Python IS slow ! [was] Re: Python too slow for real world

[Christian Tismer]

Gordon McMillan wrote:
> 
> Christian wrote:
> 
> > Markus Kohler wrote:
> > <snip/>
> > > Running pystone on HP-UX the result for 1.5.2c is 3.37% for malloc and
> > > 3.95%
> > > for free :-(
> >
> > Are you absolutely sure?
> 
> /snip snip/
> 
> I find Markus' numbers just-slightly-surprising. I don't find your
> numbers for Windows surprising at all. After all, there are some
> software companies who charge $800 or so for a drop in replacement,
> and claim 30-50% speed ups. Allowing for marketing-inflation, that's
> just about right.

Hmm. Seems to me as if I could do the same thing for Python.

> (Actually, there are, or at least were, certain pathological cases
> in which multithreading on a multiprocessor Windows box would slow
> to a crawl because of poorly designed locking within malloc / free.)

A malloc just for Python - would it need locking at all?
Since the interpreter is locking already, I guess malloc
would be always called in a safe way.

> (And, in MS's defense, let me point out that their debugging versions
> are very nice - they let you track down rogue pointers / premature
> frees much faster than most other libs.)

Absolutely, this is nice. Although I still have a bad
memory error with Python running through many Excel files,
and with the debugging library, the error just doesn't
appear :-(

ciao - chris

-- 
Christian Tismer             :^)   <mailto:tismer at appliedbiometrics.com>
Applied Biometrics GmbH      :     Have a break! Take a ride on Python's
Kaiserin-Augusta-Allee 101   :    *Starship* http://starship.python.net
10553 Berlin                 :     PGP key -> http://wwwkeys.pgp.net
PGP Fingerprint       E182 71C7 1A9D 66E9 9D15  D3CC D4D7 93E2 1FAE F6DF
     we're tired of banana software - shipped green, ripens at home