sethd.org
this is not the blog you're looking for

Resume
LinkedIn

RECENT ENTRIES

Interesting Modules 2010-07-30
Recent Interesting Talks
Book Blogging: Bloodbrothers
Book Blogging: Fires of Freedom
Duck Duck Go
Yes, Yes, Yes
About That Lack of Blogging
Using A Smart Phone
Taking the Smart Phone Plunge (Maybe)
Book Blogging: Nemesis
Perl Not Going Away
OpenX Is Hiring Perl Developers
Perl Jobs vs Perl Programmers
Free Chapter From Effective Perl Programming
Perl On Android Progressing
Book Blogging: The Mote In God's Eye
Perldoc Is Important
The Second Age of Perl
A Description of Perl 5.12
Interesting Modules 2010-04-18
The Moral of This Story
Perl 5.12 Has Been Released
Packaging and Maintaining An Alternate Perl
Perl Moving Up?
Building Dependencies Like Make
Introduction To Plack
Defining Standard Testing Methodologies
Subversion Vision Released
Interesting Modules 2010-04-01
Perl Is Thriving
Interesting Modules 2010-03-29
Assign to $0
The Looming Google AdWords Perl Problem: Followup
Subversion Vision
Interesting Modules 2010-03-26
Interesting Modules 2010-03-25
The iPhone and Perl
Perl Is Dying
Setting Up A Windows Computer: Part 3
Trouble Hiring Perl Developers
Back From San Diego 2010
Interesting Modules 2010-03-17
Remember To Use parent Instead of Base
Using Test::Class
Mar 03, 2010

Dave Rolsky Is Right

Yes he is. Granted I woudn't go quite so far as to say, "Benchmarking is pointless", but his point about the comparative usefulness of benchmarking versus profiling is right on the money.

I think the reason for this is not an inherent problem with benchmarking. The problem is that people (developers) tend to benchmark the wrong things and then they optimize the wrong things. Profiling on the other hand takes the entire system into account in its results. This is much more useful than optimizing based on a benchmark of a small part of a complex system.

Tags: .
[p] Posted @ 14:54 by Seth


This site uses the very simple and easy to use blosxom blogging software.


Advanced Search

OTHER SITES