Hardware

Possible Major Flaw In Storage Reviews!

Recommended Posts

I dont have a 250g drive here can somebody make the bench?

Where did your numbers come from then?

Do you have a link? Or are those "short-stroke" estimates?

Regardless....re-read Gilbo's post - with the excerpt from the FAQ.

Dogeared

8^|

Share this post


Link to post
Share on other sites
I dont have a 250g drive here can somebody make the bench?

Somebody already did...

How many times do we have to say, SR benchmarks are run on the first 30GB of any drive tested. One more time, SR BENCHMARKS ARE RUN ON THE FIRST 30GB OF ANY DRIVE TESTED! They have no choice, the access pattern that they replay was recorded from an 30GB partition.

SR is already doing exactly what you requested.

-JoeTD

Share this post


Link to post
Share on other sites
SR is already doing exactly what you requested.

No, you idiot!!! They use the first 30 GB and the Raptor is 36 GB. Gee, I'd thought you'd have noticed this basic flaw.

Share this post


Link to post
Share on other sites

http://www.storagereview.com/php/benchmark..._1=249&devCnt=2

check it out! SR too shows the different access time b/w the 74GD and the 36GD!! what a suprise... and SR, too, has an end xfer rate for the 74GD that is close to the beginning xfer rate for the 36GD... hmmm

Share this post


Link to post
Share on other sites

Access time doesn't tell the whole story. There's also other factors that come into play when it comes to disk performance. That's why high end SCSI drives outperform high end IDE/SATA drives when it comes to multi user access.

If it were simply access time and transfer rate, then there's be no difference between SCSI and IDE/SATA drives. That's clearly not true.

The graph that 'shows evidence' simply shows that a partial seek takes less time than a full stroke seek. That hardly takes a genius to work it out. Xbit labs made the mistake of using a tool that has no control over data locality. As data locality has a direct impact on seek times, they had no control over it.

However, if you had bothered to read the StorageReview testing methodology, you'd know that SR's tests have the concept of data locality. Traces are used because they are the best represenative of real world performance.

Real world performance isn't about boosting benchmark numbers, it's about using the most suitable drive for the job.

Access time on a 36(185) may well be lower than on a Raptor 36. If I am doing random accesses across a 30Gb file, the Raptor will provide me with less performance than the 185, as the Raptor is spending more time seeking. You've claimed that this is a major flaw, but it mirrors what happens in real life. Either you are only interesting in running meaningless benchmarks on your disk, or you don't actually care about the performance.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now