Notebook IDE RAID article: discussions, please

FatBurger · Mar 1, 2002

New idea: let's keep one thread with discussion on recent articles. The most recent article to be posted on THG is the notebook IDE RAID test. This article was actually available quite some time ago to those who receive the THG newsletter, but I digress.

First I'd like to point out the CPU utilization table. Was the IBM 60GXP test run on a PCI controller, or off the motherboard? I would assume that would make a difference with CPU utilization.

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!

As has been said in another thread, the 60GXP outperformed the RAID array, although it was fairly close in most red/write tests. The comment above could be bad translation, or simply meaning to compare the RAID to a typical desktop drive (not the fast 60GXP).

Quarter Pounder Inside
Don't step in the sarcasm!

Lars_Coleman · Mar 1, 2002

Would you really want to do this anyways, considering that the price per gigabyte for a notebook drive is more then a desktop?

People and hard drives are like bandwagon fans and sports!

slvr_phoenix · Mar 1, 2002

For some people, noise and/or heat are big issues. I could see it being an important feature for a set-top box where you don't want a noisy thing running while you're watching a DVD.

For that matter, appearantly to <A HREF="http://www4.tomshardware.com/howto/02q1/020111/index.html" target="_new">Frank Völkel</A>, having the smallest PC immaginable (without actually going to a laptop) is important to, even if it means sacrificing, well, everything.

To me though, I think an important concept was alomst entirely missed, and that's data redundancy. If you could find a drive mount converter that would let you stack two 2.5"s in the space of a 3.5", then you could run in RAID 1 mode and never have to worry about losing data from a hard drive crash again. That'd be pretty cool to put into those tiny boxes that managers prefer to use because they don't look as ugly and invasive as a mini-tower. (Or dare I suggest a full-tower.) I mean they only use MS Office kind of stuff anyway, so they don't need to worry about speed. They just need to worry about losing everyone's timesheets if their hard drive crashes. It would sure beat doing a weekly tape back-up of their whole system.

<pre>AROOOOOOOOOOOOOOOOOOOOOOO!</pre>

tas · Mar 1, 2002

The benchmark results are quite clear: only two 2.5" drives are able to outperform a modern desktop drive in terms of transfer performance

Clear? As mud.

The burst rate? big deal like the small cache is going to do anything in the real world. The transfer rate? Two different tests used so not much use.

The one real world test? The raid got toasted.

I don't think the laptop raid would even beat a modern 5400 RPM 20Gb/platter drive in real world tests.

It's silly, it's expensive, Tom must have had a slow day or something.

What's next? How to put a second engin the the front of your old VW bug?

Tom.

Edit: should have said Patrick Schmid had too much time on his hands, not Tom...Edited by tas on 03/01/02 06:45 PM.

FatBurger · Mar 1, 2002

If you'll notice, Tom didn't write that article.

Quarter Pounder Inside
Don't step in the sarcasm!

slvr_phoenix · Mar 1, 2002

What's next? How to put a second engin the the front of your old VW bug?

Even if Patrick Schmid did tell us how, we still all know that taking out the back seat to put in a big block V8 would be a lot more fun.

<pre>AROOOOOOOOOOOOOOOOOOOOOOO!</pre>

Harisahmed · Mar 1, 2002

How is it that the notebook raid was able to achieve about 2x the performance of the single notebook drives, but the performance increase for a regular IDE raid is closer to 10%?

FatBurger · Mar 2, 2002

From my personal experience, the performance increase is a lot more than 10%. Where have you seen only 10%?

Quarter Pounder Inside
Don't step in the sarcasm!

camieabz · Mar 2, 2002

Was the IBM 60GXP test run on a PCI controller, or off the motherboard?

Don't know, but my GXPs in RAID 0 use 12% of 1000MHz, so the utilisation on the review seems to tally.

I don't really go for the review at all. Few of the results tally with my results on my HDDs. My read burst speed is well over 80MBps, not concurrent with the review. My access time is at 8.3ms, which is far lower then the review data.

All my data comes from HD Tach, which I noticed was dropped as 'unfavourable' in this review. Why? Everyone uses HD Tach. It has become the popular quick benchmark. If its unreliable, why do so many recommend it in the forum?

I'm going to look into this further with some other benchmarking progs. Maybe have better answers from them.

~ They're coming to take me away. Ha Ha! ~

mbetea · Mar 2, 2002

i just don't understand why the big deal about the noise. if you're that bothered by noise that a newer hdd makes, wear earplugs or don't use a computer. even with no case fans and just the hsf on the cpu going drowns out any hdd activity that i've come across and thats with a volcano 6cu and retail amd.

I run duals because i multitask between notepad, outlook express and winamp :lol:

Guest · Mar 2, 2002

Can anyone tell me where to get the adapter for the 2.5" drive? I have an extra drive that is doing nothing and I would like to install it in a desktop as a slave drive.

Ncogneto · Mar 2, 2002

I don't really go for the review at all. Few of the results tally with my results on my HDDs. My read burst speed is well over 80MBps, not concurrent with the review. My access time is at 8.3ms, which is far lower then the review data.

All my data comes from HD Tach, which I noticed was dropped as 'unfavourable' in this review. Why? Everyone uses HD Tach. It has become the popular quick benchmark. If its unreliable, why do so many recommend it in the forum?

First off there is absolutely now way you are getting a access time of 8.3 ms off any IDE drive PERIOD! ( now if you are using a SCSI drive than I apologize, but that would not make any sense in comparison)

Secondly, time and time again HDDtach has shown to give inconsistent results and is shunned by all the major storage review sites. It exspecially is unrelaible in testing a RAID array.

It's not what they tell you, its what they don't tell you!

camieabz · Mar 2, 2002

Tell ya what. Anyone who wants a copy of the screendump can P.M. me with their e-mail address.

~ They're coming to take me away. Ha Ha! ~

camieabz · Mar 2, 2002

i just don't understand why the big deal about the noise. if you're that bothered by noise that a newer hdd makes, wear earplugs or don't use a computer.

Heh! Good one.

I use drive enclosures to avoid the noise issue. The temps are fine too. Around 38C.

~ They're coming to take me away. Ha Ha! ~

Ncogneto · Mar 2, 2002

What drive? You are reporting an access time of 4ms faster than any review site has tested any IDE drive for. With the rotational latency of 7200 rpm drives being around 4ms, your drive is faster than even the cheetah x15LP? And you wonder why some people question HDD tach? Please take a look here:

<A HREF="http://www.storagereview.com/articles/200103/20010329IC35L060AVER07_1.html" target="_new">http://www.storagereview.com/articles/200103/20010329IC35L060AVER07_1.html</A>
It's not what they tell you, its what they don't tell you!Edited by ncogneto on 03/02/02 10:23 PM.

fugly · Mar 2, 2002

Does anyone think that it would be a smart idea to include an IDE raid controller and two hard drives in a laptop? Is anyone already doing this?

Pardon me if this is a stupid question.

Ncogneto · Mar 2, 2002

Furthermore I would like to explain the differnce between two often confused terms.

Seek time: The time it takes for the heads to position themselves on the disk for a read/write assignment. This is the figure you will see when making a purchase of a drive.

Access time: The time it takes for the disk to actually perform the read/write operation. This is the figure that you will see reported form HDD Tach and others. Why is it not the same as the Seek time? Because it is a combination of Seek Time + rotational latency.

Rotational latency: This is the one you don't here much about. The hard disk platters are spinning around at high speed, and the spin speed is not synchronized to the process that moves the read/write heads to the correct cylinder on a random access on the hard disk. Therefore, at the time that the heads arrive at the correct cylinder, the actual sector that is needed may be anywhere. After the actuator assembly has completed its seek to the correct track, the drive must wait for the correct sector to come around to where the read/write heads are located. The faster the disk is spinning, the quicker the correct sector will rotate under the heads, and the lower latency will be. On average, latency will be half the time it takes for a full rotation of the disk. This figure is a constant in relation to the speed at which the disk is rotating ( ie 5400, 7200, 15,000 )

Now Raid O improves STR ( sustained transfer Rate ) but it does absolutely nothing for Access times, and in some cases actually makes it slower ( however only by a small amount). Raid O seems to be the hot item as of late, but is terribly misunderstood. Most people have little if no use for it as it only improves transfers for large chunks of Data. The much more common data transfers ( smaller chunks ) is handled by the disk cache. So, if you are using applications that are commonly moving around 200MB or more of data, then Raid O is a way to improve performance. If your mainly working around in a normal Windows enviorment you will not see much of an improvement if any. Upgrading to drives with a faster spindle speed ( ie 10k or 15k ) would give you a quicker "feel" as they inherently have faster seeks. This becomes even more obvious as the disk becomes fragmented. Remember all data is not written to a disk as you would read a page of a book. It is not uncommon for the drive to need to perform a seek, reed 16k of data, perform another seek, read 32k, perform another seek, reed 64k, etc etc. In instances like this RAID O is doing nothing for you as you are more bound by the access times of your Disk(s) then you are of its maximum STR.

It's not what they tell you, its what they don't tell you!Edited by ncogneto on 03/02/02 10:27 PM.

Ncogneto · Mar 2, 2002

No I don't. Laptops are designed mainly for portability in mind factoring in such things as size, weight, battery time etc. Better optimization of the disk cache would better serve this purpose, or even if needed, better use of a firewire drive.

It's not what they tell you, its what they don't tell you!Edited by ncogneto on 03/02/02 04:23 PM.

camieabz · Mar 3, 2002

Your link doesn't work.

~ They're coming to take me away. Ha Ha! ~

Ncogneto · Mar 3, 2002

Sorry about that the link should now be working

Also, let me first say that I in no means am trying to start a flame war here, nor am I calling you a liar. I beleived that you were indeed getting the results you claimed. It is not you that I question, rather the results you are achieving and the way in which you are testing. First let me say that I have looked at the jpegs you sent me and at first noticed a few things. Your transfer rates are bouncing all over the place with a maximum of 62,xxx and a minimum of 17,xxx. Either your controller is struggling or HD tach is having issues with your array as well. I would really like to see results with another benchmark to see if it produced a simular graph. Secondly your only testing the first 7.6 gigabytes of your array, you need to enable it to test the full array to get a more actual test. I have seen numerous sites complain of erroneous results obtained by testing raid arrays by HDDTach and your results only add to the confusion it produces ( maybe this is why TOM made a statement as such concerning HDD tach in this review ).

As for the Access times it is giving you I can only guess as to why it outputs that data ( 8.3ms ) and I would like to ask if it shows you that consistantly? Please read my above post concerning rotational latency and you will beging to understand why RAID 0 or any RAID level for that matter cannot improve upon the access time of the slowest single drive in the link. In any stripe the OS must wait until all the heads of all the disks are in position before it performs a read/write command.

It's not what they tell you, its what they don't tell you!

Crashman · Mar 3, 2002

Seagate's rubber-clad 54000 RPM 3.5" drives are nearly silent, very inexpensive, and, well, cool enough that they can run with a rubber sleeve over them! And they still outperform laptop drives. So there really isn't any excuse to run laptop drives on a desktop.

What's the frequency, Kenneth?

Ncogneto · Mar 3, 2002

U series five....yup nice drives if quiet is what your after, last forever as well.

It's not what they tell you, its what they don't tell you!

camieabz · Mar 3, 2002

I see where your coming from man.

My access times have always been in the 8.2 - 8.7 region. If you check out the specs on 60GXPs you'll find they're supposed to be 8.5ms.

The transfer rates have always been as such. Let me put it this way. If I create an image of my C: using Drive Image, it takes about 5 mins to copy it to the last partition on my array. Thats 2.2GB with no compression or any type of image or write verification enabled.

In short, it seems very fast for me, but then my last system had an ATA33 6.4GB Quantum Fireball, so I suppose there's no real comparison.

I plan to avail myself of Winbench and some other benchmarking tests soon, and hope to post results here.

~ They're coming to take me away. Ha Ha! ~

Ncogneto · Mar 3, 2002

If you check out the specs on 60GXPs you'll find they're supposed to be 8.5ms.

Yes that is seek time now add that the rotational latency of 4ms then you get 12.5 ms access which is what you should be getting.

It's not what they tell you, its what they don't tell you!

lhgpoobaa · Mar 4, 2002

bah. my 60gxp is damn near silent anyways.

"I came, I saw, I overclocked", Julius 'Smokin CPU' Caesar

Notebook IDE RAID article: discussions, please

Illustrious

Distinguished

Splendid

Distinguished

Illustrious

Splendid

Distinguished

Illustrious

Splendid

Distinguished

Guest

Guest

Distinguished

Splendid

Splendid

Distinguished

Distinguished

Distinguished

Distinguished

Splendid

Distinguished

Polypheme

Distinguished

Splendid

Distinguished

Illustrious

Share this page