Announcement

Collapse
No announcement yet.

PyPy 4.0 Released For Speedy Python

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

  • PyPy 4.0 Released For Speedy Python

    Phoronix: PyPy 4.0 Released For Speedy Python

    PyPy 4.0.0 was released today as a major update for this Python 2.7 interpreter and JIT compiler...

    http://www.phoronix.com/scan.php?pag...y-4.0-Released

  • #2
    2.7, old shit!
    3.5 or gtfo!

    Comment


    • #3
      Originally posted by uid313 View Post
      2.7, old shit!
      3.5 or gtfo!
      They target Python 2.7.10 and 3.2.5, and they're working on updating the 3.2.5 compatibility to 3.3 ( see http://morepypy.blogspot.com/2015/06...0-release.html )

      They average 7 times faster than CPython. That's not poor by any means. But on the other hand, CPython is routinely 100 times slower than C. So PyPy drops us to routinely 14 times slower than C. Huge improvement, but still ground left to cover.

      On the bright side, developing with Python is so fast that maybe the real answer is just:
      1: Write in Python
      2: Profile the result.
      3. If the result is sufficiently fast for your needs, goto 6:
      4: Rewrite slowest portion in C.
      5: Goto 2:
      6: Exit.

      Comment


      • #4
        Originally posted by uid313 View Post
        2.7, old shit!
        3.5 or gtfo!
        I certainly have no interest in a has been version of Python.

        Comment


        • #5
          Well, the PyPy project is more about a highly compatible and considerably faster version of Python. Right now their biggest hurdles is cpython extension compatibility, say with the whole scientific python stack (which is currently mostly still using 2.7).

          Also the story of CPython being ~100 times slower than C isn't really true, it is more like ~30 times in my experience, Python has definitely gotten faster over time (no big jumps, but gradually).

          Also the first Py3 version that got good traction from developers was 3.3 (3.0/3.1 was more like concept demos, 3.2 was the first 3 you were supposed to use, but 3.3 fixed a lot of the nigglies).
          Keeping an application working from 3.3 to 3.5 is pretty trivial, as most of the changes are library changes, So once they get a 3.3 version out, I think they will catch up pretty quickly.

          All in all, I love PyPy. Just too bad that it often isn't available in the distro repos, or only archaic versions are bundled. Sigh

          Comment


          • #6
            Well, the PyPy project is more about a highly compatible and considerably faster version of Python. Right now their biggest hurdles is cpython extension compatibility, say with the whole scientific python stack (which is currently mostly still using 2.7).

            Also the story of CPython being ~100 times slower than C isn't really true, it is more like ~30 times in my experience, Python has definitely gotten faster over time (no big jumps, but gradually).

            Also the first Py3 version that got good traction from developers was 3.3 (3.0/3.1 was more like concept demos, 3.2 was the first 3 you were supposed to use, but 3.3 fixed a lot of the nigglies).
            Keeping an application working from 3.3 to 3.5 is pretty trivial, as most of the changes are library changes, So once they get a 3.3 version out, I think they will catch up pretty quickly.

            All in all, I love PyPy. Just too bad that it often isn't available in the distro repos, or only archaic versions are bundled. Sigh

            Comment


            • #7
              Well, the PyPy project is more about a highly compatible and considerably faster version of Python. Right now their biggest hurdles is cpython extension compatibility, say with the whole scientific python stack (which is currently mostly still using 2.7).

              Also the story of CPython being ~100 times slower than C isn't really true, it is more like ~30 times in my experience, Python has definitely gotten faster over time (no big jumps, but gradually).

              Also the first Py3 version that got good traction from developers was 3.3 (3.0/3.1 was more like concept demos, 3.2 was the first 3 you were supposed to use, but 3.3 fixed a lot of the nigglies).
              Keeping an application working from 3.3 to 3.5 is pretty trivial, as most of the changes are library changes, So once they get a 3.3 version out, I think they will catch up pretty quickly.

              All in all, I love PyPy. Just too bad that it often isn't available in the distro repos, or only archaic versions are bundled. Sigh

              Comment


              • #8
                Originally posted by grigi View Post
                Well, the PyPy project is more about a highly compatible and considerably faster version of Python. Right now their biggest hurdles is cpython extension compatibility, say with the whole scientific python stack (which is currently mostly still using 2.7).

                Also the story of CPython being ~100 times slower than C isn't really true, it is more like ~30 times in my experience, Python has definitely gotten faster over time (no big jumps, but gradually).

                Also the first Py3 version that got good traction from developers was 3.3 (3.0/3.1 was more like concept demos, 3.2 was the first 3 you were supposed to use, but 3.3 fixed a lot of the nigglies).
                Keeping an application working from 3.3 to 3.5 is pretty trivial, as most of the changes are library changes, So once they get a 3.3 version out, I think they will catch up pretty quickly.

                All in all, I love PyPy. Just too bad that it often isn't available in the distro repos, or only archaic versions are bundled. Sigh
                For some reason your post came through three times. Weird.

                In the toy benchmarks at the "great computer language shootout" site, in some benchmarks CPython is only 2-3 times slower than C++. But in others it's more than a hundred times slower.

                I'm not trying to criticize the language. I've said it on these forums, on reddit, and elsewhere - I fear many developers, especially me wasted hundreds of hours debating the relative merits of different programming languages in search of our ideal tool. In the same time period some other people took the tool at hand - Python, Perl, Awk, Colfusion, VisualBasic, whatever - and built something useful with it. The language is a means to getting something accomplished, and the world is full of tens of thousands of examples where Python was an excellent fit.

                Comment

                Working...
                X