This is the last hard drive roundup this year. The leading manufacturers have released quite a lot of new models, it is only IBM Deskstar 120GXP which is still lacking, but this model will likely to appear only by the end of January. So, here is what we have today.

Maxtor

Maxtor D540X-4D

The D540X line has in fact two models - of Maxtor and Quantum. The Maxtor's model marked with "D" is more popular, therefore it was chosen to be tested in our lab. A standard Maxtor's case and a signal processor (DSP) are the distinguishing features of this model. The Maxtor Data Protection System and Shock Protection System, as well as the Acoustic Management System, are provided. The line consists of 20, 40, 60 and 80 GB drives. The record density is 40 GB per platter; a spindle speed is 5400 rpm. We tested a 80 GB drive.

Maxtor D740X-6L

The photo shows a model developed by Quantum. As you can see, it looks entirely different. The most important innovation is Ultra ATA/133 interface. It is the only model with 7200 rpm speed which supports this protocol. Anyway, the only model with a speed of 5400 rpm with Ultra ATA/133 support is produced by Maxtor. Here we have Maxtor Data Protection System and Shock Protection System. The line includes 20, 40, 60 and 80 GB drives with a record density of 40 GB per platter. We tested the most sizeable model.

Samsung

Samsung SP8004H

The SP8004H is a drive with a spindle speed of 7200 rpm. Its record density is 40 GB per platter and a buffer of 2 MB. The line has 20, 40 and 80 GB models, and we tested 40 GB and 80 GB ones.

Western Digital

Western Digital Caviar 1000BB

The WD Caviar 1000BB is the largest drive in our lab. This drive has a nonstandard record density: 3 platters and 6 heads instead of 5 heads as it should be with 40 GB density and a size of 100 GB. Probably the engineers think that 2 heads per platter are better; their Caviar 1200BB drive of 120 GB in size has the same combination, and its record density is 40 GB per platter. One more WD's drive we have is WD Caviar 800BB; it also has three platters and six heads.

The complete characteristics of the drives can be found in this table, and now we are moving to the tests.

Tests

In this test we decided to extend our testing technique by including the HD Tach 2.61 program. First of all, it measures an average read/write speed to/from platter, which is quite an interesting parameter. Secondly, it is interesting to compare access times and CPU utilization values obtained in different tests. Besides, the only controller I had at the time of the tests - ATA/133, was integrated into a mainboard, so I had to carry out the tests on another testbed:

• Processor - AMD Athlon XP 1800+ (1533 MHz), Socket 462
• Mainboard - Iwill XP333-R, Ali Magik 1 chipset (North Bridge – M1647 rev. C1, South Bridge – M1535D+ rev. A1)
• Controller - HighPoint HPT372 (Ultra ATA/133 RAID)
• RAM - 256 MB PC2700 DDR SDRAM DIMM Kingmax, CL 2
• Video card - ASUS V8200 (GeForce3)
• Hard drive - Seagate Barracuda ATA III, ST340824A, 7200 rpm, 40 GB
• CD-ROM - ASUS 50x

Well, it is not very good, but I think that the results depend mainly on a chipset and the HPT372 controller. Furthermore, as the Maxtor D740X was developed by Quantum, changing of an acoustic level with the AMSET utility doesn't affect an access time, and therefore, test results. Probably, integration of the Maxtor Acoustic Management technology into the Quantum product wasn't successful.

And now let's turn to the results (since the scores of both Samsung drives are very close we publish only those of the SP8004H). The diagrams also show the results of discs with a recording density of 40 GB per platter.

Access time and CPU utilization – 7200 rpm
(Ziff-Davis WinBench 99, HD Tach 2.61)

The Maxtor D740X is a leader in access time; in the ATA/100 mode the results are better, but in the ATA/133 one they are closer to the specifications (average seek 8.5ms + average latency 4.2ms = 12.7ms). I think it is caused by the HighPoint controller. The rest of the drives show almost equal results (a little worse than the specified), and the Seagate Barracuda ATA IV finishes the last. The results obtained with the HD Tach do not differ much from the WinBench, that is why in future we can use only Ziff-Davis. The CPU utilization results prove my opinion. The HD Tach results are worse, and the score of the Maxtor D740X in the ATA/133 mode is no good at all. The controller is external, but it should never take half of the CPU resources.

Exchange rate – 7200 rpm (Ziff-Davis WinBench 99, HD Tach 2.61)

The WD1000BB falls just a bit behind the leader Seagate Barracuda ATA IV in a read speed from a platter, but it recouped its losses in the read speed in the end of the disc. These results show that a decrease in the record density per platter of the WD1000BB is caused by a less number of tracks. The WD800BB is the last, while the Samsung and the Maxtor have successfully surpassed a level of 40,000 tb/s.

Maxtor D740X

Samsung SP8004H

WD1000BB

WD800BB

The difference in the read speed in the beginning of the disc in case of the Maxtor D740X can be explained by a saw-like graph. The WD1000BB has a larger saw, but the results are excellent - a read speed is more than 40,000 tb/s on the most part of the disc! The graph of the Samsung looks very pleasant.

In a write speed the Samsung falls behind. As you know, drives of this company have some problems with algorithms of operation with a cache, and this is one more proof.

Ziff-Davis WinMarks 99 – 7200 rpm

Now let's see how our participants cope with standard applications.

In the High-End WinMark the Maxtor D740X catches up with the Seagate Barracuda ATA IV in the ATA/133 mode, though it was done at the expense of applications working with small-size files. Take a file that entirely go into a buffer, add ATA/133 and a Windows cache and you will get 182300 bytes per sec in the MS FrontPage 98!!! On applications working with large files the situation differs, see the table. This is also explains leadership in the Business Disk WinMark - all files are rather small there. The WD1000BB performs very good in a fair fight as well. The WD800BB is a good player as well, in contrast to the Samsung.

This file system is more complicated, however, the Maxtor D740X performs quite good here, while the WD1000BB falls behind them.

Well, the new-comers - Maxtor D740X and Western Digital WD1000BB prove they are good players but they are still unable to dethrone the Seagate Barracuda ATA IV.

Ziff-Davis WinBench 99 – 5400 rpm

As the results of the Maxtor drives do not depend on whether "off" or "fast" was chosen for acoustic management, we will show the results only for "off" and "quiet" modes.

Well, the leader is the Maxtor 541DX with a record density of 20 GB per platter (only one head is used), that is why we can consider that we have a new leader among 40 GB-per-platter drives.

Maxtor D540X-4D

The graph of the Maxtor is almost ideal.

The quiet mode of the D540X-4D makes an access time greater (as it must be), but with the acoustic management disabled the new drive from Maxtor takes the second place with the result being much better than the specified (average seek 12.0ms + average latency 5.5ms = 17.5ms).

The scores of the HD Tach are again bad.

The new drive from Maxtor is leading here, with the results being very close in both modes.

The complete results can be found in the table.

Intel IOMeter – 7200 rpm

We got used to the fact that the Samsung sweeps the field in case of heavy loads. The WD1000BB performs good as well. The Maxtor D740X loses in the ATA/133 mode, as I think, because of the HPT372 controller.

Intel IOMeter – 5400 rpm

Here we have again a new leader! By the way, the character of the Maxtor's curves do not change, it just rises up or drops down. Besides, the difference between both acoustic management mode is still very small.

The complete results can be found in the table.

Conclusion

First of all, let me say a couple of words about 7200 rpm drives. The Western Digital Caviar WD1000BB turns to be a very good drive. It has a large size and shows very good, balanced results. The Maxtor D740X performs good as well, but usage of the ATA/133 mode is not very effective. When drives of over 137 GB become widely available (remember that Fast Drive aka Ultra ATA/133 and Big Drive technologies come together), usage of such controllers will be justified.

As far as 5400 rpm drives are concerned we have a new leader. It is interesting that IBM doesn't plan on drives with such a spindle speed in the near future, while Seagate will reportedly release a new model only by the summer. That is why this leader will probably remain such for a long time.