Sensible benchmarks for a desktop Linux system?

woogie · 2006-05-13 20:52:55

By the end of the month, I'll have done a massive overhaul of my PC - going to brand spanking new just about everything. But it's not enough to just know that my new system will be faster. I want to know how much faster it is with quantifiable numbers and everything.

So one of the big things I'll be looking at will be video games. Doing that is easy though - just use something like what review sites like hardocp do and count my framerates.

But on the other side, I want to see how my linux performance changes. I know I can measure my FLOPS, Bogomips, and other numbers, but those are kind of abstract. Doesn't show anything about the applications I use every day.

So what can you guys think of that would be sensible to do for benchmarking a Linux system to get before-and-after comparisons?

My thoughts:

Page renderings in Firefox. Make some kind of convulted, ugly mess of a page to render then see how long it takes.
Code compilation - it's something I do from time to time, and easy enough to benchmark.
Video encoding and decoding - Something I do every now and then using mplayer et al
Audio encoding and decoding - I try to keep my music collection in ogg format
Rendering - I'm starting to play with Blender, maybe make a horrifically complex scene (or better yet, download one) and time the render process?
Database queries - This isn't something I do a lot on my system, but more and more applications are starting to use database backends (like amaroK although I don't use it) I think there are benchmark suites available for MySQL and Postgres

My only other thought is that I should try to measure the increase in responsiveness I get from going from a single core to dual core system. Trying to measure that is kind of tough though. Maybe a scenario where I do something arbitrary in the background, then measure dropped frames in mplayer playing a video of some sort?

cactus · 2006-05-15 01:04:23

meh. that benchmark sucked. Not much info given on testing environment, number of runs, etc.
Some very odd marks in there as well...makes me wonder if he just ran the tests once and said "good enough".
*shrug*

Arch Linux

#1 2006-05-13 20:52:55

Sensible benchmarks for a desktop Linux system?

#2 2006-05-15 01:04:23

Re: Sensible benchmarks for a desktop Linux system?

Board footer