Am I testing right?

[TheGoatMan]

I'm sure any of you who have read my posts are wondering who sold me the crack because of my recent escapades. I went gung-ho with this computer thing and spend many sleepless nights putting things together and running all sorts of benchmarks. Now all of my results have not agreed with what like a respectable product tester showed. Like for example: My comparison between the 2600+ tbred and 2500+ barton showed the tbred to be a faster chip. Meanwhile comparing an nforce to a KT600 (which many say is no comparison) The memory bandwidth, games, and overclocking ability the VIA somehow managed to always stay ahead. I matched an Abit NF-7 against a MSI KT6 Delta lsr. So do you think my results were due to poor comparisons or lack of know how for tuning? Or is there a margin of error? Your 2 cents please.

 

[paparazi]

As with all testing, the larger the sample the more accurate the test becomes.
Testing just one motherboard against another with just one set of components leaves the chances for large errors.
How do we know if both samples were good representations of the products, one could be faulty or just be a poor performing example.
If on the otherhand there were several samples tested from both camps, with varying hardware configurations including many differant types of ram etc, the margin for error deminishes and the test becomes more accurate.
Martin

Replying helps further our knowledge, without comment leaves us wondering.

 

[TheGoatMan]

We're throwing another board and chip into the mix. Along with some corsair PC3500 memory. A gigabyte with a nforce ultra is on the way. We also tested on a ASUS A7V333 but it was pretty limited with the same results. We'll also try a 2000+ and a 1700+ chip. But all parts are brand new. So the margin of error couldn't be that bad. At least I hope. Oh and a 9600xt just arrived too for comparison. We'll see how that fairs.