PyPy Vs Unladen Swallow
sarvi
project and Unladen Swallow was supposed to attempt a 5x speedup
within a year.
Now that we are here, is that still true?
When I look at what PyPy has accomplished this last year, PyPy is now
a real contender to replace CPython.
Shouldn't a company like Google be putting its resources into PyPy and
RPython????
Specially since Oracle is beginning to assert ownership to Java. All
the more reason to beefup Python.
PyPy seems to have momentum and is rapidly gaining followers and
performance.
PyPy JIT and performance can make it on par with Java.
And it seems to be well ahead of Unladen Swallow in performance and
in
a position to improve quite a bit.
Secondly I have always fantasized of never having to write C code yet
get its compiled performance.
With RPython(a strict subset of Python), I can actually compile it to
C/Machine code
These 2 seem like spectacular advantages for Python to pickup on.
And all this by just showing PyPy, Python foundation and Google's
support and direction to adopt them.
Yet I see this forum relatively quite on PyPy or Rpython ? Any
reasons???
Sarvi
dmitrey
IIRC
1) PyPy showed great speedup for integer operations only
2) Some of PyPy modules didn't stack with some C extensions.
Have the situation changed?
D.
P.S. I'm not an Unladen Swallow developer.
Reid Kleckner
> Hi sarvi,
> IIRC
> 1) PyPy showed great speedup for integer operations only
> 2) Some of PyPy modules didn't stack with some C extensions.
> Have the situation changed?
PyPy is growing support for interoperating with C extension modules,
actually, which changes things somewhat. However, unlike PyPy, we
still have the goal of upstream integration, so users of CPython
should be able to seemlessly migrate onto CPython with a JIT.
> D.
> P.S. I'm not an Unladen Swallow developer.
>
> On 1 сен, 21:12, sarvi <sarvil...@gmail.com> wrote:
>> I understand 1 year ago, it might have looked like PyPy is decade long
>> project and Unladen Swallow was supposed to attempt a 5x speedup
>> within a year.
>>
>> Now that we are here, is that still true?
We haven't achieved our performance goals in the time that we set for
ourselves, but the approach is still interesting and viable. It just
needs a fair amount of work that we haven't been able to get time to
do.
>> When I look at what PyPy has accomplished this last year, PyPy is now
>> a real contender to replace CPython.
>> Shouldn't a company like Google be putting its resources into PyPy and
>> RPython????
>>
>> Specially since Oracle is beginning to assert ownership to Java. All
>> the more reason to beefup Python.
>>
>> PyPy seems to have momentum and is rapidly gaining followers and
>> performance.
>> PyPy JIT and performance can make it on par with Java.
How do you quantify momentum? Do you know any large services using
PyPy in production?
PyPy is certainly gaining momentum in the sense that they are gaining
performance and adding compatibility, but most people are still using
CPython, and we want to support them.
>> And it seems to be well ahead of Unladen Swallow in performance and
>> in
>> a position to improve quite a bit.
>> Secondly I have always fantasized of never having to write C code yet
>> get its compiled performance.
>> With RPython(a strict subset of Python), I can actually compile it to
>> C/Machine code
>> These 2 seem like spectacular advantages for Python to pickup on.
>> And all this by just showing PyPy, Python foundation and Google's
>> support and direction to adopt them.
Just because a language is statically compiled does not imply that it
will have the same performance characteristics as C. That said, I
haven't directly compared the performance of RPython to C, and I can't
find any benchmark comparisons for it.
>> Yet I see this forum relatively quite on PyPy or Rpython ? Any
>> reasons???
No, none in particular. There's a lot of things we can learn from
PyPy. They just haven't come up in email, mostly in face-to-face
communication and IRC.
Reid