Davin

Western Digital Raptor WD360GD

Recommended Posts

I think people here forget about some basic facts...

First, SCSI and IDE are interfaces. They are very different and you cannot expect one to be better than the other for EVERY use. In terms of raw speed, SCSI is far superior, but raw speed isn't everything, imagine a F1 or a Formula Indy on a rally course...

Second, when you use SCSI RAID arrays, you usually do it on motherboards supporting PCI 66Mhz 64bit, not the PCI 33Mhz 32bit... this makes a huge advantage for SCSI drives in terms of bandwidth ... and ... performance.

Third, as someone said it before, it is not all about mechanics when it comes to real performance. Think about the firmware, the cache... even if you have a identical amounts of cache on a drive, the way the cache is implemented affects performance...

So, leave SCSI drives to the SCSI world (servers ... some workstations) and IDE drives to the IDE world (desktop ... some workstations). I'm saying this because in the past I made some implementations drives which were clearly single-user desktop oriented with SCSI drives. That was a mistake...

I have 4 raptors on a high end desktop system, I must say they are great for this type of environment. For what I've seen on the review... forget it for serious server use...

Share this post


Link to post
Share on other sites

Hmm...were the benchmarks conducted in this review on a FAT32 or NTFS file format? I'm curious because from another Raptor review I read at Xbit Labs, the benchmarks conducted on NTFS was considerably slower than in FAT32...I'm using NTFS and when I read StorageReview's review on the Raptor, I was puzzled as to why my benches seemed slower.

Share this post


Link to post
Share on other sites
Hmm...were the benchmarks conducted in this review on a FAT32 or NTFS file format?

Surprisingly difficult to answer (in the absence of a reply from Eugene!), but a quick scan through SR's How We Test article reveals that the Desktop tests, at least, were done on NTFS. I believe that some of the other tests are done on unpartitioned disks, but if you want the full gory details, get them first hand.

It's more likely that your results are slower because of different hardware or testing settings, IMO.

Spod

Share this post


Link to post
Share on other sites
Hmm...were the benchmarks conducted in this review on a FAT32 or NTFS file format?

It's more likely that your results are slower because of different hardware or testing settings, IMO.

Spod

The fact is that almost everyone who purchased a Raptor can replicate the preview Raptor transfer rate graphs. I have heard of no one who was able to replicate the "production" review performance that SR saw.

This really needs more attention. If WD sent our "review units" that made the Raptor look like an excellent performer, but is producing and selling drives with the lackluster performance from the previews then there's a problem.

B

Share this post


Link to post
Share on other sites
Hmm...were the benchmarks conducted in this review on a FAT32 or NTFS file format?

It's more likely that your results are slower because of different hardware or testing settings, IMO.

Spod

The fact is that almost everyone who purchased a Raptor can replicate the preview Raptor transfer rate graphs. I have heard of no one who was able to replicate the "production" review performance that SR saw.

This really needs more attention. If WD sent our "review units" that made the Raptor look like an excellent performer, but is producing and selling drives with the lackluster performance from the previews then there's a problem.

B

Almost everyone? I've only seen a couple people affected here at SR, but plenty who have the expected performance.

Share this post


Link to post
Share on other sites
Hmm...were the benchmarks conducted in this review on a FAT32 or NTFS file format?

It's more likely that your results are slower because of different hardware or testing settings, IMO.

Spod

The fact is that almost everyone who purchased a Raptor can replicate the preview Raptor transfer rate graphs. I have heard of no one who was able to replicate the "production" review performance that SR saw.

This really needs more attention. If WD sent our "review units" that made the Raptor look like an excellent performer, but is producing and selling drives with the lackluster performance from the previews then there's a problem.

B

Almost everyone? I've only seen a couple people affected here at SR, but plenty who have the expected performance.

I haven't seen anyone come forward to say that they tested their reads and found normal performance. Where are the folks who have proper performance?

Or is it just an assumption that they haven't posted so their transfer rates are fine? Heck, I figured mine were fine b/c it was snappy in RAID 0. Once I decided to break the RAID array I tested them and they didn't perform as exected. Many won't even bother to check.

I'm being completely serious too. If there are people out there who have the performance of the SR production review then I need to get 2 drives replaced.

Thanks

B

Share this post


Link to post
Share on other sites

Well, I'm in that group that probably won't check....just don't have the time.

Share this post


Link to post
Share on other sites

I live in Australia and one of our own review sites has tested the WD Raptor and found different results as those posted by SR's Retail review. They were only able to match the Beta drives results and this is with production drives. They have confirmed this with many other Raptor owners, in the review they actually test 4 at once. This is a link to the second part of the review where it's tested in Raid 0 and address speed mystery. The review can be found here http://www.overclockers.com.au/article.php?id=179581.

Their results have put me off purchasing the drive since i was only going to purchase one if it could match the results of SR's Retail sample. I see no point in buying the drive now if it's only going to give me better access times over todays leading PATA drives, at the disadvantage of much less data capacity and on par performance.

Looks like WD tossed a magic firmware in SR's direction to me.

Share this post


Link to post
Share on other sites

Swift1-

It is interesting that the only website to review the Malaysian Raptors is this AU site, and they also get the low performance.

It would be nice if we could confirm whether all the retail packaged Raptors are USA made vs. Malaysian.

That could be the performance difference, rather than just cherry picked review drives.

B

Share this post


Link to post
Share on other sites

solid state drives...... need anyone say more?

the raptor drive is the best drive for the money right now.

solid state= $10 per meg...ata= $0.10 per meg...scsi= $1.00 per meg

i wish i was writing from a missle silo with google's of 19" rackmounted solid state drives in a raid array and clusters of 19" rackmounted

64 bit multi processor servers....but i am not...everyone has dreams, the question is can you afford to make dreams reality.

i will buy the raptor and keep dreaming.

Share this post


Link to post
Share on other sites
Guest Eugene
The fact is that almost everyone who purchased a Raptor can replicate the preview Raptor transfer rate graphs.  I have heard of no one who was able to replicate the "production" review performance that SR saw. 

This really needs more attention.  If WD sent our "review units" that made the Raptor look like an excellent performer, but is producing and selling drives with the lackluster performance from the previews then there's a problem.

B

We'll have a front-page update on this issue shortly. Basically, Western Digital was experimenting with various zoning configurations on the 36 GB platters to achieve an optimimum balance between transfer rates and production yields (i.e., cost). In the end, the drives that hit the channels featured zoning that was equivalent to our first sample.

That said, there's a flaw in the logic of your concern:

SR's first sample featured (lets call it) zone config A AND had prerelease firmware AND had write caching disabled.

SR's second sample featured zone config B AND had final firmware AND had write caching enalbed.

Drives that hit the channel mostly have zone config A. Because they have "A," does it also necssarily follow that they're equipped with prerelease firmware and disabled write-caching? No.

Does a reduction in STR translate really translate into into a proportional reduction in application-level performance? My take on this is well known and sprinkled throughout the site.

I've been testing a retail sample provided by HyperMicro over the weekend. We'll have some firm numbers to present soon.

Regards,

Eugene

Share this post


Link to post
Share on other sites

Thank you Eugene for listening to the readers and taking this to heart. I completely understand their decision, and agree that the performance difference to the average user should be much less significant than the STR measurements indicate. I'm looking forward to seeing this issue put to bed either way.

Share this post


Link to post
Share on other sites
Guest Eugene

Here's the "official" line from Western Digital:

"Reliability is a leading criterion for Western Digital’s enterprise-class WD Raptor product line, as evidenced by its 1.2 million hours MTBF and 5-year warranty. StorageReview.com’s initial review of WD Raptor was performed using beta samples, which, in addition to its disabled write cache, did not reflect the high volume format of first generation WD Raptor hard drives. Fine-tuning of WD Raptor hard drives has resulted in a format that boosts higher long-term reliability with a slight performance deviation from early production WD Raptor hard drives. This balance of long-term reliability and overall performance makes WD Raptor a true enterprise class drive and the highest performing ATA hard drive available on the market.

  Since the time of the reviews of both the beta and the subsequent Western Digital-provided WD Raptor samples (to StorageReview.com), the majority of WD Raptor drives shipped into the channel have been formatted to deliver maximum long term reliability along with the optimum overall performance available today in enterprise Serial ATA hard drives. The high yields Western Digital has achieved in its manufacturing of WD Raptor hard drives are typically considered a key indicator of long-term reliability -- the foundation of the WD Raptor enterprise Serial ATA product family.” - Ted Deffenbaugh, senior director of product marketing at Western Digital

I have complete SR results for the retail sample provided by HyperMicro. A little while ago, though, we had to update our SATA controller's bios and driver to iron out an STR problem that we encountered in conjunction with testing WD's Caviar WD2500JD. Hence, a retest on all previously-reviewed ATA drives was warranted. The "reset" article is complete, pending publication. As soon as we get it out, I'll be able to sneak the latest Raptor figures into the database.

Thanks,

Eugene

Share this post


Link to post
Share on other sites

Hmm. I read that as, "the model we submitted for review was tuned for high performance at the expense of reliability; the model you'll get is not." Granted, it's put much more diplomatically, but it reads a lot like an AMD or Intel submitting an overclocked processor for a review, then underclocking in the retail channel for improved stability.

I'm a bit concerned only because WD is putting SR's words of support onto the side of its flashy color retail boxes, fully attribution and all.

Of course, I'm quite happy with both of my Raptors - a fantastic drive by any means.

Share this post


Link to post
Share on other sites
Hmm.  I read that as, "the model we submitted for review was tuned for high performance at the expense of reliability; the model you'll get is not."  Granted, it's put much more diplomatically, but it reads a lot like an AMD or Intel submitting an overclocked processor for a review, then underclocking in the retail channel for improved stability.

I'm a bit concerned only because WD is putting SR's words of support onto the side of its flashy color retail boxes, fully attribution and all.

Of course, I'm quite happy with both of my Raptors - a fantastic drive by any means.

I agree with you, but conditionally. It comes down to the results we are waiting for on the Hypermicro supplied unit. If the reduced STR proves to have noticable negative impact on the results then I could understand, and even agree with, the belief that WD have intentionally supplied a unit with higher performance than can be expected from a purchased unit.

Share this post


Link to post
Share on other sites

Somewhat mixed results. I'd say they've fallen somewhere in between the expected 'minimal' impact and possible 'significant' impact.

A drop in STR of around 9% has caused a drop in the Drivemarks of roughly 4-5% in most cases. But over 11% in the Bootup Drivemarks? Can anyone offer any explanations for that?

Eugene, I'd like to clarify if the standard Raptor entry is for the results with the updated BIOS and drivers?

Share this post


Link to post
Share on other sites
it has different firmware as well, and that is what eugene has attributed the app-level perf drop to.

Really? I didn't see that. I only saw the info on the updated SATA controller firmware.

Share this post


Link to post
Share on other sites

http://forums.storagereview.net/index.php?...showtopic=11424

Keep in mind that this HM sample features both a different zone configuration (which affects STR) as well as a newer firmware revision. As exemplified in the SATA update article, newer software doesn't always mean faster. STR alone wouldn't result in these measurable decreases; its highly likely that the newer firmware is also more conservative in its caching strategies.[/quote

Share this post


Link to post
Share on other sites

OK, that said...would I notice that much difference from two large seagate SATA drives in a RAID vs. two raptors if I just played games and did some photo/video editing? THANKS!

Share this post


Link to post
Share on other sites
http://forums.storagereview.net/index.php?...showtopic=11424
Keep in mind that this HM sample features both a different zone configuration (which affects STR) as well as a newer firmware revision. As exemplified in the SATA update article, newer software doesn't always mean faster. STR alone wouldn't result in these measurable decreases; its highly likely that the newer firmware is also more conservative in its caching strategies.

I've had a few days to ponder this, and came up with the following conclusion.

It's seems reasonable to expect a good correlation between the drop in STR and Bootup Drivemarks. So a reduction of 9% in STR can mostly be applied to the reduction of 11% in Bootup Drivemarks, plus maybe 2-3% lost due to firmware. This seems reasonable, if slightly undesirable.

Applying that 2-3% loss due to firmware to the other results, leaves the remaining 2-3% to be attributed to the reduced STR of 9%. This seems in line with the expectations expressed by Eugene and others.

Share this post


Link to post
Share on other sites
OK, that said...would I notice that much difference from two large seagate SATA drives in a RAID vs. two raptors if I just played games and did some photo/video editing?  THANKS!

There's a good chance you'd notice the difference between a single Seagate and a single Raptor (although you didn't specify which Seagate). This may not extrapolate equally to the RAID scenario, particularly considering that other factors such as the RAID controller play a part.

Share this post


Link to post
Share on other sites

Hi,

Can i just ask if the WD Raptor situation has been clarified, and if it has, what the outcome is.

If it turns out that WD did indeed submit 'souped up' drives for testing that are not available through retail channels, should that benchmark not be removed from the database?

Chrs

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