they be compatible? How long will it take to make them compatible?
    All this is answered on their page. The goal is 100% compatibility with CPython extensions.

    Originally posted by coder111 View Post
    These resources should be spent working on PyPy IMO.
    As I already explained, trying several different approaches to see what works best is common in the python ecosystem.

    Comment


    • #12
      JIT is crap compared to using an intermediate representation/language/bytecode and AOT(Ahead Of Time) Compilation.


      As used in Androids new runtime called ART.
      https://software.intel.com/en-us/blo...id-x86-runtime

      http://www.chupamobile.com/blog/2014...art-vs-dalvik/

      Using IR(Intermediate representation/intermediate language) and AOT(Ahead Of Time) compilation is the way to go!

      Comment


      • #13
        python release

        not so many upgraded as we wanted.

        Comment


        • #14
          Originally posted by plonoma View Post
          JIT is crap compared to using an intermediate representation/language/bytecode and AOT(Ahead Of Time) Compilation.
          How is this different from what CPython already does?

          Comment


          • #15
            CPython does not do AOT compilation to native code.

            Comment


            • #16
              Originally posted by SciK View Post
              CPython does not do AOT compilation to native code.
              You are right, it does't compile to native code, nor is it likely such a thing would even be possible in general in python. Efforts to compile python code to machine code have only been successful by imposing additional restrictions on the language compared to normal python (shed skin) or making assumptions about how the code runs (Psyco, Pypy).

              This is where JIT benefits a programming language like python. By looking at how the code is used in practice, peformance hot spots that are often used in a consistent manner can be identified and compiled to machine code, allowing python to maintain its flexibility while still getting performance benefits. I am nopt sure about Pyston, but I know Pypy does this when the JIT compiler is enabled.

              Comment

              Working...
              X