iXBT Labs - Computer Hardware in Detail






Digest 2004: Desktop Hard Disk Drives

Not long ago interesting events on the drives market (hard disks and disk controllers) used to happen as frequently and would mean to users as much as the ardent fight between the desktop processors (AMD versus Intel) or gaming video accelerators (ATI versus Nvidia). But 2004 was not one of such years, I guess – it was a placid year for hard disks compared to the succession of previous years. Indeed, two or three years ago the data density on hard disk platters used to grow at an outstanding (if not menacing) speed, surpassing noticeably the notorious Moore's law: platter capacity would double less than in a year (!), while according to Moore the number of transistors on silicon dyes doubles each couple of years. By the way, the typical minimum sizes of side members of advanced chips and magnetic recording bits are almost equal now – about 50 nm, that is we have real nanotechnologies in modern hard disks.

The rise of this "magnetic boom" was given by the invention of heads with giant magnetoresistive effect in mid 90s of the last century and the use of some new materials for magnetic media. Hard disk manufacturers were well aware that it couldn't go on forever, but they strived for being the first to "skim the cream" off the advanced technologies, by releasing new models of larger capacities and higher performance almost each six months. Having barely appeared on mass sales, the previous models would be simply discontinued. Large OEM hard disk consumers even started to complain and demand from manufacturers to slow down the pace of development in this industry. They barely had time to validate the new models and to start installing them into their products as the manufacturer's model line would change and the time (and money) consuming validation process had to be started from the very beginning. However the development reserves, conditioned by the mentioned technological leap, gradually receded. And the growth rate of platter density dropped from the empyreal heights approximately to the level, which has set for thirty years already. Data density growth rates started to slow down in 2003 and continued in 2004. And now its growth rates are estimated as redoubling each two years or even worse (see Picture 1).

Pic. 1. Data density growth in desktop hard disks in recent years.

Indeed, data density redoubling for a year is out of the question for a long time: the first desktop hard disk with 80 GB platters (these hard disks constitute the lion's share of the market) appeared (just think about it) over two years ago – that is in autumn 2002! And though all manufacturers upgraded their desktop hard disk lines a tad later (in autumn 2002 most desktop hard disks available on sale were based on 60 GB platters and only in spring 2003 all of them started to use 80 GB platters), it has actually become the last mass-scale upgrade all manufacturers did to the new data density (in 3-inch ATA hard disks). After that, there were only separate steps, which did not lead to complete upgrade of the industry to the new data density of platters.

Out of such "steps" first of all I want to note the 200 GB Seagate Barracuda 7200.7 (UATA and SATA modifications), which were announced in the end of 2003 and actually appeared on the market only in spring 2004.

200 GB Seagate Barracuda 7200.7

Only them and, perhaps, 300 GB Maxtors (MaXLine III series), which appeared a tad later, became the only models with 100 GB platters for a long time. Most manufacturers (practically all) have announced their intent to upgrade to 120 GB platters in the nearest future. But in fact no real actions followed – most of them still feed us with promises. But in 2004 they successfully (as far as it's possible with the average income per one hard disk being less than two dollars) continued to sell good old (in case of really "good" models) 80 GB platter hard disks made in 2002-2003. The only new models worth mentioning are recently appeared (autumn 2004) three-platter 300 GB Maxtor DiamondMax 10 (7200 rpm – the Plus suffix is not necessary, because Maxtor, as well as many other manufacturers, gave up 5400 rpm models),

300 GB Maxtor DiamondMax 10

and the five-platter (!) 400 GB Hitachi Deskstar 7K400, only now appearing on the shelves.

Five-platter 400 GB Hitachi Deskstar 7K400

as well as three-platter hard disks of the same capacity from Seagate (Barracuda 7200.8), which are still on their way. The last two series are notable for the fact that IBM (pardon me – young Hitachi Global Storage Technology) returned to its unique multi-platter design (previous three generations of IBM desktop hard disks had maximum three platters), while Seagate finally mastered not only 133 GB platters (the first in this industry! – we'll see when these disks actually appear on the shelves), but also the three-platter design, on which they would work for the last years and still didn't manage to apply in the previous generation of its hard disks (according to unofficial statements of Seagate employees, they had certain problems with the three-platter design in desktop ATA hard disks).

So, we can say for sure that the hopes for the platter capacity to continue redoubling in desktop hard disks in 2004 completely evaporated: while manufacturers confidently redoubled platter density in the past (10, 20, 40, 80 GB), now even 25% gain is considered a progress and is exposed as a new generation of hard disks. Manufacturers strain every nerve to suck dry the existing magnetic recording technologies, they all switch to "adaptive formatting" of platters depending on individual properties of each "head-platter" pair (Hitachi already tested it in its mobile hard disks and Samsung – in its desktop models; besides almost all companies plan on using it in desktop hard disks). And again the industry returns to the time-proven method of increasing hard disk capacities – by increasing the number of platters and heads. However, the desktop market (unlike the professional data storage market) is presently not in need of increased hard disk capacities – the most popular models now are 80-160 GB hard disks, which are quite sufficient for the vast majority of personal and corporate (office) users. That's why the offered 250-400 GB ATA hard disks rather forestall the real demand, though they are quite welcome for professional near-line data storage devices, having provided terabyte storages without much hustle and expenses.

Meantime the desktop storage industry has been increasingly active for the last couple of years in promoting another tendency of raising storage capacity and performance (with the platter density growth slowed down, hard disk performance growth has also slowed down – linear writing speed is not growing either). I mean "consumer" RAID. Thanks to the efforts of some "storage brands" (Adaptec is not the last among them), the simplest RAID 0 and 1 controllers (and sometimes even 10) have appeared even in cheap PCI-controllers and motherboard chipsets. And while such controllers were still unusual for most users in 2003, in 2004 it's nonsense not to have ATA RAID on your motherboard or system chipset. Indeed, at the current humane prices for desktop hard disks, it would be cheaper to buy two 80 GB hard disks and set up a practically "free" RAID 0 with its double performance and capacity, instead of spending the same sum of money for a single 160 GB hard disk, which will have practically the same performance as a single 80 GB hard disk. Besides, the famous Intel Application Accelerator (caching Windows-driver for hard disks), which used to work with single drives, has been working only with RAID for two years already, it deliberately leaves off "non-RAID".

WD Caviar RAID Edition — these hard disks are designed specially for RAID.

I want to note two things here. Firstly, 2004 has become a first night for an interesting technology called Matrix Storage, used by Intel in its desktop i925X/915 chipsets. It allows to organize two independent RAID arrays with just two SATA hard disks – stripe for performance and mirror for reliability. This approach is no news for professional SCSI systems, but they cost dear, while in this case it's offered practically free. The second positive tendency of 2004 is the infiltration into users' desktops of cheap RAID 10, consisting of four SATA hard disks (reliability of a mirror and high performance and capacity of a stripe combined). Built-in RAID 10 controller appeared in desktop chipsets from SiS, VIA, and even Nvidia. Unfortunately, Intel persists in "holding back" this feature in its ICH6R (according to those who designed the RAID controller for Intel in his chipset, this feature is out there). Perhaps, this is done for marketing reasons.

16-port SerialATA 1.0 RAID-controller Adaptec AAR-21610SA for PCI-X bus.

But capacity and RAID issues were not the only ones to occupy the market of desktop ATA drives in 2004. In my opinion, the successful progress of Serial ATA interface has become a more important issue in this segment – both in conquering the hearts of users (you won't believe this, but there are a lot of hunkers among Russian PC integrators, who still don't want to deal with Serial ATA) as well as the interface development proper. It's not even that hot swapping of cheap SATA controller was improved; prices for SATA and UATA models became almost even, there appeared a new generation of SATA 1.0 controllers of a higher performance; and we are on the brink of the next generation of controllers (and hard disks) supporting the next standard for this interface – Serial ATA II – with doubled data transfer rate (3 Gbit/sec). What is much more important at this stage of Serial ATA development is the appearance of hard disks and controllers supporting Native Command Queuing (NCQ) – intellectual mechanism that changes the sequence of commands in a disk queue, which provides noticeable performance gain in applications.

NCQ is part of the Serial ATA II specification, but it's already available in some Serial ATA hard disks and controllers. Among the hard disks already supporting NCQ there are only some models (senior ones, as a rule) from Maxtor and Seagate. The situation with controllers is a tad better – Intel ICH6 south bridge, several models on dyes from Silicon Image, Promise and Marvell, some of them are already designed for Serial ATA II. A number of major manufacturers (like Adaptec and LSI) prefer to launch professional Serial Attached SCSI (SAS) controllers right away, which are back compatible with SATA II. In general, demonstration of Serial Attached SCSI in various technical forums, which is going to replace the traditional parallel SCSI, has become one of the main break-throughs of the hard disk interfaces in 2004. However, it's still too early to include SAS into our yearly digest – let's wait for the products for SAS in 2005.

At present "consumer" hard disks are growing their buffers pending the appearance of new high-performance and intellectual interfaces: in 2004, the 8 MB cache (first appeared in ATA-hard disks from Western Digital less than three years ago) has become de facto standard for most ATA drives, while some models are even equipped with 16 MB buffers. I think that it's not the limit and the 32 MB cache is round the corner.

But the other Western Digital initiative was not supported by other manufacturers in 2004 – in 2003 this company released the famous WD Raptor with rotational speed of 10000 rpm and SATA interface. In 2004 WD doubled its capacity to 74 GB and also equipped it with a simplified NCQ modification (so called TCQ). According to the leading figures (Seagate and Maxtor), the time of 10000 rpm desktop models has not come yet. The demand for such drives is low and bears a specific character; though they will be able to present Raptor counterparts, if it suddenly becomes commercially expedient. WD has also made a mistake with TCQ – it should have applied NCQ right away (hard as it might be), because in practice TCQ demonstrates a controversial effect. However, the hero of 2004 — WD Raptor 2 (WD 740GD) — looks good even without it, awards from mass media are falling on this model like a rain (the best read and seek rates among ATA drives, noticeably quieter and faster than the first Raptor, doubled capacity). We should also note that in the second half of 2004, due to its worthy financial operations, Western Digital grew the second largest (after Seagate) world manufacturer of hard disks, having pushed Maxtor to the third position, which situation has been slippery of recent. WD's turnover is based solely on "low income" desktop ATA drives, while its competitors offer a much wider range of products. However, WD already has conquering plans on other markets, but it's another story.

Thus, the market of desktop hard disks in 2004 suffered no tempests and revolutional changes. The model range of hard disk changed only moderately, more attention was paid to new controllers and interfaces. Though we still had some outstanding and joyful events. Happy New 2005 Year!


Alex Karabuto (lx@ixbt.com)
January 06, 2005

Write a comment below. No registration needed!

Article navigation:

blog comments powered by Disqus

  Most Popular Reviews More    RSS  

AMD Phenom II X4 955, Phenom II X4 960T, Phenom II X6 1075T, and Intel Pentium G2120, Core i3-3220, Core i5-3330 Processors

Comparing old, cheap solutions from AMD with new, budget offerings from Intel.
February 1, 2013 · Processor Roundups

Inno3D GeForce GTX 670 iChill, Inno3D GeForce GTX 660 Ti Graphics Cards

A couple of mid-range adapters with original cooling systems.
January 30, 2013 · Video cards: NVIDIA GPUs

Creative Sound Blaster X-Fi Surround 5.1

An external X-Fi solution in tests.
September 9, 2008 · Sound Cards

AMD FX-8350 Processor

The first worthwhile Piledriver CPU.
September 11, 2012 · Processors: AMD

Consumed Power, Energy Consumption: Ivy Bridge vs. Sandy Bridge

Trying out the new method.
September 18, 2012 · Processors: Intel
  Latest Reviews More    RSS  

i3DSpeed, September 2013

Retested all graphics cards with the new drivers.
Oct 18, 2013 · 3Digests

i3DSpeed, August 2013

Added new benchmarks: BioShock Infinite and Metro: Last Light.
Sep 06, 2013 · 3Digests

i3DSpeed, July 2013

Added the test results of NVIDIA GeForce GTX 760 and AMD Radeon HD 7730.
Aug 05, 2013 · 3Digests

Gainward GeForce GTX 650 Ti BOOST 2GB Golden Sample Graphics Card

An excellent hybrid of GeForce GTX 650 Ti and GeForce GTX 660.
Jun 24, 2013 · Video cards: NVIDIA GPUs

i3DSpeed, May 2013

Added the test results of NVIDIA GeForce GTX 770/780.
Jun 03, 2013 · 3Digests
  Latest News More    RSS  

Platform  ·  Video  ·  Multimedia  ·  Mobile  ·  Other  ||  About us & Privacy policy  ·  Twitter  ·  Facebook

Copyright © Byrds Research & Publishing, Ltd., 1997–2011. All rights reserved.