Sign in to follow this  
davidedney123

Tom's been at his crack pipe again

Recommended Posts

Yeah, that's just wacky. Just get a pair of Samsung drives or even a couple of Seagate U series - low noise and heat and worlds faster than those notebook drives.

Seeing this got me thinking though... I'm sure some people here have seen those RAID enclosures that take IDE drives but have a SCSI interface. If one was made to specifically take notebook drives you could fit 10 of them in a full height 5¼" bay, imagine 600GB using the IBM Travelstar 60.... 8)

Share this post


Link to post
Share on other sites

Well, the test itself isn't without interest. Trying to design a completely noiseless computer, the tips provided by Tomshardware might actually come in handy.

I cannot agree more with you about the comments he provides. Especially this one comes to mind:

"The benchmark results are quite clear: only two 2.5" drives are able to outperform a modern desktop drive in terms of transfer performance - and without the high temperatures and obnoxious noise!"

Obnoxious noise? From the X15-36LP? Even my Atlas 10K-III isn't that horrible, the sound can be annoying, but if you move your ear atleast 1m from the drive, I don't think anyone would call the drive-noise obnoxious...

As usual with Tom's, his reviews or "tests" as I would call them, springs from good ideas but ends with bad results because of incompetent people writing the reviews. Perhaps the lack of time to write the "tests" are the main factor in degrading the quality of them, I cannot know for sure - but it better be time-frames destroying the reviews. I cannot think of any other reason to publish reviews as flawed as the one we saw today :(

-Uffe

Share this post


Link to post
Share on other sites

Tom is not the author of that article, as noted on every page of the review, so claiming he is on the crack pipe is a bit misdirected. Might want to reevaluate who really is here.

Share this post


Link to post
Share on other sites

He might not have written it, but the site is his and the content that he allows on it reflects on him. You'd have to take a great big hit on that pipe to allow crap like that on your site!

Share this post


Link to post
Share on other sites

Putting two notebook drives in RAID 0 might seem like a good idea, but considering that they are slower than a Cuda IV, costs more, has a higher chance of failure and the Cuda is pretty much silent as is, there really doesn't seem any reason for it.

Share this post


Link to post
Share on other sites

Nomatter if the article is well done or not, I hope it will increase people's interest in 2.5" drives. If the standard size of our hard drives would be reduced to 2.5" instead of the current 3.5", we could have much smaller enclosures.

I really hope hard drive manufacturers will push the performances of these smaller drives. The new Toshiba 40GB hard drive with a 16MB buffer is a step in the right direction. If this drive ends up being a marketing success, maybe other manufacturers will follow. With the very high platter density found in today's IDE drives, capacity is no longer an issue against the use of 2.5" hard drives in mainstream computer systems. 20GB to 40GB is enough for most people. According to the article at THG, the transfer rates, while not matching those of modern 7200rpm 3.5" IDE drives, were still acceptable. Rest to shave the access time. The challenge must be to fit fast actuators into such a small design while maintaining an acceptable cost. I have high hope they will eventually find a way to achieve this.

Anyway, I'll follow the developments of 2.5" hard drives closely.

Share this post


Link to post
Share on other sites

Personally, I think the results Patrick came up with in that review speak for themselves... and how he came to the conclusions he did I will never know. Allow me to rant :)

First - he suggests that a RAID array of 2.5" notebook drives can rival the performance of a desktop drive - when clearly his results show the array lagging behind the IBM he used as a comparison in both access time and overall benchmark performance. In fact, the lack of discussion on drive access time is scandalous. Anyone who has tried and tested a fair amount of hardware will realise that access time is everything when it comes to choosing your main system drive. No matter if you had 20 drives in that RAID array, the access time would still be pants, and Windows would crunch and grind just the same.

Second - he gets excited about the low power consumption of the drives. Now, whilst this is true, he fails to realise that he had to use two 15w drives to get somewhere near the performance of one 35w drive.

Third - laptop drives *do* make some noise. Not a lot, but not that much more than a quiet desktop drive in the first place. And you get double the noise if you have to fit two drives to get any decent performance.

Fourth - he fails to mention the additional risk involved of running two drives in a RAID-0 configuration. Twice the drives = twice the risk of failure.

And to top it all off, his concluding words are bizarre. It's as if he wrote the conclusion first, full of anticipation and optimism, then ran the tests and decided to say it anyway. The only thing he got vaguely right was his price comparison. Poor old Tom will need a crack on his pipe when he reads that article.

Share this post


Link to post
Share on other sites
Nomatter if the article is well done or not, I hope it will increase people's interest in 2.5" drives.  If the standard size of our hard drives would be reduced to 2.5" instead of the current 3.5", we could have much smaller enclosures.

...

With the very high platter density found in today's IDE drives, capacity is no longer an issue against the use of 2.5" hard drives in mainstream computer systems.  20GB to 40GB is enough for most people.

...

First, I want to say I agree that 20GB to 40GB is enough for most people. It is more than enough space for all the apps and junk people put on their machines. I think the problem is when you get into things like audio/video work and when you consider performance.

I think it is important to remember that the disk capacity is not linearly proportional to the diameter of the platter. The surface area of a 2.5" platter is 4.9 sq. in. The surface area of a 3.5" platter is 9.6 sq. in. - almost twice that of the 2.5" platter (ignoring the area used by the spindle, which would only magnify the impact).

The reduced platter surface has two impacts:

1. Lower capacity - are you willing to lose half the space to go to 2.5" form factor?

2. Lower performance - at the same rpm, more area passes under the drive heads on the outer edges (hence the better performance at the "begining" of disks). Are you willing to give up this performance?

Personally (what other way is there???) I would not be willing to give up the space and performance. Audio/video files require a lot of storage. Video capture requires a certain sustained performance. Application loading also benefits from the improved performance.

I think the difference in noise and heat does not justify the lower performance and capacity.

2.5" drives may be fine for most people, but I would think that you and I (like most people here) will not find this trade-off acceptable.

YK

Share this post


Link to post
Share on other sites
Nomatter if the article is well done or not, I hope it will increase people's interest in 2.5" drives.  If the standard size of our hard drives would be reduced to 2.5" instead of the current 3.5", we could have much smaller enclosures.

I really hope hard drive manufacturers will push the performances of these smaller drives.  The new Toshiba 40GB hard drive with a 16MB buffer is a step in the right direction.  If this drive ends up being a marketing success, maybe other manufacturers will follow.  With the very high platter density found in today's IDE drives, capacity is no longer an issue against the use of 2.5" hard drives in mainstream computer systems.  20GB to 40GB is enough for most people.  According to the article at THG, the transfer rates, while not matching those of modern 7200rpm 3.5" IDE drives, were still acceptable.  Rest to shave the access time.  The challenge must be to fit fast actuators into such a small design while maintaining an acceptable cost.  I have high hope they will eventually find a way to achieve this.

Anyway, I'll follow the developments of 2.5" hard drives closely.

For mechanical storage, the 2.5" does sacrifice a number of points, as was previously mentioned in this thread. :(

Personally, I think our best bet for the future is FRAM, MRAM, or some other form of solid-state storage. 8)

I just wish it happen sooner than later, with a reasonable price. :roll:

Share this post


Link to post
Share on other sites
Personally, I think our best bet for the future is FRAM, MRAM, or some other form of solid-state storage. 8)  

I just wish it happen sooner than later, with a reasonable price. :roll:

I am still waiting for bubble memory to replace the 5MB Profiler interfaced to my Apple.

I have been waiting almost twenty years now.

Count on later, rather than sooner.

Share this post


Link to post
Share on other sites

First - he suggests that a RAID array of 2.5" notebook drives can rival the performance of a desktop drive - when clearly his results show the array lagging behind the IBM he used as a comparison in both access time and overall benchmark performance."

One should not expect 4200RPM laptop drives to perform as well as the speed king of IDE drives from a year ago, RAID'd or not. They're not wrong in saying the the drives performed comparably to a desktop drive, they just pick the wrong one to compare it to. Pick any 5400RPM drive and the results will be quite similar. Those Toshiba drives aren't exactly the fastest on the market either, 5400RPM laptop drives are out on the market, including 16MB cache version about to be released by Toshiba.

"In fact, the lack of discussion on drive access time is scandalous. Anyone who has tried and tested a fair amount of hardware will realise that access time is everything when it comes to choosing your main system drive."

There is a section named about access time, and their analysis of it is spot on as well as the definition. Access time is important, but is pretty much irrelevant in the IDE market where it has stayed constant for 3 years or so. Look at the WDJB drives vs the Maxtor D740X. The Maxtor has a 1ms lower access time and is soundly beaten across the board. Even the regular BB version without the cache advantage beats it across the board. Access time is only part of the equation, and can be offset by other performance characteristics. Until a 10k IDE drive is released, access time will remain an afterthought in the market, as it doesn't really affect performance that much.

"Now, whilst this is true, he fails to realise that he had to use two 15w drives to get somewhere near the performance of one 35w drive."

This scenario only occurs at bootup, which often can be staggered, splitting up the requirements, and I'm guessing these numbers are pulled out of thin air anyway, and not completely accurate. Read/write usage for the laptop drives is a measly 5W combined, putting it at half of a standard IDE drive.

"Fourth - he fails to mention the additional risk involved of running two drives in a RAID-0 configuration. Twice the drives = twice the risk of failure."

If you have any intention of actually duplicating this experiment you better already know the basics of RAID.

This article does have real world applications that can be very useful. An example would be the rash of home theater PC's that are appearing. The smaller the case the easier it is to place, these drives can be split up and squeezed into any hole they will physically fit in without worrying about cooling or anything, unlike a desktop IDE drive which is considerably larger to begin with. Reduced power consumption means smaller PS, and cooler operation which means a fanless design is definitely possible which will make a system much quieter on top of the drives themselves being quieter. The benefits of a quiet PC in a HT setup is obvious. There is no need for cutting edge performance in a system like this, nor is a RAID failure of much consequence in a system that is unlikely to have anything mission critical on it.

This also ignores the obvious application of RAID'ing 2 laptop drives within a laptop. Using win2k/XP a software stripe can easily be created that would give very good performance for a disk subsystem in the mobile market.

You're nitpicking the article to try and make someone look bad. To some people this article will be very informative and maybe give them an option they hadn't thought of to fulfill the needs of something they have been trying to do. Just because you have no use for the information, doesn't mean that everyone else should be deprived of it.

Share this post


Link to post
Share on other sites

Could any one here tells the difference between a hardware RAID 1 with two disk of each having one platter; and a single HDD with 2 platters, other being equal. Isnt it a more effective and efficeient way to have more platters in a single drive. :?:

Share this post


Link to post
Share on other sites

RAID 1 is mirroring, the point of which is to copy the data to two physical drives in case one fails. When one fails, you remove the dead one and put a new, blank drive in it's place. You can then instruct the RAID controller to copy the data from the remaining good drive to the new blank drive, and you're back in business, ready to rebound from another failure at any time. Your single drive does not have this feature, so there really is no basis for comparison. Could a person tell the difference between a single drive and a mirrored set just by using the system? Probably not, unless they noticed two drive LEDs flashing simultaneously :)

Share this post


Link to post
Share on other sites
This also ignores the obvious application of RAID'ing 2 laptop drives within a laptop. Using win2k/XP a software stripe can easily be created that would give very good performance for a disk subsystem in the mobile market.

I am not sure how effective this would be as it would require three drives in a laptop to achieve it. You cannot software mirror your boot drive.

Look at the WDJB drives vs the Maxtor D740X. The Maxtor has a 1ms lower access time and is soundly beaten across the board. Even the regular BB version without the cache advantage beats it across the board. Access time is only part of the equation, and can be offset by other performance characteristics. Until a 10k IDE drive is released, access time will remain an afterthought in the market, as it doesn't really affect performance that much.

That would depend on how you describe "across the board" In all fairness if the maxtor drives have a 1ms faster access time over the WD, then it has won one of the competitions, thus making an across the board "sweep" as you claim by the WD impossible. Your right about access only being one part of the equation, but then again so is STR only one part as well. And here I disagree with you. Most users will notice faster access times more so then they would a higher STR.

The review does have a few obvious flaws, take for example this quote:

"The access time of the RAID 0 array is a bit quicker than that of a single notebook drive. Still, it is a far cry from the excellent results of the DeskStar 60GXP."

Somebody want to explain this too me? I thinking the reviewer probably should run the tests a few more times to average his results? Raid on any level cannot improve the access times of the slowest drive in the chain.

Furthermore has anyone else found it odd that the review had absolutly no mention of the write speeds of the two drives while together in the array? Shouldn't we expet to see at least something in this respect?

Share this post


Link to post
Share on other sites
I am not sure how effective this would be as it would require three drives in a laptop to achieve it. You cannot software mirror your boot drive. 

WTF are you talking about?!

Share this post


Link to post
Share on other sites

I'm sure he meant to say "Stripe your boot drive in software"

Anyway, I think a mass transition to 2.5" drives would be interesting. Notice how the Seagate 15K drives are using reduced platter sizes already to improve access time? Given that "we" can cram 160GB into a 3.5" drive easily these days, it would be interesting to see an 80GB 2.5" drive with tiny platters. The only real flaw with the idea is that the small actuators you need to cram into 2.5 inches can't move the heads fast enough to create real performance.

Apply some economy of scale to that problem, and I'll bet it would go away.

RAM is so cheap, I'm tempted to make a 512MB RAMDrive and put my paging file and temp directory on it. That would solve any perceived performance problems associated with a slow/quiet boot drive. :)

A $100 motherboard can accomodate 4GB of DDR memory these days.

Less than $500 will get you 2GB.

Share this post


Link to post
Share on other sites
WTF are you talking about?!

Simple, you cannot have your OS on a software Raid ( ie a dynamic disk array create via the OS) array in windows 2k or xp.

Share this post


Link to post
Share on other sites
WTF are you talking about?!

Simple, you cannot have your OS on a software Raid ( ie a dynamic disk array create via the OS) array in windows 2k or xp.

Oh really? :)

M$ Server OS'es have softwaremirroring. Netware does also. Linux does as well.

M$ DESKTOP OS'es do not support mirroring but many others do and on boot drives to.

Share this post


Link to post
Share on other sites

The man isn't saying that they don't support software RAID, he's saying that you can't boot of a software RAID array and he's perfectly correct when it comes to Win2k & XP Pro. If you want to use software RAID under these OS's you need to boot off a single drive, hence the need for at least three drives.

Dave

Share this post


Link to post
Share on other sites

In Windows

0. Simple volume (partition) is BOOTABLE.

1. RAID1 (Mirror) is BOOTABLE.

2. RAID0 (Stripe) is NOT bootable.

3. RAID5 (Stripe With Parity) is NOT bootable.

4. Spanned Volume (associated non-contigous space) is NOT bootable.

So the only thing that you can boot off costs you:

1. 50% of disk space lost.

2. No increase in write speed (best case). Decrease in write speed (worst and common case).

3. No increase in read speed unless you are extremely lucky.

If you want RAID0 (i.e. speed boost), you need either 3rd drive to place system at, or you need to organise two drives so they fit the requirements (assuming 2 disks 40Gb each, and X is about 3):

Scenario 1:

Disk 1: X Gb OS; (40-X) Gb RAID0 part

Disk 2: X Gb simple volume; (40-X) Gb RAID0 part.

Partition breakdown is:

1. X Gb for OS

2. X Gb arbitrary storage (may be temp but w/o speed boost)

3. 80-2X Gb RAID0 with performance boost.

Scenario 2:

Disk 1: X Gb OS Mirror Master; (40-X) Gb RAID0 part

Disk 2: X Gb OS Mirror Backup; (40-X) Gb RAID0 part

Partition breakdown is:

1. X Gb for OS with fault tolerance and likely degraded performance.

3. 80-2X Gb RAID0 with performance boost.

Share this post


Link to post
Share on other sites
But with a heavily used RAM-dive you'd probably lose some bandwith for other apps...

The paging and temp files don't get THAT much use. It's just that when they do, you want maximum responsiveness. Also, PC2100/PC2700 is a whole lot of bandwidth. More than the system is really capable of using in most situations.

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