Smart Drive 2002C + WD Raptor.. Still noisy! Any ideas?

SuitCase874 · Post by **SuitCase874** » Mon Apr 05, 2004 3:20 am

The subject is fairly self explanatory, but I'll try and elaborate. :P

I'm attempting to put together a quiet PC, and have been doing stuff like adding an Arctic Cooling VGA silencer to my Radeon, slowing down the RPM of my P4's HSF, and will soon be replacing my crappy 120mm generic PSU fan with the quietest Papst one I can find.. but one annoying part of my computer has always been the Raptor 36gb drive I bought.

Admittedly buying a 10krpm drive was a pretty stupid move when I was aiming for a quiet computer, but I like the performance of it and I've been working to eliminate both the noises it creates - the rattling vibration when seeking, and the constant whine (I read this was fixed in the 74gb model, etc, but that's not really an option for me..)

Anyhow, I spent $100 on a copper Smart Drive 2002 from Siliconacoustics.com and while I was really impressed by the way it was designed, etc (It's so solid-looking! :P), there is still some whining and if I screw it into one of my drive bays, the vibration is absolutely terrible. I am quite disappointed in its performance, considering the price and the heft of the thing.

Previously I had experimented with temporary solutions such as wrapping the drive in newspaper or foam, and this cancelled out all noise completely. Of course, the drive got to an uncomfortable temperature quite quickly when I did that, but I am surprised that solid layers of foam and copper in the Smart Drive are less effective at removing the whine than a few layers of thin paper. I also tried putting the hard drive in a small box, elaborately cut out with rubber bands in a sort of large suspended enclosure design - this brought the drive to almost uncomfortable temperatures but was very effective in stopping noise.

It seems like the only way to stop the seeking noise safely is by resting it on the bottom of my case (which is filled with anti-vibration pads) and I'm mostly okay doing this - it was the setup I had before. I have found resting it on bubble wrap or something similar makes it even more quiet when seeking and doesn't affect the temperature much.

However, the whining noise persists unless I cover the Smart Drive unit with something sort of thick.. and I'd like to know what materials I can use to safely wrap the drive in, as this noise is very irritating. Currently I have the entire hard drive wrapped in bubble wrap, which does stop the noise, and the drive is running at around 38-43 degrees, which is a fine temperature (max temp is meant to be 55 degrees, I think?), but it seems like a very hacked-together solution to the problem. The bubble wrap is opened at the back to allow more airflow into the drive, etc. If I wrap it any tighter the temperature climbs and climbs and doesn't really stop.

I'm just wondering - does anyone have any simple sort of modification I can do to seal up this noise while not causing the hard drive to melt? It'd be greatly appreciated.

idealcrash · Post by **idealcrash** » Mon Apr 05, 2004 1:30 pm

Does suspending an HD reduce the whine? I don't think so. I have a WD1200JB inside a Smart Drive 2002c and I can still hear it whine. I placed the Smart Drive at the bottom of the case on top of foam and I could still hear it. Looks like the only solution to this is to buy a quieter HD.

slimeballzz · Post by **slimeballzz** » Mon Apr 05, 2004 1:49 pm

idealcrash wrote:Does suspending an HD reduce the whine? I don't think so. I have a WD1200JB inside a Smart Drive 2002c and I can still hear it whine. I placed the Smart Drive at the bottom of the case on top of foam and I could still hear it. Looks like the only solution to this is to buy a quieter HD.

So it's not a good idea to invest on a Smart Drive 2002c eh?

SuitCase874 · Post by **SuitCase874** » Mon Apr 05, 2004 1:54 pm

Yeah, it's not gonna be solved via suspension. If I hold it with my hand it doesn't get any quieter at all.

Slimeball, I'd say that the Smart Drive is pretty nice, for cooling and slight dampening of vibration, anyway. Remember my Raptor is one of the loudest drives you can buy, and it'd be great if you wanted a neat and simple solution to silencing a fairly quiet\"normal" drive already. But I personally don't think it's worth the money, and you're probably better off putting together some sort of dampener yourself - like my current bubble-wrap. It's more effective, and so far it's only been a little bit more warm.

slimeballzz · Post by **slimeballzz** » Mon Apr 05, 2004 2:06 pm

SuitCase874 wrote:Yeah, it's not gonna be solved via suspension. If I hold it with my hand it doesn't get any quieter at all.

Slimeball, I'd say that the Smart Drive is pretty nice, for cooling and slight dampening of vibration, anyway. Remember my Raptor is one of the loudest drives you can buy, and it'd be great if you wanted a neat and simple solution to silencing a fairly quiet"normal" drive already. But I personally don't think it's worth the money, and you're probably better off putting together some sort of dampener yourself - like my current bubble-wrap. It's more effective, and so far it's only been a little bit more warm.

What if you were to put the Raptor into the Silent Drive and then suspend the entire silent drive?

I'm looking for a pair of Silent Drive 2002C for my pair of Hitachi 7K250 80GB drives and plan on either suspending them or putting them at the bottom of the case with some sorbathane underneath them. Do you have any pictures with your bubble wrap method? I may try that though for some reason I'm afraid of the bubble wrap melting or my drive getting too hot.

esn · Post by **esn** » Mon Apr 05, 2004 2:48 pm

slimeballzz wrote:I'm looking for a pair of Silent Drive 2002C for my pair of Hitachi 7K250 80GB drives and plan on either suspending them or putting them at the bottom of the case with some sorbathane underneath them. Do you have any pictures with your bubble wrap method? I may try that though for some reason I'm afraid of the bubble wrap melting or my drive getting too hot.

I've got two Smart Drive 2002 (not C); one with an IBM 30Gb and the other with a Maxtor 80GB. The IBM rests on the insulated case floor and is wedged in place by the foamed case walls. The Maxtor had to be mounted in the lowest 5.25" bay due to space considerations. Just by checking with my finger, the Smart Drives may be running slightly hotter than the system case itself, it's hard to tell; certainly not hot enough to melt bubble wrap. So long as the bubble wrap isn't touching something else that's really toasty I would think you'd have no problem with it. But, now that I think about it, I have done the finger test after prolonged max HDD temps (44C in my case); only at idle (upper 30's). Stil, I wouldn't expect a problem.

Interesting side note: If I gently touch the top of IBM Smart Drive, the HDD will whine noticeably. Can't remember if this happens to the Maxtor also. Other than that, they work great. HDD's are quiet (not silent), running without AMM enabled. The hard mounted Smart Drive does transmit some vibration to the case, but damping material has, for the most part, taken care of that.

wumpus · Post by **wumpus** » Mon Apr 05, 2004 3:50 pm

Admittedly buying a 10krpm drive was a pretty stupid move when I was aiming for a quiet computer

Er, yeah. Want an air raid siren to go with that? Hard drive performance is overrated. Just buy 2-3gb of memory instead. The fastest and quietest hard drive of all.. is a giant disk cache.

So it's not a good idea to invest on a Smart Drive 2002c eh?

I am VERY happy with my Smart Drive-- I have a Maxtor 250gb in there and I can barely hear it now. However, it can't work miracles. 10k drives are way loud.

Trip · Post by **Trip** » Tue Apr 06, 2004 12:51 am

http://www.silentpcreview.com/article22-page1.html

I bought some copper colored aluminum RAM sinks to try on my HDD but they are too tall (1/2")*. The aluminum is made the same way as the blue Zalman NB cooler.

If this worked, perhaps you could then wrap bubble wrap tightly around everything with a stream of air flowing through each heatsinked side of the HDD.

*The reason HS are too tall is I'm using a NoVibesIII. I'm going to look around for some 1/4" tall RAMsinks - real copper this time.

What was your HDD temp. in the 2002C?

this is thought was copper... it's aluminum

SuitCase874 · Post by **SuitCase874** » Fri Apr 09, 2004 8:18 am

I pretty much "solved" my problem, and I'll be selling the Silent Drive. Details on this thread about what I did, if it's of any interest to you Raptor (or otherwise) owners out there..

What if you were to put the Raptor into the Silent Drive and then suspend the entire silent drive?
That'd decrease vibrations (seek noise), certainly, but the high pitched motor whine would remain. That and suspending such a big and heavy lump of metal could be difficult..

I'm looking for a pair of Silent Drive 2002C for my pair of Hitachi 7K250 80GB drives and plan on either suspending them or putting them at the bottom of the case with some sorbathane underneath them. Do you have any pictures with your bubble wrap method? I may try that though for some reason I'm afraid of the bubble wrap melting or my drive getting too hot.
Yeah, I had that concern too. If you look at what I did above, I tried to circumvent those problems in various ways. I suppose you could buy two of them.. they do work, but they're very expensive and simply putting the drives on anything that absorbs sound does far more than the 2002Cs can hope to. I don't personally recommend the drive except for HDs that are quiet already. Or for cooling purposes, they're quite adept at that.

Er, yeah. Want an air raid siren to go with that? Hard drive performance is overrated. Just buy 2-3gb of memory instead. The fastest and quietest hard drive of all.. is a giant disk cache.
Maybe if I was running Oracle or something. :| But I absolutely despise load times for games, programs, windows, anything, so I decided to get something reasonably fast. I suppose RAM can help, but I can't imagine it giving better general performance than a fast hard drive.

However, it can't work miracles. 10k drives are way loud.
But when two layers of thin bubble wrap give way better results..

What was your HDD temp. in the 2002C?
Around 35 degrees idle to 50 degrees peak, I believe. My drive's max recommended operating temperature is 55. The heat dissipated quite quickly, though.

wumpus · Post by **wumpus** » Fri Apr 09, 2004 10:33 am

However, it can't work miracles. 10k drives are way loud.
But when two layers of thin bubble wrap give way better results..

That's a pretty substantial ghetto enclosure you built, with a fan.

But I absolutely despise load times for games, programs, windows, anything, so I decided to get something reasonably fast. I suppose RAM can help, but I can't imagine it giving better general performance than a fast hard drive.

Do you have any benchmarks you can cite to support this "substantial" reduction in load times with the 10k Raptor versus a 7200rpm, 8mb cache IDE drive? Sounds to me like a massive increase in noise and heat for a modest, incremental improvement in load times.

That's kind of a strange decision for a silent box, but to each his own.

edcrane · Post by **edcrane** » Fri Apr 09, 2004 11:18 am

In my experience the answer to WD whine is DIY carving rubber enclosures. The single catch is that the enclosure has to have good thermal contact with the drive (which means thermal paste + tight construction) in order to effectively dissipate the heat. Two drives in this type of enclosure typically require slight active cooling (a 5v L1a is more than sufficient), but the whine is vastly diminished.

Trip · Post by **Trip** » Fri Apr 09, 2004 3:37 pm

SuitCase874 wrote:Er, yeah. Want an air raid siren to go with that? Hard drive performance is overrated. Just buy 2-3gb of memory instead. The fastest and quietest hard drive of all.. is a giant disk cache.
Maybe if I was running Oracle or something. But I absolutely despise load times for games, programs, windows, anything, so I decided to get something reasonably fast. I suppose RAM can help, but I can't imagine it giving better general performance than a fast hard drive.

Cenatek's RAMDisk is pretty fast. If you keep your computer running, you could install a game or whatever on it and load it almost instantly as long as the power isn't shut off. I've used it but don't have that much RAM. I imagine the speed would be comparable to a Rocket Drive.

What was your HDD temp. in the 2002C?
Around 35 degrees idle to 50 degrees peak, I believe. My drive's max recommended operating temperature is 55. The heat dissipated quite quickly, though.

How much airflow was it receiving? This is very good in very poor airflow area and very bad in a very good airflow area. (if the measurement was taken while in a 5.25 slot and you used no extra fans, then these are pretty good temperatures imho)

SuitCase874 · Post by **SuitCase874** » Fri Apr 09, 2004 4:56 pm

That's a pretty substantial ghetto enclosure you built, with a fan.
Indeed, but if you're really ideologically opposed to a fan (compared to the rest of my current components, it's fairly quiet.. I might replace or remove it later when I upgrade my PSU fan and stuff) you can simply not use one. It stood up to a fair while of unrealistic stress-testing without touching the max temperature. And it seems that silent PC people can tolerate that, from the looks of the P4 silent heatsink and similar devices that aren't completely safe.

Do you have any benchmarks you can cite to support this "substantial" reduction in load times with the 10k Raptor versus a 7200rpm, 8mb cache IDE drive? Sounds to me like a massive increase in noise and heat for a modest, incremental improvement in load times.
Maybe it is, but I can live with it. I haven't done any comparisons with 7200rpm drives but from what I can tell it's faster than my older drive and I'm content with that. This computer was never intended to be silent, merely quiet, and I think I'm doing okay so far.

Cenatek's RAMDisk is pretty fast. If you keep your computer running, you could install a game or whatever on it and load it almost instantly as long as the power isn't shut off. I've used it but don't have that much RAM. I imagine the speed would be comparable to a Rocket Drive.
I suppose so. But that sort of thing is very inconvenient to run (and in the case of the Rocket Drive, ludicrously expensive) and I don't repeatedly access just a few gigs of data, most of the time.

How much airflow was it receiving? This is very good in very poor airflow area and very bad in a very good airflow area.
Poor airflow. The closest fan to it was the low-speed VGA silencer fan. I had no case fan directly on it, and I seemed to get similar results when resting on the front part of the case floor and when screwed in below the CD drive. They both had no real airflow.

Trip · Post by **Trip** » Fri Apr 09, 2004 6:06 pm

It seemed like the 2002C could handle the raptor, though it's not rated for it.

I've heard that the HDD is often one of the greatest bottlenecks in a good system and that the raptor is fast enough to almost fully relieve the bottleneck (and to make the bottleneck something else) but I suppose it depends on what you are using the computer for.

wumpus · Post by **wumpus** » Fri Apr 09, 2004 8:51 pm

Alternately, if HDD performance is your bag, baby, some Spinpoints in a Raid array would probably be just as fast if not faster than a 10k drive, with much lower noise and heat levels..

Trip · Post by **Trip** » Fri Apr 09, 2004 9:04 pm

A suspended 740 Raptor should not be too terribly loud nor should it run much hotter than a spinpoint.

01 RAID (minimum 4 drives) is the fastest for us consumers, right? I didn't realise RAID could be that fast! I'll search around for some performance data, this is interesting!

or is 0 RAID faster

dag maybe 1 RAID is, hell I dunno

jimveta · Post by **jimveta** » Sat Apr 10, 2004 12:53 am

for the most part, i think (and has been my experience) a faster hard drive really does make for a substantial increase in overall system performance because it is usually the slowest link/largest bottleneck. for example:
if you were to travel somewhere, and had to drive by car halfway then ride a bicycle for the other half -- and were given the choice of doubling the speed of *either* car or bicycle, which would you choose? suppose originally, you drove the car at 60 mph and the bicyle at 10 mph.

choosing to double the car's speed sounds like a good idea initially right? because now you have the car @ 120 mph and bicycle @ 10 mph -- which sounds like it should get you to your destination faster than choosing to double the bicycle's speed with the car @ 60 mph. and the bicycle @ 20 mph.

this is like with the hd vs. memory example. where memory = car and hd = bicyle. .. well, if you work that example out, increasing the bicyle's speed to a measly 20 mph would actually yeild a *whole lot* more performance than the car's 120 mph. back to hard drives vs memory now, the only way i see that a couple gigs of memory would make a difference is in specialty applications; i think the disk cache argument is only valid if you're just doing reads--and reading a whole lot of the disk at that. even the same rpms but different protocols i.e. scsi vs. ide shows significant improvement see here:
http://hardware.devchannel.org/hardware ... 3249.shtml
same 7200 rpm, in a situation of heavy load, ide w/ much faster cpu = 7 mins, scsi w/ slower cpu = 1 min 10 sec (this is actually not surprising). note that, like the hush atx article showing the synth benchmark that sata is faster than u320, the very simple benchmark using hdparm shows ide to be "faster", see buffered disk reads (mainly because of command overhead, imo). but real world results show otherwise.

.. and with regard to raid, depending on the controller and how raid is implemented -- a lot of cheap "raid" controllers actually implement the raid in software (controllers with raid 5 is usually full hardware raid though)--from hardware to driver/ o.s. .. if raid 1 is implemented, or has the option to perform "geometric" reads, meaning it can spit the requests across drives, then raid 1 is the fastest for reading while raid 0 is fastest for writing. the performance increases with more drives. but raid 1 is actually ok for writing with just a two drive array and a good amount of cache (controller buffer + disk buffer) and no high large write throughput demands.

Trip · Post by **Trip** » Sat Apr 10, 2004 12:58 pm

jimveta wrote:
http://hardware.devchannel.org/hardware ... 3249.shtml
same 7200 rpm, in a situation of heavy load, ide w/ much faster cpu = 7 mins, scsi w/ slower cpu = 1 min 10 sec (this is actually not surprising). note that, like the hush atx article showing the synth benchmark that sata is faster than u320, the very simple benchmark using hdparm shows ide to be "faster", see buffered disk reads (mainly because of command overhead, imo). but real world results show otherwise.

What do you think of this review?

haha, did you read the discussion after the review you linked to?

"And you low life stupid moron will soon be ruled by those Chinese guys. "

"No they won't, I read The Economist and WSJ and they say otherwise. You probably don't read on a 6'th Grade Level"

"You don't read on a 6'th Grade Level - you wrote a run-on sentence"

...

hyperslug · Post by **hyperslug** » Sun Apr 11, 2004 6:41 pm

SuitCase874 wrote:Do you have any benchmarks you can cite to support this "substantial" reduction in load times with the 10k Raptor versus a 7200rpm, 8mb cache IDE drive? Sounds to me like a massive increase in noise and heat for a modest, incremental improvement in load times.
Maybe it is, but I can live with it. I haven't done any comparisons with 7200rpm drives ...

Even better, in an effort to find suitable competitors Storage Review benchmarked it against 15k drives. Desktop results are here: Raptor. Note the bootup test, the IDE Raptor beats the fastest SCSI's.

"Substantial" is an understatement.

dimva · Post by **dimva** » Sun Apr 11, 2004 7:18 pm

hyperslug wrote:
SuitCase874 wrote:Do you have any benchmarks you can cite to support this "substantial" reduction in load times with the 10k Raptor versus a 7200rpm, 8mb cache IDE drive? Sounds to me like a massive increase in noise and heat for a modest, incremental improvement in load times.
Maybe it is, but I can live with it. I haven't done any comparisons with 7200rpm drives ...
Even better, in an effort to find suitable competitors Storage Review benchmarked it against 15k drives. Desktop results are here: Raptor. Note the bootup test, the IDE Raptor beats the fastest SCSI's.

"Substantial" is an understatement.

That's the 74gig one. Look at the 36gig one. Only modest improvements when compared to the hitachi.

So basically, my recommendation would be to get the 74gig one: more space, more speed, less noise.

wumpus · Post by **wumpus** » Sun Apr 11, 2004 7:41 pm

http://tech-report.com/reviews/2004q1/r ... ex.x?pg=13

In my opinion, the 10k drives are a horrendous choice for a quiet PC, even a quiet performance PC. The ratio of performance to noise/heat is way, way off.

If HDD performance is really that important-- which IMO is very debatable except for highly specialized niche applications-- I'd go for an array of 2-4 Spinpoints way before the raptor.

Talz · Post by **Talz** » Sun Apr 11, 2004 9:52 pm

Raptors vary a bit, especially their idle noise. For instance I can't hear mine at idle over my case/cpu fans (all 80mm fans spinning below 1500 rpm). But I acoustipacked my case to get this result, before that I could faintly hear a high pitch whine when everything was at idle. Now I do clearly hear it seek, but hard drive seeks are the one noise I like to hear. Really it bugs me a bit if I can't hear the seeks. Now I have another system built for complete silence, able to run downloading through the night or whatever else without my noticing. And I have to say for my own daily usage the difference between a raptor and barracuda for a main drive were easily noticable. I do have to say the new Samsung in the 2nd system is impressing me though, also noticably faster than the Barracuda, but still noticably slower than the raptor on loads that are clearly HDD limited etc.

My real point though was that idle noise and seek noise vary in importance even to those looking for a quiet pc. Almost everyone will agree loud high pitched idle noises are annoying, aside from a few of the performance nuts that regularly run tornado fans at full speed etc. For seek noise you will get more mixed opinions though. Some want complete silence, especially likely in this forum, many are simply looking to for the best performance they can get without going beyond their personal sound level threshhold for annoyance. I would fall into the last category, with a relatively high threshhold for seek noise since I actually like to use it to keep track of what my system is doing. Though even I would stay away from most SCSI 10K+ drives.

As for disk arrays, I've never been impressed with them outside of RAID 5 or similiar for servers. RAID 0 is just insane for most anything but a scratch disk imo, and RAID 1 has very limited overall performance benefits. Protection against failed hdd's is definately a nice bonus for some uses though. I'd rather stick to straight up backups most of the time though.

Even having said that, my ideal drive would be a raptor that had a mode similiar to the Samsungs with AAM enabled that kicked in after a certain amount of user idle time with the screen saver. Maybe in a few years.

Trip · Post by **Trip** » Sun Apr 11, 2004 10:27 pm

wumpus, that graph shoes the 740raptor as noisier than the 360raptor. Isn't the 740 supposed to be a quieter version of the 360?

I can't stand high pitched whine either.

Talz, http://www.gamepc.com/labs/view_content ... sct&page=8 , RAID 0 performance looks pretty good here - of course, it's just another review. There's probably another out there with contradictory results...

Talz · Post by **Talz** » Tue Apr 13, 2004 4:10 pm

Gamepc.com's testing is fairly limited there but my problem with RAID 0 isn't speed, it's the more than doubling of the chance of a catastrophic failure. I hate hard drive failures. Now RAID 1 decreases this chance but adds very little performance wise. RAID 1 is something I've considered using but the combination of increased costs and noise have held me back so far

The rumoured Matrix RAID on intels upcoming chipset allowing a mix of RAID 1 and 0 with only two drives is much more interesting to me though. If it's in my budget I think I'll likely try this with my next system build.

wumpus · Post by **wumpus** » Tue Apr 13, 2004 7:40 pm

Gamepc.com's testing is fairly limited there but my problem with RAID 0 isn't speed, it's the more than doubling of the chance of a catastrophic failure.

Totally agree. HDD performance, unless you're in one of those crazy niche application areas, is a case of rapidly diminishing returns for a lot of risk. Better to get a decent performing HDD and spend that "extra" money on the fastest CPU, motherboard, video card, and the most memory you can afford.

IMO of course.

Trip · Post by **Trip** » Tue Apr 13, 2004 9:00 pm

When would more than a gig of RAM be helpful?

In your opinion, generally the "HDD is the bottleneck" thinking is wrong?

Even with Dual-Channel DDR400, RAM is still an issue?

wumpus · Post by **wumpus** » Tue Apr 13, 2004 9:13 pm

Well, again, YMMV, but in the real world most users won't hit the disk that often.

Example. Say I load the game Far Cry. Well, I hit the disk once for maybe 1-2 mins tops, then I spend an hour playing the game without touching the HDD. Now, this should go without saying, but my CPU/memory/video are getting a hellacious non-stop workout for that hour.

Ditto for Visual Studio .NET 2003.. once my solution files are in the cache, disk usage is really quite minimal for hours and hours. But I'm compiling like a mother, so CPU/memory is a serious ongoing bottleneck. There were some VS.NET benchmarks on Xbit recently which showed the Athlon64 was almost 50% faster at compiling code than a 3.0ghz northwood! So yeah, integrated memory controller = teh win.

As for >1gb of RAM, you can never be too rich, too thin, or have too much memory. Most of it goes to disk cache. Sure there's a point of diminishing returns, but since the physical disk is about a billion times slower than main memory, even if your 1gb+ disk cache hit rate is .0001 percent, you're still winning.

Not that hard drive performance ISN'T important, it's just that it is very low on the totem pole of relative performance gains for the typical user.

jimveta · Post by **jimveta** » Fri Apr 16, 2004 9:04 am

Trip wrote:
jimveta wrote:
http://hardware.devchannel.org/hardware ... 3249.shtml
same 7200 rpm, in a situation of heavy load, ide w/ much faster cpu = 7 mins, scsi w/ slower cpu = 1 min 10 sec (this is actually not surprising). note that, like the hush atx article showing the synth benchmark that sata is faster than u320, the very simple benchmark using hdparm shows ide to be "faster", see buffered disk reads (mainly because of command overhead, imo). but real world results show otherwise.
What do you think of this review?

haha, did you read the discussion after the review you linked to?

"And you low life stupid moron will soon be ruled by those Chinese guys. "

"No they won't, I read The Economist and WSJ and they say otherwise. You probably don't read on a 6'th Grade Level"

"You don't read on a 6'th Grade Level - you wrote a run-on sentence"

...

i think that storagereview article is actually a valid review. however, unfortunately like most pc storage articles, not much emphasis is given for the IOmeter benchmarks, nor realworld app benchmarks. but at least they do IOmeter benchmarks. .. so, yes, the sata drive is fast for bootup and simple i/o. however, this page in the same article:
http://storagereview.com/articles/20040 ... 0GD_4.html
shows ALL the ide drives at the bottom of the list -- and illustrates the reason why the hardware.devchannel.org article got the results it did (i.e. 7 mins for IDE, 1 min 10 sec for SCSI) even though hdparm benchmarks in the same hardware.devchannel.org article showed ide having higher marks (49 mb/s for ide vs. 29 mb/s for scsi) but this is again more the nature of the benchmark being limited to how it's applicable in various situations.

here's more somewhat anecdotal evidence, from the book "PC hardware in a Nutshell" published by O'Reilly:

"We'll devote less space to SCSI than IDE because IDE drives dominate the PC platform, but we will try to hit the high points of SCSI."

".. ATA drives, whether PATA or SATA, simply cannot compare to SCSI drives in performance under heavy load (although ATA drives may actually be a bit faster under light load because their simpler protocols impose minimum overhead). In our real-world testing, under very heavy disk access, the slowest SCSI drives we used were faster than the fastest ATA drives, particularly under Windows NT/2000/XP, Linux, and other multitasking operating systems. ..."
" .. Although ATA may actually beat SCSI under light load, when disk activity starts to climb, SCSI is simply faster. Don't let anyone convince you otherwiese."

"To verify our impression of SCSI versus ATA, we did an experiment in mid-2002. At that time, Barbara's main workstation used a 7,200 RPM SCSI Seagate Barracuda drive. We built an identical system, but substituted a 7,200 RPM Seagate Barracuda ATA IV drive. During normal operation, performance of the two PC's was indistinguishable.

We then started an XCOPY operation that streamed gigabytes of data comprising hundreds of directories and thousands of files from a third system across our 100BaseT network to the hard drive of the ATA system. While that data was being copied,the ATA system was nearly unusable. Loading Word from the hard drive took literally a full minute, and opening a large document took even longer. We then repeated the experiment, but this time to Barbara's SCSI Barracuda. The drive banged away, certainly, but we were able to load programs and run things normally with very little performance degridation.

All this doesn't mean that the Seagate Barracuda ATA IV was a bad drive. It wasn't. In fact, it was one of the best ATA drives available at the time. But it does establish that ATA bogs down under load, whereas SCSI just keeps on ticking."

(Note that the hardware.devchannel.org article situation is similar, but using Linux and a maildir format comprising of 50,000 folders and files needed to be read, parsed and sorted; this is the kind of stuff IOmeter can test that other benchmarks won't show)

(Now with regards to comparing HD performance with other parts of the system)

".. One of our friends, who'd just spent a bundle on the fastest Pentium 4 system available at the time, made the mistake of sitting down in front of one of our systems that has a 15,000 RPM Seagate Cheetah X15 SCSI drive installed. He fired up Word, turned to Robert, and asked if this was a dual-processor system. Robert told him it wasn't and asked what made him think it was. He said everything just flew up onto the screen as soon as he double-clicked the icon, much faster than on his new system. When Robert told him that the system used a Celeron processor, the conversation became a bit strained. Robert finally stopped teasing him and explained that the system had an Adaptec Ultra160 SCSI host adapter and a 15,000 RPM Seagate Cheetah X15 drive. Kind of like one of those undercover police cars that looks like a junkyard reject but has a 500HP engine."

.. Trip: now with regards to the reader comments in that article, yeah i agree that most were idiotic .. i didn't remember those you pointed out, but i stopped reading early on when some doubting readers started asking about defragging, cluster size and using ghost to get better results (on linux).

wumpus wrote:Ditto for Visual Studio .NET 2003.. once my solution files are in the cache, disk usage is really quite minimal for hours and hours. But I'm compiling like a mother, so CPU/memory is a serious ongoing bottleneck. There were some VS.NET benchmarks on Xbit recently which showed the Athlon64 was almost 50% faster at compiling code than a 3.0ghz northwood! So yeah, integrated memory controller = teh win.

i will also preface by saying ymmv, but (see above snippet from book) And from my own experiences at work on my system going from a 7200 rpm drive to 15k made a BIG difference. secondly, the disk performance also makes a substantial difference when doing large, multithreaded compiles (at work we use large machines with a combination of software and hw raid, e.g. 11 cpu, 45 GB mem, large fiber channel scsi arrays; and since the perf is typically held back by the disk, we're very keen on making it as cpu bound as possible; also i don't think we'll be moving to SATA anytime soon

)
in fact, i may do a comparison between a 10k 36 gb raptor and various scsi drives, just to see for myself when ide is faster and when scsi is faster. lastly perhaps the amd64 is faster in compiling more because of its much more branch friendly architecture than the P4 as the p4 does win in most memory benchmarks.

jimveta · Post by **jimveta** » Fri Apr 16, 2004 9:17 am

Talz wrote:Gamepc.com's testing is fairly limited there but my problem with RAID 0 isn't speed, it's the more than doubling of the chance of a catastrophic failure. I hate hard drive failures. Now RAID 1 decreases this chance but adds very little performance wise. RAID 1 is something I've considered using but the combination of increased costs and noise have held me back so far

The rumoured Matrix RAID on intels upcoming chipset allowing a mix of RAID 1 and 0 with only two drives is much more interesting to me though. If it's in my budget I think I'll likely try this with my next system build.

.. i'm with you here too. but how about having seperate arrays like 2 disks for raid 1 and another pair for raid 0? the mirrored array can be used for "source" files and the striped array can be used for temporary or work files. that way each raid array is used for its strong points (and raid 1 should be as fast, if not faster, than raid 0 for reads). .. i do know the dilemna of noise though.. i wonder how the various external sata enclosures are...

Smart Drive 2002C + WD Raptor.. Still noisy! Any ideas?

Smart Drive 2002C + WD Raptor.. Still noisy! Any ideas?

on hard drive performance