benchmarking FC1
Chadley Wilson
chadley at pinteq.co.za
Wed Jan 28 13:21:00 UTC 2004
On Tue, 2004-01-27 at 17:19, George wrote:
> >>>>> "Emiliano" == Emiliano Brunetti <emiliano.brunetti at fastwebnet.it> writes:
>
> Emiliano> On Tue, 2004-01-27 at 15:51, George wrote:
> >> >>>>> "Ben" == Ben Stringer <ben at burbong.com> writes:
> Emiliano> ...
> >> a lot of time in the process. If you really need performance, apply
> >> the money you save to more RAM.
>
> Emiliano> George,
>
> Emiliano> i was 'asked' to do this job. I can agree with you on many points, yet i
> Emiliano> have to do some kind of testing.
>
> Emiliano> E.
>
> I read that again and I sounded pretty pompous. Sorry.
>
> --
> *********************************************************************
> * George Yanos * *
> * UTC at UIC * *
> * 312-413-0059(w) * *
> * 708-848-4221(h) * *
> * gyanos at uic.edu * *
> *********************************************************************
195803
OK Guys, I understand whats needed here, without detail I will try shed light
It is what we hardware technicians call a system benchmark,
How this works is we run a program that shows you numerically the maximum performance
you can get out of the combination of the hardware, the loaded drivers combined with the
software (OS) installed .
So take word for example how fast can the hardware process documents of a certain size or how long
will it take to compress a series of files of a certain size, also when playing games how many
frames per seconds [FPS] at the various resolutions and color depths can the card produce.
Just remember that often the manufacturers spec and the actual performance differ due to the
current state of the systems and many other restrictions.
Our company PROLINE Computers for example will test several different configurations
we then choose the best performers and those that we find to be most stable under the load.
These configurations are what we would use as the ideal systems for that operating system,
OK I know what some are going to say, the kernel runs at the same speed on this hardware
as on any other hardware, this maybe true, however your drivers will certainly influence the
speed of a system, the mix of components also effects the speed of your system.
Here is an example we tested 2 PCs for gaming loaded with Windoze Xplease and the difference in the
results just on the graphics card in FPS was 150,The difference in the specs was the on had a via chip-set and the other had
an Intel board and chip-set, both had same ram, hard disk and screen-cards etc..
We then load the latest Intel inf update for windows Xplease and we found that the upgrade agp driver combined
with the new directX and the latest Detonator driver for the card brought the Intel up by 98 FPS.
So we repeated the exercise on the VIA PC and it rocketed the difference to an astonishing 234 FPS.
We need tool like 3dbenchmark an win-bench for Linux.
As a manufacturer we are expected to provide the fastest machines on earth for as little as possible money
Its challenging but fun
Anyway thats my two cents
Have a great day
--
Chadley Wilson <chadley at pinteq.co.za>
More information about the fedora-list
mailing list