[Bioperl-l] [Perldl] Perl Performance Project?

MARK BAKER mrbaker_mark at yahoo.com
Thu Jan 5 12:31:11 EST 2012





________________________________
 
 

The operating system is important with with benchmarking as well 

my most important perl scripts run 2 twice as fast on OpenSUSE 11.3 

then on windows server 2008 /XP  / windows 2003 at both 64 and 32 bit levels 


Which I think is a important point to mention here, may be we can get 

tests done  to find the fastest OS...


Cheers 


Mark R Baker



________________________________
 From: David Mertens <dcmertens.perl at gmail.com>
To: Steffen Schwigon <ss5 at renormalist.net> 
Cc: bioperl-l at lists.open-bio.org; perldl <perldl at jach.hawaii.edu> 
Sent: Thursday, January 5, 2012 7:24 AM
Subject: Re: [Perldl] Perl Performance Project?
 

Steffen (and BioPerl and PDL folks) -

This looks interesting, and I suspect that the PDL and BioPerl communities might be interested in this, so I'm CC'ing them. PDL and BioPerl folks - take a look! At one point I read about somebody who wrote some git plugins that don't allow commits to the main branch unless they improve the test coverage. I suspect that one might be interested in implementing a similar approach for one's own work, focusing also on benchmarking.

Steffen, a couple of suggestions:

	1. You should include a link to perlformance.net in the docs
	2. you should include a link to the graphs from the main web page
You give a skeleton for writing a plugin, but I'm not quite sure how to set up my own server to test my plugin. Would I install Benchmark::Perl::Formance, build and install my own plugin, and then run

david> benchmark-perlformance --plugins=MyPlugin

Is it that simple?

Thanks!

David


On Mon, Jan 2, 2012 at 10:32 AM, Steffen Schwigon <ss5 at renormalist.net> wrote:

Hi all!
>
>I now have my Perl benchmarking infrastructure ready and already some
>coverage over several Perl versions.
>
>The infrastructure consists of a Tapper and Codespeed instance, an own
>*not* regularly updated CPAN mirror (to keep dependencies stable), and a
>dedicated benchmark machine.
>
>One server is running a Tapper raw result database and website
>(http://perlformance.net/), the Codespeed graph rendering website
>(http://speed.perlformance.net/) and the CPAN mirror
>(http://perlformance.net/CPAN/).
>
>The second server (perl64.org [6 core AMD Opteron 4180]) is dedicated to
>only run benchmarks, without any disruption from email, web, or other
>services. I also took care of disabling all OS features that typically
>lead to deviation, like ASLR and Core Performance Boost. And, yes, Perl
>is built using Yet Another Great Perl Bootstrap Script(tm)
>(http://search.cpan.org/%7Eschwigon/App-Bootstrap-Perl%29%5B1%5D.
>
>Read more about the overall vision in my YAPC::EU 2011 slidedeck:
>
> http://perlformance.net/res/yapc_eu_2011_perlformance-net.pdf
>
>I blogged this also with some more details here:
>
> http://blogs.perl.org/users/steffen_schwigon/2012/01/perlformance.html
>
>Principally the infrastructure is able to consume benchmarks from other
>providers. Talk to me if you want to track numbers from your machines.
>
>A side effect is a nice collection of many Perl installations. I can
>easily upgrade and rerun new benchmarks over them. So whoever is
>interested in benchmarks, please write a code snippet, ideally a
>Perl::Formance plugin, and talk to me.
>
>Theoretically I also backup the data, let's see how reliable... :-)
>
>Happy New Year!
>
>Kind regards,
>Steffen
>
>Footnotes:
>[1]  mine is better than yours :-), it can distroprefs! ANDK++
>
>PS: I regularly struggle with dependencies when bleadperl breaks CPAN
>   or on 5.8.x, so there is still some maintenance effort and the
>   “automation” is more or less still a while(true) loop -- but who
>   cares...
>--
>Steffen Schwigon <ss5 at renormalist.net>
>Dresden Perl Mongers <http://dresden-pm.org/>
>


-- 
Sent via my carrier pigeon.

_______________________________________________
Perldl mailing list
Perldl at jach.hawaii.edu
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl


More information about the Bioperl-l mailing list