[TechNet]

Spod, on Jan 17 2006, 02:19 PM, said:

TechNet, on Jan 16 2006, 08:22 PM, said:

I reckon the extra drive in my RAID0 array makes a big difference also, which is apparent in Eugene's benchmarks when he tested the WD1500.

Can't work out why WDC restricted the controller interface to just 1.5Gb/s instead of 3.0Gb/s as it does make a difference when the drives are RAID'd

1. For the things that RAID 0 helps with, i.e. STR heavy tasks, then yes, 4 drives are better than two (though often limited by the bandwidth to the SATA controller. As seen above, this STR advantage can easily be matched or exceeded by using a drive with better firmware or a bigger cache when it comes to real world applications.

Some benchmarks give too much importance to STR in the overall scheme of performance. STR benchmarks are the obvious example, giving no consideration to cache, access time or firmware optimisations. That's why SR benchmarks drives using recorded (and thus repeatable) traces of real applications.

As you can see from Eugene's results, if you increase STR while keeping everything else constant, you (usually) get an improvement. By using a single drive with lower STR, similar access times but with major improvements in firmware and cache, the improvement by increasing STR alone is surpassed. That should give you some perspective on the relative importance of STR - it helps, but not that much.

2. The maximum STR of an individual Raptor 150 GB is under 100 MB/s. Since each SATA drive has its own dedicated link to the controller, and that link has a bandwidth of 150 MB/s, the link to the drive will never be a bottleneck no matter how many drives you have.
(As long as you're not using port multipliers to run several drives off one connection to the controller. But you'd know if you were.)

With SATA, it's the bandwidth to the SATA controller that will limit STR from a RAID 0 array - anything not integrated into the chipset itself will be limited to the bandwidth of the PCI bus, usually about 120-130 MB/s.

You were probably thinking of PATA, where two drives might share the same cable (channel), so if you didn't have enough PATA channels free to give each drive its own cable (channel), then two drives had to share the bandwidth of one channel (up to 133 MB/s for ATA133).


So, basically, I stand with Eugene on this one. But I'm glad you challenged him - it's important to question any authority if you suspect that it is misleading you. I hope we've reinforced your trust in SR's benchmarks and opinions, but even if not, thanks for giving us the opportunity to try.*

* I should point out that I don't represent Eugene or SR at all. That last sentence was a bit ambiguous, and I don't want to give the wrong impression.

Hi Spod

It wasn't me who challenged Eugene. Some other guy on here who was comparing his review to one in GamePC.

My comments were almost off topic to be honest and was a discussion on benchmarking with JLN.

[cemal gurel]

Hi again,

I have a typing problem for the Sil0680ACL144 chip. The chip I have on Digitus PATA RAID controller is not from 'Silicon Integrated Systems', it is by 'Silicon Image!' Sorry, for that!

I know the therotical limits of both PATA 133 and SATA 150 and Sata 300. For high bandwith, SATA is good to use on 9xx chipsets by Intel. With 8xx chipset, SATA has NO[b] significant advantage!

I have given a comparison with 1xWDC 1500 Raptor and 2xWDC2500SB by saying 2x expensive but 1,5 times slower!

And also, 1xWDC 1500 Raptor is not faster compared with a basic RAID system with 2x80GB 2x2MB cache by any kind of current market drive, but 4x expensive including the RAID card!

Another problem of so called 10000rpm drives by any kind of manutacturer(s) including SCSI ones, is the diameter of the 'inside disc' problem. WDC1500 Raptor's discs are SMALLER than 7200rpm desktop drive disks! That makes, run over exterior disk surface 'speed on pass' problem, which means, 10000 / 7200 = 1,38 - lower disk diameter = about 1,2 times maximum only actual rotational difference! It causes storage decrease + 1 head more for same storage cost by the manufacturer, (so we are paying for the extra cost!). Why not making them by same 7200rpm disks? are a technical problem. This makes 10000rpm drives roughly 8600rpm drive. Do not forget! Read process starts by the outer side of the disk(s)!

We have reached to the end of PATA interface on 133MB's. NEWER drives can not give SO MUCH Brilliant performance because eventually there is no higer transfer rate than the actual limit of 120MB's by ATA133 controller and this effects average read performance negatively!

I want to add one more thing, WDC1500 Raptor is designed for desktops but NOT for 7/24 critical NEARLINE usage. One can make a RAID chain by WDC1500 Raptor, but the 7/24 drives has bios support for every 7 seconds write check. SB and JB drives has not much phisycal difference actually. I have examined one WDC1200JB with WDC2500SB and only seen, 1 single capacitor extra, on C39 place near the biggest WD70C26 controller chip. So the dependancy work is done by SOFTWARE, rather than the HARDWARE! Probably adding a capacitor to that C39 place and firmware flashing can make a JB, SB drive!

Good luck for all, enjoy spinning drives for speed!

[shoek]

TechNet, on Jan 17 2006, 02:16 PM, said:

It wasn't me who challenged Eugene. Some other guy on here who was comparing his review to one in GamePC.

I hope I didn't come across as challenging Eugene. I respect the work that he's done over the years here at SR, and I was curious to know how he'd respond to another review of this drive where the results seemed to contradict his (specifically in the area of multiple Raptor 74's in RAID0 versus a single Raptor 150). He does appear from his responses to be a bit intolerant of questions like these (or at the very least exasperated by them), so I may be wary to post similar questions in the future...

To summarize, his response was that the benchmarks used in that review are too low-level to draw conclusions about preceived performance, and the benchmarks SR uses model that better so the results are not necessarily contradictory. One drive or array can have higher low-level benchmarks than another, but perceived performance running "real world" applications may actually be lower.

Like most of us here, I have an interest in storage devices and their performance but it is not my full time job and of course I don't have a web site dedicated to it. I try to read as much possible here but missed the saga of IOMeter and Eugene's history with that benchmark. I must have been distracted in November of 2001 to have missed that...

Cheers!
shoek

[JLN]

Eugene (or other folks with knowledge),

Do you think it would be advantageous for me to build my RAID array on smaller data-block sizes than 128KB? Maybe 64?

This would be for single-user performance, working primarily with large files but also using a lot of Office applications.

JLN

[Spod]

JLN: check out SR's stripe size FAQ.

More generally, I'm not saying there's anything wrong with challenging Eugene. Other review sites ought to know better than to use IOMeter for workstation results, and I think that's possibly what may have exasperated Eugene, not the fact that shoek was asking about it. It's in the FAQ list, but most people won't look in the FAQs unless they expect the answer to be there.

It helps to raise the issue now and again, because maybe then some more of the sites that misuse it will see the error of their ways.

[JLN]

Hey, thanks much!

I should have searched the site more before posting the question.

Seems like 128 should be fine.

JLN

Spod, on Jan 20 2006, 04:59 AM, said:

JLN: check out SR's stripe size FAQ.

More generally, I'm not saying there's anything wrong with challenging Eugene. Other review sites ought to know better than to use IOMeter for workstation results, and I think that's possibly what may have exasperated Eugene, not the fact that shoek was asking about it. It's in the FAQ list, but most people won't look in the FAQs unless they expect the answer to be there.

It helps to raise the issue now and again, because maybe then some more of the sites that misuse it will see the error of their ways.

[horst]

shoek, on Jan 17 2006, 04:19 AM, said:

Eugene,
How do you respond to GamePC.com's recent Raptor 150 review, where they show that a dual, triple, or quad RAID0 array of WD740GD's beats a single Raptor. They used the highly respected Areca 1220 PCIexpress card. I assume you used a Silicon Image SATA card? How do you think the nForce4 RAID would fare?



Thanks,
shoek

Well looking with a half cocked eyed I can see that in fact the single 74 beats the single 150,
surely even to the people writing the review, that's got to raise the we've-really-ballsed-this flag

[Alsone]

Eugene would it be possible for you to obtain a copy of Battlefield 2 from someone and do some tests using this program?

I ask because this would be a true gaming test for these drives as BF2 has huge memory requirements - on HIGH settings it requires 2GB of memory to be fitted (one of the few games that requires this - I talk here from personal experience and refer you to this article as well: http://corsairmemory...1GB_vs_2GB.pdf). Loading time for BF2 can be anything from 20 seconds to 5 minutes depending on the machine spec as a result of the huge data transfers that take place.

As a result, I personally would like to see the Raptor 1500 tested with:

1. Battlefield 2

2. SI 3114 Chipset Controller - as this is the one fitted to the Asus A8N Sli line of boards - one of the most common and most popular gaming boards on the market

..and tested with the above vs the 36GB and 74GB Raptor in both single and RAID configs with the 1500 also tested in single and RAID if possible (I appreciate you might not have 2 1500's at the moment).

I think the results would be interesting and a true indication of this drives gaming potential.

Al.

[XeoNoX]

whats a good firmware version to buy if im getting the 74GB Raptor ?

[enderox]

Just another thankyou for a detailed and comprehensive review.

Eugene is a legend! (and NOT just in his own lunchbox)

Reading all the way through this thread enlightened me to the IOMeter workstation pattern issue (yep - I must have skimmed over the sidebar too). One can often gleam so much more on these great technically-orientated forums.

I feel like a bit of a putz for buying a swag of 36g Raptors recently for a server's raid array. I wanted more drives (for raid5 redundancy) and didn't need hordes of capacity. If I'd known these we coming so soon, I might have just looked at raid1 a bit closer. It would be certainly be nice to see more smaller capacity drives on the market - not all of us need multi-hundred megs of raid storage space.

Mike...