Jump to content
Sign in to follow this  
Angry_Games

hard drives & raid - benchmark and compare!

Recommended Posts

Guest Tmod

Ok so I am a little late to the party.

 

Here is my Ultra Infinity with the addition of a LinkDepot 3112A Add-In PCI card.

 

DFI%203112%20Raid%200.JPG

 

Tmod

Share this post


Link to post
Share on other sites
But slower then my Maxtor 200GB with 16MB cache. The two guys that ran the raptor tests didn't run the test to large enough data size. You'll notice on their graphs they only go to about 148GB.

 

Take a look at my graph at 148GB mark and let me know what you see :)

 

http://www.dfi-street.com/forum/attachment...ttachmentid=286

 

 

What is your ATTO at the same settings ?

Share this post


Link to post
Share on other sites

I bees all "corn-fused" JeffP. :) 2x74G is only 148G. ? Am I missing something here?

 

The Raptor Random access time is TWICE AS FAST as your array.

 

And again If you'll check read my 1st post in thread you'll see that the HD Tach forums acknowledge that RAID burst speed test is not accurate.

 

Nice run though. Those Intel RAID controllers Rock!

 

But faster - No way Jose :)

Share this post


Link to post
Share on other sites

I think what he's trying to say is...

 

If you look at the graph of a pair of Raptors they sustain reads at about 105MB/sec through their entire capacity.

 

If you look at his graph you will see that for the first 150 GB of the test his sustained reads are higher.

 

Not sure exactly how to interpret these results but on the face of it I thinks he's got a point.

Share this post


Link to post
Share on other sites

Well I certainly accept that fact, and as long as you ONLY use the 1st half of the 400GB array I guess your point has some validity. (Even though HD Tach has trouble measuring RAID arrays).

 

But his ATTO scores look lower than Raptors to me.

 

And their random access speed IS TWICE as long.

 

LOt of folks think the WD Raptor series are overpriced and noisy. Just like many people think a Thermaltake Heat sink is overpriced and a Tornado fan is noisy and the OEM AMD HSF combination is just fine. I just don't agree with either statement.

 

I personally have had nothing but ROTTEN luck with Maxtor drives. But I am also sure this was an aborration and their latest drives are indeed State of the Art.

 

JeffP's rig should make anyone envious. He did an excellant job testing and posting. I know I am coming accross like a FanBoy, but I really do love my Raptors

 

WDraptorslcdc.jpg

Share this post


Link to post
Share on other sites

I've been playing around with this some more and found that changing your stripe size to a small value really seems to help benchmark results. I've run a 16k and a 128k stripe size and find that ATTO shows a result difference of nearly 20-25MB on read and write throughput.

 

Not sure if it's impacting that much in real world performance or not. Of course maybe this was common knowledge and I should of known better. I certainly expected to see a difference but not this much.

 

soundx: Sorry I wasn't very clear with what I was saying in my post. I was trying to point out that referencing at ~148GB point the larger Maxtors were showing better throughput. Although thinking about this after the fact it makes sense they would. Given both drives have the same phsical size platters a larger disk should be gathering more data on a single pass even though it's spinning at a lower rpm. At least that holds true when the head is reading on the outer part of the platter. As it moves inward you can see that advantage dissapears and thats where the Raptors hold their own and then some.

 

Jeff

Share this post


Link to post
Share on other sites

jeff, if you get a chance, try making an array 16k stripe, then loading WinXP off another drive (with your RAID attached tho!) and using WinXP to set the cluster size...if you have a 16k stripe and a 16k cluster then from what I understand, the results are even better (more efficient?).

 

I'm gonna try it when i hook up the SATA II hitachis today

Share this post


Link to post
Share on other sites

I've been looking at the makeup of the system files for a Windows XP Pro install. Judging the average file size and number of files, I can see that you will increase the efficiency of the storage srtucture by going with the 16K clusters with a 64K stripe. This allows for a tighter grouping of the actual data areas with less slack space.

 

Gonna keep looking and learning until I completely understand how all these variables work together.

 

Edit:

Forgot to provide a link to some of the source info I'm using...

 

http://www.storagereview.com/guide2000/ref...perfStripe.html

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
Sign in to follow this  

×