Re: Why does writing STDOUT affect my Python LDAP performance?

[Joe L. <jl...@op...>]

I'm using the same RawHide kernel (rebuilt) for my builds of python-ldap an=
d
the like. RedHat was experimenting with Ingo Molnar's new scheduler, and
this is likely the suspect difference. There is a new patch he has just
release, which I may apply against the kernel and rebuild. If you want the
best performing 2.4.9 kernel, I'd go with 2.4.9-21 (RedHat 7.2 proper)

BTW -- Hiya. At a new company?

On 2/12/02 9:33 AM, "Leif Hedstrom" <le...@pr...> wrote:

> Michael Str=F6der wrote:
>=20
>> I've added measuring the time to the script itself and can't
>> reproduce your results (tested with python2.0, python2.1.1 and
>> python2.2). In my case redirecting stdout to /dev/null was faster.
>> How did you do time measuring? Maybe somebody screwed up /dev/null
>> on your system?
>>=20
>=20
> Well, the good news is that I've solved my problem, and it's not related
> to Python LDAP at all. :-).
>=20
> I wrote a small C-program that also reproduced the same problem. So,
> this narrowed it down to a few things: C-compiler, glibc or kernel. I
> hadn't touched the first two in quite a while, but I did upgrade the
> kernel a week ago... So, I rolled back to 2.4.7 (from 2.4.17), and now
> my problem is gone. Don't ask me what happend with the 2.4.17 kernel, I
> hadn't noticed any problems or errors at all after the upgrade (RPM from
> Rawhide), but something is not good...
>=20
> The module still leaks memory though, I'll poke at that little later.
> Thanks again for the help, so I could narrow things down to the real
> problem. :-)
>=20
> Thanks,
>=20
> -- Leif
>=20
>=20
> _______________________________________________
> Python-LDAP-dev mailing list
> Pyt...@li...
> https://lists.sourceforge.net/lists/listinfo/python-ldap-dev