Home » Opinions » Benchmarking Blues

Benchmarking Blues

by | Go to comments

Share:

So you’ve heard the hype, read the reviews and pored over the benchmark results. Good for you - you’ve got the easy part.

While it takes a short time to skim through a review and the accompanying results, what you probably don’t know is the sheer effort that goes in to testing hardware. To put it mildly, it can be a real pain.

For any benchmark to be worth its salt it needs to be demanding. This requires that it take a certain amount of time to run. And as no single test can be relied upon to properly evaluate something benchmarking tends to be a long winded process. And that's when everything goes smoothly.

Recently though, things have been far from smooth. In the last few months we’ve seen major new technologies introduced in the form of dual core CPUs and SLI graphics. While this is great it seems that there’s been too much new technology rushed out of the door by manufacturers before it has been properly tested.

/94/4b0246/7f11/1594-1.jpg

SLI motherboards are a real case in point. In one particularly painful benchmarking experience we had two boards from different manufacturers running concurrently and neither was able to complete a full set of tests. In order to paint as full a picture as we can in our reviews we try and get as complete a set of results as possible. But both boards would run without problems for a while before suddenly falling over without any logical explanation.

Often BIOS revisions fix these sorts of flaws but when you’re testing a board that isn’t even available yet the chances are that the BIOS hasn’t even reached version 1.0 so the chances of sorting the particular problem in any useful time-frame is minimal.

Of course, when you call someone up from the company to find out what’s going on they usually claim that they had no problems during testing. After further questioning we generally discover that the only test run prior to it being shipped out to us is 3DMark03, as if this is the be-all and end-all. This is quite frankly ridiculous as our tests prove that it alone is no indicator of stability.

Go to comments
comments powered by Disqus