DM+9 review finally up at SR
- Thread starter CityK
- Start date
CougTek
Hairy Aussie
Weeks? How about months?CityK said:
Mercutio
Fatwah on Western Digital
Looks like typical Maxtor. Jack of all trades, master of none.
Not to diss SR, but the last few times I went to look up a drive review...they didn't have it. One being this DM 9, the other escapes me at the moment.
Anyone know why it takes them so long to review a drive? Also, does anyone REALLY analyze the data they derive from testing?
Anyone know why it takes them so long to review a drive? Also, does anyone REALLY analyze the data they derive from testing?
Buck
Storage? I am Storage!
I'll analyze some of it, such as the legacy tests and server performance. I'm not sure why they're behind in testing drives. It could be that Eugene is busy making decent money somewhere other then SR.
I was, er, trying to be diplomatic about it.Cougtek said:
The reviews have been a little thin on the analysis side haven't they....which makes me wonder, given that the raw values are placed in the database weeks beforehand, why is it taking so long to pump out the write up.Handruin said:
----
Any thoughts on this:
The Maxtor's possess a faster access and (in two cases) higher areal density then the WD2000JB, yet fail to top them in the SR desktop drivemarks....Can we simply chalk this up to Maxtor drives employing inferior caching strategies? Or do you think there is more to it than that.
CK
Mercutio
Fatwah on Western Digital
Maxtor hasn't ever really been cozy with SR. WD has.
Just something to keep in mind. At least I have since I heard that there was some exploration along those lines.
On the other side of things, Eugene had to in essence do 3 sets of testing for this review, so I guess that mitigates things a bit.
Wish someone could tell me which DM9 I have, though.
Just something to keep in mind. At least I have since I heard that there was some exploration along those lines.
On the other side of things, Eugene had to in essence do 3 sets of testing for this review, so I guess that mitigates things a bit.
Wish someone could tell me which DM9 I have, though.
CityK said:
Well, I do have a thought about the posted review. If you notice near the end of the review there seems to be glitch where they draw values from the reliability database. My guess is that Eugene writes the review, and then gives it to Davin to post into some home made coookie cutter review system.
Maybe since Davin has a full time job (so I assume) maybe this has something to do with reviews taking a longer time to get posted to the web?
zx
Learning Storage Performance
I really don't like the new approach that Maxtor is taking with the DM+9. Of course, I mean that varible platter desity stuff.
That being said, fortunately all three drives have almost the same performance. However, they do not have the same noise level. Maxtor's DM+9 seem a less interesting option if noise is a factor.
That being said, fortunately all three drives have almost the same performance. However, they do not have the same noise level. Maxtor's DM+9 seem a less interesting option if noise is a factor.
e_dawg
Storage Freak
CityK said:
You got most of it, I'd say. SR Desktop DriveMarks and ZDBench testing are greatly affected by caching strategy and short seek performance (which is different from access time measurements, which test seek performance over 1/2 to full length stroke distances). On that second point, HD mfr's can shape the seeks in such a way that short seeks (<5% stoke length) are as fast as possible because they do not generate as much seek noise, but longer seeks (20+% seek length) are slowed and shaped in a way that they generate minimal noise. Since longer seeks are not nearly as frequent as short seeks, they are willing to make this tradeoff in consumer applications. SCSI drives do not make this same tradeoff.
e_dawg
Storage Freak
... the point being that access times are neither representative of real world performance nor are they a good predictor of ZD or SR benchmarking performance because short seek behaviour will often be programmed differently than long seek behaviour.
e_dawg
Storage Freak
... therefore, a drive may have the worst long seek performance around (giving poor access times), but may also be programmed for ultimate short seek performance (giving excellent ZD and SR benchmarks and real world performance).
time
Storage? I am Storage!
Nicely explained, e_dawg.
However, I'm pretty sure that when I'm actually waiting for my hard drive, it's seeking further than 5% stroke. Which is why I don't think the benchmarks are as representative of the real world as they could be.
See, it depends on what you want from benchmarking. Take Photoshop for instance. Benchmarking usually includes a mix of tasks that complete quickly and ones that take a long time. If a task takes 1 second instead of 2, the program will seem snappier, but the effect on the user's productivity is minimal. If you ask a user if they have noticed any improvement in this scenario, most of the time they will either stare blankly or just agree with you to keep you happy. Invariably, it doesn't feel twice as fast to them.
On the other hand, if the task goes from 1 second to 2 seconds, they'll complain. :roll: I digress ...
But what if the thing they find slow about their computer is a task that takes ten seconds, or even two minutes? If you can halve these times, it has a big effect on the user's productivity and psychology. And these are the sorts of things I would like to see benchmarked.
In real life, files are scattered to the four corners of a disk (poetic license), not packed into the first few cylinders where a big cache, smart prediction and fast short seeks rule. I don't believe you can claim real life testing of a 120GB drive with just a few hundred megagbytes.
However, I'm pretty sure that when I'm actually waiting for my hard drive, it's seeking further than 5% stroke. Which is why I don't think the benchmarks are as representative of the real world as they could be.
See, it depends on what you want from benchmarking. Take Photoshop for instance. Benchmarking usually includes a mix of tasks that complete quickly and ones that take a long time. If a task takes 1 second instead of 2, the program will seem snappier, but the effect on the user's productivity is minimal. If you ask a user if they have noticed any improvement in this scenario, most of the time they will either stare blankly or just agree with you to keep you happy. Invariably, it doesn't feel twice as fast to them.
On the other hand, if the task goes from 1 second to 2 seconds, they'll complain. :roll: I digress ...
But what if the thing they find slow about their computer is a task that takes ten seconds, or even two minutes? If you can halve these times, it has a big effect on the user's productivity and psychology. And these are the sorts of things I would like to see benchmarked.
In real life, files are scattered to the four corners of a disk (poetic license), not packed into the first few cylinders where a big cache, smart prediction and fast short seeks rule. I don't believe you can claim real life testing of a 120GB drive with just a few hundred megagbytes.
e_dawg
Storage Freak
Oh I agree with you, time. I did stretch the truth to make a point. While I agree with SR, ZD, Seagate, and others that short seeks predominate, I also find that from personal experience that one needs a drive that can seek quickly at all stroke lengths to feel snappy.
This is why I still prefer using a fast seeking 10k SCSI drive as my main drive than going to a WD800JB, for example. I am glad I didn't, because I have a WD800JB in my second desktop, but it doesn't feel nearly as snappy as my new 10k3 in my primary desktop, even though it has SR benchmark results that are pretty similar.
--------
I feel that I must reiterate your important point about optimizing performance to deliver satisfactory user experience (paraphrased). I abide this variation on the concept:
I have always felt that it is more important to raise the "minimum" speed of your system than trying to increase the "top end" speed. Some of you remember that I was seriously considering a dual CPU system last summer to address those annoying times when a single CPU system is often "frozen" by an app. For example, when Acrobat is capturing and distilling, the computer becomes unresponsive and you pretty much can't do anything else until it finishes.
For my uses, I don't care if my CPU is 2.4 GHz or 1.2 GHz, as my computer wouldn't be any more responsive, nor would it prevent those "frozen moments". I'd rather have dual 800 MHz CPUs where I can guarantee I will be humming along at 800 MHz minimum instead of flying along at 2.4 GHz doing a single task and then dropping down to effectively a 25 MHz 486sx when the CPU is occupied.
For someone else who does DivX encoding, he would be thrilled if he had a 3 GHz P4 that could encode a movie in half the time of his current 1.2 GHz Duron.
This is why I still prefer using a fast seeking 10k SCSI drive as my main drive than going to a WD800JB, for example. I am glad I didn't, because I have a WD800JB in my second desktop, but it doesn't feel nearly as snappy as my new 10k3 in my primary desktop, even though it has SR benchmark results that are pretty similar.
--------
I feel that I must reiterate your important point about optimizing performance to deliver satisfactory user experience (paraphrased). I abide this variation on the concept:
I have always felt that it is more important to raise the "minimum" speed of your system than trying to increase the "top end" speed. Some of you remember that I was seriously considering a dual CPU system last summer to address those annoying times when a single CPU system is often "frozen" by an app. For example, when Acrobat is capturing and distilling, the computer becomes unresponsive and you pretty much can't do anything else until it finishes.
For my uses, I don't care if my CPU is 2.4 GHz or 1.2 GHz, as my computer wouldn't be any more responsive, nor would it prevent those "frozen moments". I'd rather have dual 800 MHz CPUs where I can guarantee I will be humming along at 800 MHz minimum instead of flying along at 2.4 GHz doing a single task and then dropping down to effectively a 25 MHz 486sx when the CPU is occupied.
For someone else who does DivX encoding, he would be thrilled if he had a 3 GHz P4 that could encode a movie in half the time of his current 1.2 GHz Duron.
e_dawg
Storage Freak
... how you optimize your system's performance will depends on what you're using your computer for. This will determine what the limiting factor (bottleneck) is. And IMO, this is what you need to address. If you take care of the bottleneck, you will optimize the user experience.
zx
Learning Storage Performance
Very strange though....The 80gig/platter version of the DM+9 outscores all SCSI drives in both winmarks. What does it mean :
1. winmarks are outdated ?
2. ATA drives are getting very close to SCSI in terms of "desktop" performance ?
3. manufacturers do everything they can do to have good winmarks in ATA drives ?
However, SR's drive benches indicate that while the gap between ATA and SCSI is smaller than before, you still have a performance advantage with SCSI drives, especially 15K drives.
1. winmarks are outdated ?
2. ATA drives are getting very close to SCSI in terms of "desktop" performance ?
3. manufacturers do everything they can do to have good winmarks in ATA drives ?
However, SR's drive benches indicate that while the gap between ATA and SCSI is smaller than before, you still have a performance advantage with SCSI drives, especially 15K drives.
yeti
What is this storage?
ATA drives typically have a much lower command overhead than SCSI drives (they don't queue and reorder read commands). Thus for relatively simple benchmarks, with mostly short seeks, ATA drives stand a good running chance when compared to SCSI ones.
All random IOmeter scores will show that the SCSI drives do much better there, as these are much more seek dependent for a single thread, and also queue management dependent for multiple threads.
All random IOmeter scores will show that the SCSI drives do much better there, as these are much more seek dependent for a single thread, and also queue management dependent for multiple threads.
Pradeep
Storage? I am Storage!
e_dawg said:
Acrobat Distiller will freeze nearly any computer
Howell
Storage? I am Storage!
Pradeep said:Acrobat Distiller will freeze nearly any computer
I have recently run into this problem at work (2G P4, 512MB). Is there any kind of tuning possible? Why does it act differently from PDFwriter? Why would you use each one?
Adcadet
Storage Freak
Back in the days when D&E would talk with me, this is the distinct impression I got. And Davin does have a full time job last I heard - a pretty nice one, too.
Adcadet
Storage Freak
CityK said:
I remember Seagate in a deal with SR, not WD. It may be that WD is better than Maxtor about getting drives to SR, however.
I kindof wonder about that Seagate deal now. Who's their main advertiser now adays?
Mercutio
Fatwah on Western Digital
Seagate had that gigantic quarter-screen flash abomination on SR in early 2002. I think that's what you're remembering.
Adcadet
Storage Freak
Mercutio said:
No, I remember D&E telling me that Seagate was onboard with SR. Not sure if there is still a relation, and to what extent. I've always been pretty good at ignoring ads.
Will Rickards
Storage Is My Life
So I see this review at xbitlabs.
It says the second number in the Serial is the number of heads/surfaces.
I had $70 in gift cert to best buy and they have the 120GB version for $80 after rebates. So I bought it. But I got one that says Y4. There were also ones that say Y3. So I'm trying to figure out which ones better and whether I should return the one I have.
The one I have lists Average Access Time < 9.1ms
The other Y4, lists Average Access Time < 9.4ms
The one I have clearly states on the front label 1 year warranty
The other Y4, does not list warranty information
All I can get at without opening it is the Kit number L01P120 and the Serial number.
It says the second number in the Serial is the number of heads/surfaces.
I had $70 in gift cert to best buy and they have the 120GB version for $80 after rebates. So I bought it. But I got one that says Y4. There were also ones that say Y3. So I'm trying to figure out which ones better and whether I should return the one I have.
The one I have lists Average Access Time < 9.1ms
The other Y4, lists Average Access Time < 9.4ms
The one I have clearly states on the front label 1 year warranty
The other Y4, does not list warranty information
All I can get at without opening it is the Kit number L01P120 and the Serial number.
Will Rickards
Storage Is My Life
Returned the Y4 version and bought the Y3 version.
Manuf Jun 01 2003 and has newer firmware.
I'll try to run some benchmarks and let you guys know how it performs.
Manuf Jun 01 2003 and has newer firmware.
I'll try to run some benchmarks and let you guys know how it performs.
Prof.Wizard
Wannabe Storage Freak
- Joined
- Jan 26, 2002
- Messages
- 1,460
Mercutio said:
Both of you are right.Adcadet said:
I also remember a Seagate technician browsing and posting/responding on threads. I had a thread of mine arguing about the pluses and minuses of having HDD technicians onboard. On the old, lost forums I think...