6 IDE Hard Disc Drives between 15 and 30 GBytes – THG.RU

Introduction

Tom’s Hardware Guide is known as the premier source for computer hardware information. Running several hardware labs and proudly serving one of the largest readerships worldwide, we are ready to expand our activities to the storage sector as well. For me it is an honor to be the editor who does the first hard disc drive review, as this is meant to introduce a series of periodical articles. This initial one deals with six hard drives from Fujitsu, Quantum, Seagate and Western Digital. I’m sorry that we did not include drives from Maxtor and IBM, but I promise that we will include their products for the next articles.

IDE vs. SCSI

Of course there was the question of whether to start with an IDE or a SCSI hard drive review. SCSI is still the most flexible and of the two, it is the faster bus system. Ultra2Wide-SCSI is the common standard, allowing up to 15 drives in the SCSI chain and enabling a bandwidth of up to 80 MByte/s. High performance drives are available with an Ultra160 interface, which gives you double the bandwidth. But those terrific speeds are highly priced. SCSI drives are at least a third more expensive than IDE models. Furthermore, you will need an SCSI host adapter. IDE interfaces can be found on every motherboard for the past few years now, whilst only a minority of all motherboards is being produced with an on-board SCSI chip. If you want to spend all that money, you will get a very solid base. Regarding the pure data transfer rate, there is no difference between two identical IDE and SCSI drives. SCSI systems can be faster if you use several drives in a RAID configuration. That is also possible with IDE, but currently still restricted to the maximum IDE bandwidth of 66 (ATA66-spec) and very soon 100 MBytes/s (ATA100-spec). Some drives can also be faster thanks to the special SCSI features (command queuing) or just by being the latest generation high end drives. As high end computers usually require best performance, the fastest drives are primarily developed for the SCSI interface.

IDE, on the other hand, is much easier to handle. As only two drives can be attached to an IDE port, there is only the Master or the Slave setting. No drive ID is required and the bus does not require any termination. Incompatibilities between two drives have become very unusual in the last years. Moreover, I’d guess that most of our readers make use of IDE drives, which is apparent hundreds of emails asking for advice.

IDE vs. SCSI, Continued

As already mentioned, it is also possible to run IDE RAID systems at very low cost, as IDE RAID controllers are very cheap compared to SCSI models – but they are usually limited to four drives. Moreover, you can use cheap IDE drives. When using two drives, the data transfer performance will practically double, and you will get even more performance using three or even four drives. Anyhow, you should know that performance does not increase linearly. Two drives still double your performance, whilst three drives do not provide triple the transfer speed any more. Using four drives will reach the limit of UltraDMA/66 at the most, so that you will usually get approximately 3.5x performance.

This photo shows the 40 pin IDE connector. Please note that a UDMA/66 cable comes with 80 pins, while the connector only has half the pin count.

That’s how the older SCSI connectors look like (SCSI, SCSI-2 or FastSCSI, UltraSCSI).

The latest SCSI devices make use of the Ultra2Wide interface, but normal Wide and UltraWide SCSI (20 and 40 MByte/s) make use of the same 68 pin connector.

This one is the 34 pin FDD connector!

Test Criteria

Some people might believe that hard drives only differ in their performance. As a matter of fact, performance is only one chapter of the hard disc story. In an attempt to make drives faster, one popular measure has been to increase the rotational speed. As a result of higher friction, the drives become hotter and also make more noise due to high rotational speeds. That issue might not be a problem for servers, but home and office computers ought to run cool and should avoid noise as much as possible.

Test Criteria – Performance

A fast hard drive has to provide both short access times and high data transfer rates. The rotation speed has a big influence on both numbers, as it can both increase the transfer speed and reduce access times, due to the read/write heads having to wait less time until the requested data comes past the heads.

In order to increase performance, all modern disc drives come with an UltraDMA/66 (or UltraATA-2) interface and an internal cache memory. All drives in this article come with 2 Megabytes, apart from Seagate’s 20 GByte Barracuda ATA I (which is in the process of being replaced by the faster Barracuda ATA II), which can only make use of 512 KBytes. In spite of this one vanishing from the market, I decided to include it in this review, as it has been our reference hard disc for almost all tests in the last six months.

Test Criteria – Noise Level

Every time you enter an office you can hear if computers are present. It is the power supply fan and the hard drive which are mostly responsible for the noise you can hear. High quality power supplies come with a thermistor and automatically increase or decrease the fan’s rotation speed depending on the temperature inside the power supply.

The first hard drives that were running at 7,200 rpm produced a high frequency wheeze, which most people describe as almost unbearable if you have to work at or near them. Today, this wheeze could be reduced to a minimum, so that drives running 7,200 rpm are not any louder than 5,400 rpm models.

Test Criteria – Bad Vibrations

Apart from the spindle motor and all interrelated drive motions causing noises, the resulting vibrations can pass onto the computer case as well. Depending on the quality of your case, one or even several parts will be affected by them. I think you can imagine that vibrant parts of sheet metal are quite nasty and will destroy any remaining concentration. It’s quite difficult to find a scientific method to mark out a number, which represents for the vibration level, so I hope you can live with my personal evaluation for the time being.

Test Criteria – Temperature

As already mentioned above: Increasing rotation speed entails higher friction, which in turn leads to higher heat dissipation. Drives running at 10,000 rpm (SCSI only) necessarily need active cooling, as the magnetic discs can lose the data integrity at increased temperatures. IDE drives are meant to run without cooling, as the expense factor is more important than with SCSI models. Taking a look at the temperature numbers of the test drives, you can run all the hard drives in this review without active cooling. However, it is always an investment for data safety to use a hard disc cooling system. It can even become a performance issue, as most drives will regularly recalibrate the heads at high temperatures. That may be a small argument in favor of expensive SCSI drives: Everybody who needs continuous data streams (such as with hard disc recording or video editing) will be grateful if the drives only recalibrate when it’s absolutely necessary. That’s in fact an issue which is usually considered during the development of high end drives.

Please remember that there are other components, which suffer from high temperatures as well. Every drive comes with its own electronics (IDE means integrated disc electronics), featuring at least a controller chip, a firmware chip and maybe one or several cache memory chips. High temperatures also easily affect those chips. As you already know from processors, the chip temperature may not exceed a certain value. Of course there is hardly a way of measuring the real chip temperature, so the common option is to use the surface temperature. Most Pentium III CPUs are specified for up to 80 degrees Celsius, for example. Improper cooling results in even higher temperatures, which may result in turn with a dead chip. The same applies for hard disc components of course.

Test Criteria – What’s too hot?

You will probably not be able to determine when the drive is running too hot, but you can easily find out if the drive temperature exceeds approximately 50 degrees Celsius. Run your computer for some time (at least two hours). After this time, you should carefully open the case and put your hands on the disc drive (please don’t touch any electronic part on the bottom). It will be best to keep the computer running, as turning it off will cool down most devices pretty fast – at least out of potential critical temperatures.

A direct contact of hot objects to your skin causes pain at approximately 55 degrees Celsius. That’s what you should also try to find out with your hard drive. If you feel that you have to take away your hand from the disc’s surface, you can be sure that the temperature exceeds approximately 50 degrees. In this case, I would consider getting a hard drive cooling system to ensure a long lifetime for your disc.

Fujitsu MPF-3204AH

The drive is able to provide 27 MByte/s within the first half of the medium. In the second half (10-20 GBytes), the transfer speeds decrease slowly to 17 MByte/s.

There is no difference in these results between Windows 98 and Windows 2000.

Fujitsu MPF-3204AH, Continued

Fujitsu, just like Seagate, makes use of an aluminum case. This material seems to be best suited for heat dissipation. The drive heats up to 47 degrees Celsius, which is as hot as our reference hard drive, the Seagate ST320430A. Both drives should be installed correctly to make the heat dissipate over to the computer case. There was nothing to complain from the vibration point of view. Even though Fujitsu uses aluminum, the drive runs comparatively calm.

With this drive, the jumper descriptions are situated at the top of the drive on a sticker (see photo). Beginners possibly will not know on which side to start, but as the worst case scenario is a non-functioning hard drives, you will just have to move the jumper.

The 3204AH is an average performer. On the one hand, it is able to provide very high maximum transfer speeds, but it will also drop more than other drives. Its big advantage is the low noise level, which was not surpassed by any other candidate. The seek time was quite good both in Windows 98 and Windows 2000, making this drive best suited for data base applications or applications, which require lots of little file accesses.

Click here for detailed specs on the Fujitsu website.

Quantum Fireball Plus LM 20.5

Great so see that the Quantum drive can sustain the transfer speed between 27 and 25 MByte/s within the first half of the medium. It slowly goes down to 18 MByte/s in the second half.

Absolutely the same in Windows 2000.

Quantum Fireball Plus LM 20.5, Continued

The Fireball Plus LM looks like a typical Quantum drive, coming without any stickers. All information on the top is engraved instead. The Fireball Plus LM can’t compete in the Business Disc WinMark, but it performs quite well in the Highend part. It scores worst in Ziff Davis’ Content Creation Winstone 2000, which is of course only interesting if you need to run applications like Adobe Premiere, Photoshop, Macromedia Director or the audio software Sound Forge. There is mixed news regarding CPU utilization – while the drive occupies the CPU for almost half the time (!) in Windows 98, it requires less than most competitors in Windows 2000 (22%).

The drive provides a decent maximum of data transfer performance, while it is great to see that the minimum performance got also a good result. All in all you will get a very balanced hard drive, which you should not even hear working. The Quantum drive almost got the lowest noise measurement result and just the Fujitsu drive makes less noise. But there’s quite an controversial issue again: Seek times under Windows 98 are quite bad, while the drive gets much better results in Windows 2000. It’s in fact the fastest drive regarding access performance. The drive heated up average during our test. We measured 45 degrees max, which is tolerable of course.

The backside reveals all electronic chips. Such a drive should be handled with some more care than others, because the chips at the bottom are not protected at all.

In the end I have to say that I wouldn’t go for this drive if I planned to use Windows 98. Quantum’s Fireball Plus LM will provide much better performance using Windows NT or Windows 2000. It’s a cool and quiet drive with great NT performance, but it was not able to convince us for use with Windows 98 at all.

Please click here for detailed specs on the Quantum website

Seagate Barracuda ATA ST320430A

This drive will provide between 25 and 28 MByte/s within the first half on the medium. In the second section, the data transfer speed will be decreasing from 25 to almost 18 MByte/s.

Seagate’s Barracuda ATA scores just as fast as under Windows 98. The only difference is the reduced CPU utilization in Windows 2000.

Seagate Barracuda ATA ST320430A, Continued

This drive should be no longer available today. Nevertheless, we wanted to include it, as it has been our reference hard drive for the last months. More than half a year ago we decided to get some of those drives, as it was one of the fastest models then. Moreover, Seagate covers the bottom of their drives (see photo below), so that the electronic parts are better protected. That is usually not too important, but as we have to change system setups over and over again, those drives have proven to be very suitable for testing.

All Barracuda ATA series hard drives are packaged inside an aluminum die casting case, which has both an advantage and a disadvantage. It’s generally known that most CPU heat sinks are made of aluminum or aluminum alloys, as this material is quite suited for heat conduction. The same applies to the Seagate hard drives. Thanks to their aluminum case, all produced heat will pass over to the case quite fast, spreading over the heavily increased face. The bad feature is that aluminum is not able to eliminate the drive’s vibrations as good as other metals. That’s why you can clearly feel that a Barracuda ATA is vibrating a lot more than, for example, the Western Digital drives.

With regard to performance, the ST320430A does not belong to the very top any more. In all real world benchmarks it cannot hold up with its competition, most likely due to its smaller cache size (512 KBytes instead of 2 MBytes of all others). Only the pure data transfer speed is still quite good. Nevertheless, the Seagate drive is one of the hottest and also noisiest candidates. If you don’t care about that, you will of course get a modern hard drive for little money.

Click here for detailed specs on the Seagate website.

Seagate Barracuda ATA II ST330630A

The drives provides more than 29 Mbyte/s over the first third of the medium. It steadily drops to 24 between 10 and 20 GBytes and is between 24 and 18 MByte/s within the last third of the drive.

The data transfer speed in Windows 2000 is almost the same as that in Windows 98. Please consider that the drive only needs 22% CPU time in Windows 2000, while this time is about 39% in Windows 98.

Seagate Barracuda ATA II ST330630A, Continued

We got the latest 30 GByte model from Seagate for this review. As you can already see from the name, the Barracuda ATA II is meant to be the successor to our reference hard drive. The drives comes with 2 MBytes cache memory, which I believe is responsible for the better benchmark results. The ST330630A scores best in most benchmarks. The Western Digital models are slightly faster in the Content Creation Winstone running under Windows 2000 Pro. Nevertheless, the Barracuda ATA II provides the best data transfer rates of this review, making the drive suitable for all applications, which require constant data streams.

The great performance has its price of course, as the new Seagate drive is clearly the hottest one. Being the only drive which heats up to 50 degrees Celsius, my personal advice is to get a hard drive cooling kit. Nothing big, but just enough to reduce the working temperature by a few degrees.

I wasn’t surprised to see both those temperature values and some vibrations with this drive, as it makes use of the same aluminum casting case as the predecessor ST320430A. That shows this drive’s designation: High end IDE systems. If you want best performance, you will be happy with the Seagate drive. Nevertheless, beginners should maybe go for another drive, as this one needs to be taken care of in order to reduce case vibrations and drive temperature.

Click here for detailed specs on the Seagate website.

Western Digital WD153BA and 205BA

The drive’s performance decreases quite evenly over the medium. It provides 22-25 MByte/s within the first 5 GBytes, 21-24 MByte/s between 5 and 10 GBytes and down to 16 MByte/s in the last sector.

I was quite skeptical when I saw the performance drop at ~ 2 GBytes. There is obviously some driver issue giving the benchmark this incredible performance drop. My hard drive contained approximately 2.1 GBytes of data, so all benchmarks were running in the area where this performance drop is supposed to be. As there were no issues with the real world benchmarks, I do not believe in this result.

The 20 GByte drive gets a similar graph as the 15 GBytes model.

There’s also a similar effect with the 20 GByte drive. The graph tells us about a performance drop at approx. 12% of the drive’s capacity which simply is not there at all.

Western Digital WD153BA and 205BA, Continued

The drives from Western Digital do not make use of an aluminum case, that’s why we could feel hardly any vibration coming from the drives. The good impression stays after checking the noise level as well. Even though you can clearly hear the drive whistling, the noise does not bother at all. Most people won’t have any reason to complain about too high noise level or about unwanted noises.

Both drives got good results in most real world benchmarks except the Highend Disc Winmark and the Seek Time tests. Nevertheless, you will get decent performers, even though the transfer speed tests (using the WinMark’s Disc Inspection Test) don’t attest to those drives being the fastest. Instead of heading for pure performance, Western Digital optimized other technical details, which don’t seem to be that important on the first view. For example the CPU utilization time of their drives has been successfully reduced, requiring only half the time of all competitors in Windows 2000 and being among the most frugal models in Windows 98. That’s also why the WD drives can keep up with their competitors in the Disc Winmark tests, even taking the lead in ZD’s Content Creation Benchmark – although the pure data transfer speed is higher in all other models.

All chips are placed on the top side of the electronic board to avoid any kind of harm.

Here you can see the jumper block of WD hard drives. The jumper is on spare right now, while you can see how you have to plug it on for Master or Slave configuration.

Western Digital shows impressively that getting high synthetic benchmark results does not mean you have to have a fast drive. Both WD models basically cannot deliver the highest performance, so I would recommend others for hard disc recording or video editing. But they come with a good noise level, almost no vibrations, low working temperatures and brilliant electronics. When you add all these together, the Western Digital drives can be recommended for almost all users. Due to their smooth handling, beginners and noise sensitive persons should be happy with them.

Click here for detailed specs on the Western Digital website: WD 153BA

Click here for detailed specs on the Western Digital website: WD 205BA

Test Setup and Benchmarks

Test Setup: Explanations

We were looking for a test system which would be easy to handle. The little motherboard from IWill makes use of the Intel i810 chipset, which includes both on-board graphics and a decent UltraDMA/66 controller. The graphics chip is certainly not able to impress anybody, but it is enough for the purpose of these tests, as we want to show the differences between different hard drives. We could have used a PCI IDE controller, but as the on-chip solutions from Intel have always been among the very best, there is no necessity to do so.

Intel’s Celeron 500 is a cheap and fast processor, which provides enough performance for our hard drive tests. Even though the CPU runs at only 66 MHz FSB, the system provides the same IDE interface performance as with i820 or i840 systems and a CPU running at 133 MHz system clock. Finally, the IDE controller also works with the same speed as the PCI bus (33 MHz).

Of course we made use of Intel’s latest Busmaster drivers for Windows 98 and Windows 2000. Just a little hint: You should know that there is no chance of removing them once they are installed. Even today, the Intel storage drivers still come without uninstalling options.

In case of Windows 98 and Windows 2000, we used Intel’s latest graphics reference driver ver. 4.1, which has been released recently.

Temperature Measurements

IWill’s W100 motherboard provides one header for a thermal sensor (called thermistor). It was originally meant to monitor the system temperature. Nevertheless, it can be abused to measure the temperature of certain surfaces (e.g. a CDROM or hard drive). We placed the sensor on the top and all four sides of each hard drive in order to get its temperature. Usually we got differing values from the top and a side. However, we always took the highest result, even if we got it at the side of the drive.

Noise Measurements

In order to be able to get a noise level for each drive, we built a special crate, lined up with panels of quietizing material. It is 50x50x40 centimeters in size and can also be used for other devices as well (fans, CDROM drives etc).

That’s a photo of our noise measurement crate. We make use of a little digital analyser from a German mail-order company to get a dB(A) value. Let me please make clear that this method is absolutely proprietary, and therefore our results cannot be compared with results of other reviews.

In the middle you can see our high fidelity microphone from Sennheiser inside the metal pipe. Directly below is the place where the device will be put. The surface is covered with a special rubber (the black pad), which is able to absorb all kinds of vibrations, so we can be sure that the only noise inside this crate comes from the drive itself.

Disk WinMarks

The Business Disc WinMark of ZD’s WinBench 99 Suite shows that both Western Digital drives as well as the new Seagate Barracuda ATA II are prepared for all kind of office applications.

Quantum’s Fireball Plus LM drive is able to compete in the HighEnd Disc Winmark. Our reference drive, the Seagate Barracuda ATA, is clearly slower, which is due to the small cache size of only 512 KByte.

Content Creation Winstone 2000 – Windows 98 SE

Even though the Barracuda ATA II is at the top again, please be aware that the differences are very little.

Content Creation Winstone 2000 – Windows 98 SE

The same applies for the Windows 2000 benchmark run. The Western Digital drives are able to rock the boat this time.

Data Transfer Test: WinBench 99 Disc Inspection Test

Here you can see which drives are the fastest – and which one are the slowest. Some applications will run better is the difference between minimum and maximum transfer speed is not too big (e.g. Quantum).

The Seagate and Fujitsu hard drives are the clear performance winners here, while these numbers don’t always reflect real world performance, as we can see with the Western Digital drives.

CPU Utilization Test – Windows 98 SE

The fujitsu and Quantum drives consume a lot CPU time. Western Digital and Seagate are much better here.

CPU Utilization Test – Windows 2000

Again, Western Digital did the best work. Their drives’ CPU utilization does not reach such low values like SCSI, but 10% is still a great result.

Seek Time – Windows 98 SE

Seagate and Fujitsu got the best access times in Windows 98. The Quantum drive cannot hold up to them. Let’s take a look at the Windows 2000 results…

Seek Time – Windows 2000 Professional

Even though the Western Digital drives don’t seem to be fast performers, they got quite good results so far. Voilб: The Quantum drive is the best drive in our Windows 2000 access benchmark.

Disc Operating Noise

Our results are Decibel values dB(A). Mathematically speaking, a 10dB difference on power measurements is a 10X (1000%) increase in actual power. Hower, the human ear does perceive a 10X increase in audio power to be only twice as loud.

Disc Temperature

The Seagate and Fujitsu drives are quite hot models! Here, Western Digital did very good work once again, as the 15 GByte drive is still less than 40 degrees.

Installation of a Hard Drive

There is basically not much to say about the installation of a hard disc drive. Nonetheless, I want to emphasize some issues.

The faster your drive’s rotation speed, the more will be its heat dissipation. Thus it is very important to install your drive properly. Make sure that the drive gets in close touch with your computer case. Or in other words, make sure both sides of the drive bay are in close contact with the drive by using all four screws. That’s very important, as the case takes over the function of a heat sink.

Some of you may want to reduce the passing-on of the drive’s vibrations to the computer case by putting some material between the drive and the case. Please be aware that this will take away the only option of passive heat derivation to the computer case. If you still want to use insulating materials, I’d strongly recommend using a hard drive cooling system for drives running at 7,200 or 10,000 rpm.

Conclusion

I think it is extremely difficult to give a recommendation for just one drive. After reading this article you should know now that every drive has both benefits and handicaps. The Seagate Barracuda ATA II is the fastest one, but will likely make your case vibrate as well. The Fujitsu drive is very quiet and provides short seek times, but gets almost as hot as the Seagate drives. Quantum builds a decent drive as well, With only the Windows 98 performance letting it down. Western Digital provides two great drives as well, but performance freaks will complain about their basic performance.

I hope you understand why I’m not going to recommend any particular drive at this time, as choosing the right one depends highly on your personal requirements. If you want performance or capacity in the first place, I’d go for the Seagate drive. But you should spend some money for a hard drive cooling kit as well. Windows 98 users should avoid the Quantum drive, while database applications will run fast with the Fujitsu drive. You should also take either the Fujitsu or the Quantum drive if you want to get a quiet model.

Finally, all users who just don’t know what they want may take a look at the Western Digital drives. They are not the very fastest, but come with little CPU utilization, no annoying noises and acceptable temperatures.