Re: Pure Python implementation of hashcash v.1
"David Mertz, Ph.D." <[email protected]>
| Newsgroups | gmane.mail.spam.hashcash |
|---|---|
| Message-ID | <[email protected]> |
On Sep 23, 2004, at 10:59 PM, Mitch Denny wrote:
> http://www.gnosis.cx/download/gnosis/util/hashcash.py
> Hi, it would be interesting to see whether it ran faster on IronPython
> which
> compiles Python code to a .NET backend. Apparently the CLR backend can
> actually speed up execution times too.
Well, sure. I'd be interested in knowing that. But I don't have
access to a .NET environment (don't run Windows, on principle; haven't
installed Mono because of a lack of a compelling reason).
If someone with .NET wants to try my hashcash.py module on IronPython,
I'd love to see results. Just run:
./hashcash.py -sm [email protected]
To get some rough benchmarks. The time for generating a megahash
varies wildly, but the hashes per second seems stable across runs on
the same machine. If you use a machine with Psyco installed (JIT for
Python, basically), it boosts it another 20% or so.
In any case, it's a lot slower than the C version. It's even more
slower than the *amazingly* good G4/Altivec version, which just screams
on my Powerbook (exactly the same as the published benchmark, FWIW)...
I guess a G5 is even better.