The test, hard drives

The test

To reduce the number of hard drives tested we made two choices. The first was based on a hard drive capacity of 120 GB. This is currently the most sold on the single component market thanks to an excellent price/capacity ratio. Also 60 GB platters used to have significant differences depending on the platter (inferior access time on an 80 GB platter, for example). With standard capacity, 80, 120 or even 160 GB, differences are only in the number of faces and platters.

Our second choice was on the cache level, which is 8 MB for all hard drives. Performance gains brought by this cache are more than significant in the application area and its interest is limited for pure transfers. Price differences for the moment are minimal.

For our tests we used the H2Bench utility, which has theoretical (access time, cache reading/writing rate) and practical tests based on reading/writing reproductions in various uses (disc swap, software installation, file copy and the following applications Word, Photoshop and F-Prot). This test is made on the first 40 GB of the hard drive.

Tests were made on an ASUSTeK P4C800-E Deluxe motherboard. Hard drives were connected either to the ATA Parallel port or to the Serial ATA. There is inborn management of these two interfaces by the Intel ICH5 Southbridge. To activate Audio Acoustic Management, we used the Hitachi feature tool utility.

Hard drives

For theses tests we brought together 8 hard drives purchased from a retail outlet:

- Hitachi Ofskstar 7K250 120 GB, ATA, 8 MB of cache (HDS722512VLAT80)
- Hitachi Ofskstar 7K250 120 GB, SATA, 8 MB of cache (HDS722512VLSA80)
- Maxtor DiaMBnd Max Plus 9 120 GB, ATA, 8 MB of cache (6Y120P0)
- Maxtor DiaMBnd Max Plus 9 120 GB, SATA, 8 MB of cache (6Y120M0)
- Seagate Barracuda 7200.7 Plus 120 GB, ATA, 8 MB of cache (ST3120026A)
- Seagate Barracuda 7200.7 120 GB, SATA, 8 MB of cache (ST3120026AS)
- Western Digital WD1200JB 120 GB, ATA, 8 MB of cache
- Western Digital WD1200JD 120 GB, SATA, 8 MB of cache

To compare performance we added an IBM/Hitachi 180GXP:

- IBM/Hitachi 120 GB, ATA, 8 MB of cache (IC35L120AVV207-1)

For your interest we also talk about the Raptor Serial ATA 10 000 Rpm by Western on the second to last page.

Recognizing a hard drive

How to recognize the 8 MB of cache?

If all of these Serial ATA hard drives are equipped with 8 MB of cache, this is not necessarily the case with classic ATA hard drives equipped with 2 or 8MB. How do you know the cache size without using software? Simply by looking at the model number.

- Hitachi 7K250: 2 last figures of the number, 20 for 2 MB and 80 for 8 MB

- Maxtor DM+9 : 6th letter of the number : L for 2 MB, P for 8 MB

- Western Digital : 2 last letters of the hard drive name : BB for 2 MB, JB for 8 MB

- Seagate 7200.7 : 3 last three figures of the number :
40 GB : 2 MB only (ST340014A)
80 GB : 2 MB only (ST380011A)
120 GB : 2 MB (ST3120022A) or 8 MB (ST3120026A)
160 GB : 2 MB (ST3160021A) or 8 MB (ST3120023A)
200 GB : 8 MB only (ST3200822A)

How to recognize a hard drive with 80 GB per platter

This is more difficult. With Hitachi and Seagate it’s easy, the 7K250 and 7200.7 have inevitably 80 GB per platter (except for the Seagate 200 GB with two 100 GB platters). The situation with Maxtor and Western is different. The DiamondMax Plus 9 usually had 60 GB platters, and the transition to 80 GB platters was made progressively. Maxtor hasn’t officially informed us of the way to recognize these hard drives even if it was possible to find both on the market. Now, the transition is over and the problem no longer exists.

For Western Digital it’s even more complicated. The manufacturer has never released a line of hard drives based on platter size. Since its release, the 8 MB cache 120 GB ATA is named WD1200JB. This WD1200JB denomination, however, is also used for hard drives with three 40 GB platters, two 60 GB platters or two 80 GB platters (with one face disabled for one of them). It’s not possible for the final user to recognize any of them.

Western initially told us that all hard drives manufactured since the second quarter had 80 GB platters, but we noticed this wasn’t accurate. We found a WD1200JB manufactured in August and its performance showed it was equipped with 40 GB platters!

The hard drives tested here all have 80 GB platters and were manufactured in September. It would be in Western Digital’s interest to modify the name of their products according to platter capacity but as they’ve used this configuration for some time now this option is less likely.

Specifications

Specifications

For Hitachi, the higher platter density of the 7K250 increases debit rate. The maximum debit rate reached by this media is increased by 8%. Strangely, the sustained debit rate is unchanged on the technical sheet. We will see later that it’s not accurate. So apart from the higher platter density, only the number of switch on/off cycles changes from 40,000 to 50,000. Hitachi has officially increased the engine reliability for the long term.


It’s not possible to compare Maxtor hard drives since none have been released since the DiamondMax Plus 9, and the manufacturer doesn’t inform us on debit rates. One year ago, they announced an access time “inferior to 9 ms”, and now it’s 9.3 ms. Could this figure be due to the 80 GB platter upgrade?


For Seagate, improvements in the Barracuda 7200.7 are on two levels; significantly increased debit rate compared to the Barracuda ATA V, and a reduced access time. There is an increase of 1dBA in noise as a result.

Finally, regarding Western Digital’s WD1200, we can’t show the specification table, because even the tables on the Western Digital’s website are based on the previous versions equipped with 40 GB platters.

Physical characteristics

Physical characteristics

Here are a few pictures of the hard drives:


On a practical point of view, the Serial ATA Seagate hard drive is the only one of this comparative test which isn’t equipped with a 12V standard electric input in addition to the Serial ATA electrical input. So you will have to buy an electrical adapter if not provided with the hard drive.


The Serial ATA interface apparently isn’t inborn on any of the hard drives. Even if it isn’t visible on Western Digital’s hard drive, we know that they use an 88i8030 Marvel ATA to SATA bridge. Hitachi and Maxtor also use this bridge but it’s visable. The same controllers are used on ATA and SATA hard drives (Infineon and Ardent, respectively), the difference on SATA is on the Marvel chip level which is the chip closest to the port that it manages.


For Seagate, the ATA hard drive uses an ST controller (as shown on the picture) or Agere (like another disc in this test). The SATA hard drive combines this Agere controller with a LSI chip identical to the Marvell chips. None of these hard drives have a controller with inborn Serial ATA management.

Transfer rate

Transfer rate

We start with the average debit rate for the entire hard drive.


Logically, the 180GXP is last, because of its 60 GB platters. The performance gap with the fastest hard drive is quite small, 2.5-2.6MB /s. The Hitachi, Maxtor and Western 80 GB platters hard drives have equivalent performances. Only Seagate is slightly behind. You will also notice that the difference between Serial ATA and ATA is logically quite small, because the mechanisms are identical.

Now if we take a closer look at the advancement of the reading head. The X-axis is disc advancement expressed in GB and the Y-axis represents the debit rate measured in KB/s.


All hard drives start with debit rates between 55 and 60 MB/s, then at half way the debit rate falls to 45 to 50 MB/s. Until the 2/3 disc mark the debit rate remains over 40 MB/s. At the end of the disc only the Maxtor and Western have a transfer rate of over 30 MB/s. Looking carefully at the disc shows the 180GXP and 7200.7 are slightly slower. The Maxtor and Western behave almost the same and the 7K250s are faster in the beginning of the disc and slower at the end.

For writing results are identical, with a slight accentuation of the reduced debit rate for IBM at the end of the disc.

Seek time

Seektime

The next test is the measurement of seek time. In the table are included access times for reading and writing. To obtain a reliable figure we deactivated the cache in writing for this test.


In writing, only the Hitachi went below 13 ms, the 7K250 is 0.2 ms faster than the 180GXP. Western’s hard drives provide results close to the previous generation, still below 14 ms. However even with 60 GB platters with seek times approximately under 1 ms, the Maxtors aren’t able to reach 14 ms. We think that the density increase accompanies the increase of time required by the head placement for Maxtor.

You may notice that writing seek times are more significant. The difference is small for Hitachi and Western and bigger for Maxtor and Seagate.


Once the AAM is activated, seek times noticeably increase. To some extent, only the 7200.7 and Hitachi 7K250 have competitive seek times. It’s a plus compared to the 180GXP whose access time increased to 20ms like the other hard drives.

You may wonder why the SATA version is included in the graph without AAM and the ATA version is with the AAM. The difference in access time between these two hard drives led us to believe this function was activated on the ATA hard drive and not on the S-ATA version. The only problem with the current ATA and SATA firmware is the AAM function is hidden and non-accessible! We were able, however, to verify our hypothesis thanks to an ATA disc equipped with firmware 3.16 with AAM (not available by retail).

Cache transfer rate

Cache transfer rate

None of the hard drives were able to sustain 100 MB /s. To get close to this limit two discs had to be used simultaneously. The contribution of the Serial ATA and its 150 MB/s was only to transmission speed of information included in the hard drive. We measured the cache debit rate on the following hard drives:


The only two hard drives able to reach over 100 MB/s were the Maxtor DiamondMax Plus 9 and Hitachi Deskstar 7K250 (the ICH5 is restricted to ATA 100 and isn’t compatible with the ATA133). For Western and Seagate the transfer rate was equivalent in SATA and ATA. This was disappointing.

Application test performance index

Application performance index

The next test was the H2Bench application test, based on a reproduction of reading/writing made on the hard drive in various uses (disc swap, software installation, file copy and the following applications Word, Photoshop and F-Prot). This test is made on the first 40 GB of the hard drive. These results also depend on the test configuration and shouldn’t be directly compared with previous tests.


Hitachi was first and even the good old 180GXP was ahead of the new hard drives, despite a density of 60 GB per disc. The 7K250 provided the best performance. There was a 4% difference between the SATA and ATA version. This gap can also be due to differences on the firmware level.

Next were the Western and Seagate SATA hard drives. Maxtor’s hard drives had poor results probably due to the concession on access time. In last place was the 7200.7 in ATA version. This result is only logical because it has the highest access time due to a standard and non-modifiable AAM activation.

Formatted capacity

Formatted capacity

Here is the real hard drive capacity as is reported by any OS after formatting. The Western Digital and Seagate had the smaller capacity and the Maxtor and Hitachi had better results. The best result is still below the claimed 120 GB.


This result is due to the calculation method. Computer scientists and machines count in binary. For us a Kilo octet is equal to 1024 bytes and not 1000 as Kilo should be defined. For quite some time now, hard drive manufacturers changed their calculation method and use Kilo-byte = 1000 octets. Officially they are correct because since 1998 the international norm is 1 Kilo-byte = 1000 octets, 1 Kibi-byte = 1024 octets.

When a hard drive manufacturer sells a 120 GB hard drive, it sells you 120 Gigabytes and not 120 Gibibytes. Our OS continues logically in binary to count Giga as Gibi (Gibi pour Gigabinary, Mebi for megabinary and kibi for Kilobinary).

According to this Giga definition, the Western and Seagate hard drives reached 120.0 GB, the Maxtor 122.9 GB and Hitachi 123.5. The result is the same in the end as there will be more space with a Hitachi or Maxtor than with a Seagate or Western.

Hard drive temperature

Hard drive temperature

The next graph indicates hard drive temperature in use. Figures correspond to the hottest part of the disc after one hour of intensive use (ambient temperature 21°C).


There is a 3°C difference between the Western and Maxtor. In second is Hitachi’s 7K250 followed by Seagate.

Noise pollution

Noise pollution

The last test concerned hard drive noise levels. We took several measurements with a sound measuring device from 3 cm from the hard drive.


Of course, the Hitachi, Maxtor and Seagate with their hydro-dynamic (Fluid Dynamic Bearings) engines were in the lead. If Hitachi and Seagate had almost the same amount of noise, Maxtor is clearly ahead and offers truly silent hard drives in this area.

Western Digital with its ball bearing engine for 7,200 RPM hard drives came in last. The Western’s equipped with FDB engines are usually destined for the OEM market. Though difficult it’s also possible to find them on the retail market. Their extension for ATA with 2 MB of cache is LB, and PB with 8 MB of cache.

The next graph includes measurements during intensive use.


Without AAM (Audio Acoustic Management), in performance mode, the Maxtor hard drives were noisy even if they were almost silent in rotation only. This problem is resolved if the AAM is activated, but access time is 20 ms.

For Western, the noise level difference from rotation is reduced. However, even if the difference was small, starting from 45 dBA necessarily gives higher figures. Hitachi’s 7K250s had better results and these hard drives are the least noisy with or without AAM.

The Seagate 7200.7 in the ATA version with the non modifiable activated AAM is discreet, but provides reduced performances. The 7200.7 SATA is the noisier hard drive of this comparative test, and it is a first for Seagate. It’s a shame that the firmware provided with the disc doesn’t provide the opportunity to activate the AAM to reduce the noise level.

UpdateThe 7K250s have an ITF function (Idle Time Function), which after ten minutes of inactivity moves the reading head arm (this time laps may vary). If our test hard drives purchased in retail outlets didn’t produce any specific sound, this doesn’t seem to be the case for all 7K250s. Some users had hard drives that made noticeable noise in silent configuration in which the disc isn’t active during certain periods. This noise can last from 0.5 to 1.5 seconds.

Raptor & Maxboost

WD Raptor, an isolate case

We initially wanted to include a Western Digital Raptor to compare performances. This hard drive is the first ATA (serial) with a 10,000 RPM rotation speed. Unfortunately, the small capacity (36 GB) isn’t enough to use the H2bench applicative index, and our test necessitated 40 GB of disc space for correct processing.

It should be noted, however, that this hard drive reaches an average debit rate of 46.6 MB/s in reading/writing and an average access time of 8.5 ms for reading (8.9 ms for writing). The debit rate isn’t extraordinary but thanks to the reduced access time, it has no SATA competitor in terms of performance if used as a main hard drive.

Of course there are trade offs for higher performance. First of all, for the price of a 36 GB raptor it’s possible to purchase a 120 GB SATA 7200 RPM (the warranty is 5 years). Temperature remains reasonable at 51°C under the same conditions as other hard drives, however, the noise level was higher. The Raptor reached 47.5 dBA in rotation and 50.5 dBA during data access, which is relatively high. It’s up to you to choose the right hard drive according to your own requirements and budget.

Maxtor Maxboost

Maxtor used to provide until recently a free beta-test software, Maxboost. The final version of this software, designed to increase your hard drive performance with windows 2000 and XP, will be free of charge (contrary to what we said earlier). How does it function? Simply by creating a large hard drive cache in RAM memory managed by algorithm via the CPU.

You may have understood that this performance improvement is brought about by a large writing cache. However, at the same time part of your RAM CPU will be used. Another disadvantage is that you may loose data if the computer is unexpectedly turned off.

This initiative deserves some praise but it does have its downsides. The beta version isn’t compatible with the SMP whether it’s physical or logical as is the case with HyperThreading. At any rate, we would prefer manufacturers to concentrate their efforts directly on the hard drive card controller to increase cache size and improve the embedded cache controller’s algorithm.

Conclusion

Conclusion

The Hitachi Deskstar 7K250 is the one hard drive with a better performance than the others. Whether in ATA or Serial ATA version, the 7K250 provides the best results, especially due to its small access time. In addition it has low noise (besides the ITF problem on certain discs-see section on noise results) and has a slightly higher platter capacity than average. This series is faultless and we hope for Hitachi/IBM that there is no repeat of the 60GXP and 75GXP’s lack of reliability.


The Maxtor hard drives were disappointing. It seems that the DiamondMax Plus 9 with 80 GB platters has a reduced access time compared to the previous generation. They came in last place in terms of performance (with AAM deactivated). The low noise level during disc rotation could have given the DiamondMax Plus 9 a good grade in this area but this result is ruined by noise in access.

Seagate was the Dr. Jekyl and Mr Hyde of this test. One hard drive was the efficient but quite noisy S-ATA, and the other was the silent but worst performing ATA. It’s a shame that they didn’t give us the possibility to activate or de-activate the Audio Acoustic Management. One of Seagate’s advantages, however, is that it has the most reliable hard drives-according to figures provided by our retail partner LDLC.

The lack of information regarding the type of platters(40, 60 or 80 GB?) in Western Digital’s hard drives is more than annoying even if in the short term all production will be with 80 GB platters. Beside this, the hard drives provide great performances and are in second place. This good result is tarnished by a high level of noise due to ball bearing engines.

And what can we expect for 2004? We are reaching limits in terms of platter density and LMR type writing heads. Seagate recently announced the release of hard drives with 100 GB platters, which could be the latest evolution in this technology. PMR technology, which is in the last stage of development should permit a higher storage capacity of around 175 GB. In the meantime, manufacturers could work on reducing the access times of their ATA / SATA hard drives, as Western Digital did with the Raptor.