by Sergey Romanov , Nikita Nikolaichev
10/14/2003 | 09:35 PM
Last December Seagate announced a new generation of hard disk drives aka Barracuda 7200.7, having become the second manufacturer to reach the data density over 50Gbit per square inch. And the first one here was Maxtor, as you all know, which announced this data density in September already. It is very interesting that two last HDD generations from Seagate are only a little over five months away from one another. For various reasons Seagate decided to force the events and prepared the next generation of high performance hard disk drives for desktops (successors to Barracuda ATA V) in a really short period of time. This way the entire product range from Seagate has been slightly rearranged.
After they discontinued the 7200rpm SCSI solutions, Barracuda brand has been fully occupied by ATA HDDs, which automatically eliminated the necessity to us this “ATA” work in the model naming. They have also terminated the U Series family, and 5,400rpm HDDs have been transferred to Barracuda, which resulted into numeric indexes added to the name, such as “7200” and “5400”. And the generation name is now indicated with a digit after a point. As a result, the next generation after Barracuda ATA V is called Barracuda 7200.7.
In this case a question arises: and why is the fifth generation followed by the seventh? No, this is no marketing mistake, Seagate really has already had six HDD generations with 7,200rpm spindle rotation speed. The thing is that the very first model of this class (by the way, it was not only the first model from Seagate, but the first model for the entire industry) announced in October 1997, was manufactured under Medalist Pro brand name. Actually, it was the world’s first hard drive with fluid dynamic bearings (FDB), which allowed to combine the server spindle rotation speed with an acceptable noise level for a home system.
Only a year and a half later when the competition came into this market segment (and all HDD manufacturers were very eager to start working on the new technologies), Seagate marketing department got a great idea to take advantage of the already well-known in the SCSI market Barracuda brand name and to use it in the ATA market as well. The second generation of 7,200rpm HDDs from Seagate, which appeared in June 1999 was already called Barracuda ATA.
Later Seagate tried to stick to aggressive 6-month cycle for the new product announcements. The third HDD generation came into this world in January 2000. Barracuda ATA II acquired larger buffer increased up to 2MB and Ultra ATA/66 support. In September 2000 they released Barracuda ATA III, which boasted third generation fluid dynamic bearings, almost twice as high data density and new Ultra ATA/100 interface.
Starting with this product family, Seagate gave up the production of three-platter models. Once again the data density and the hard disk drives storage capacity doubled in June 2001 when Seagate launched barracuda ATA IV. This HDD family didn’t boast as much of a speed boost as the previous one, but instead brought a revolution in terms of noise parameters (even now that more than two years have passed already, it is still one of the quietest hard drives in the market). After that a period of calm set in. :)
Only a year later, in June 2002, Seagate prepared a replacement for this one of the most successful HDD families. Barracuda ATA V didn’t boast any brilliant performance characteristics, didn’t turn into a breakthrough in terms of the noise level, but enlarged the Seagate’s product range quite noticeably. Barracuda ATA V Plus acquired a larger 8MB buffer, and became the world’s first hard drive supporting new SerialATA interface. And now that only 5 months have passed a new HDD generation seems to be coming out. Does it mean that Seagate is coming back to a half-year cycle in terms of new product announcements?
I assume that the reason for this rush with the new generation launch was the recent activity on Maxtor’s side. However, you already know that Maxtor’s announcement appeared a kind of “paper” one (read our article called Real Maxtor DiamondMax Plus 9 HDD with 80GB Platters Reviewed! for details), because only one model aka 6E040L0 acquired a high-density platter. This is the eldest hard disk drive from the budget DiamondMax Plus 8 family (see our Maxtor DiamondMax Plus 8 HDD Family Review. Part I), therefore its arrival didn’t affect the market that much. It turned out that the hard drives with 80GB platters from two largest manufacturers started selling at about the same time, in February this year. After a while Seagate started a massive attack along all front-lines offering the customers a rich choice of models featuring four different storage capacities, two different buffer sizes and two types of supported interfaces.
The new family of Seagate’s hard disk drives is unusually numerous. It includes 9 models, which are already in mass production. In the table below you can see what the difference between them is:
OEM customers have a much richer choice, actually. In particular they can also order ST80013A model (80GB, 8MB buffer and Ultra ATA interface), and 7 modifications of SerialATA hard disk drives.
Starting with Barracuda ATA V the models with 8MB buffer and ATA interface get a “Plus” word in the name.
In order to get a better idea about how good or bad the new family is, we collected all nine officially manufactured Barracuda 7200.7 models and for a better comparison added the performance results for Barracuda Serial ATA V. To ensure that the results are all easy to read and analyze we will use the following naming legend:
As you see, the hard drives in our today’s test session feature different firmware versions, which will allow us to compare different firmware versions with one another. And due to the fact that we have so many testing participants, we will be able to evaluate highly objectively the performance of all Seagate Barracuda 7200.7 HDDs. The matter is that sometimes the review sample you get may not be quite OK, therefore with a big variety of test pieces today we are safe from these “incidents”. Moreover, when we will work on further reviews we will already know if the results obtained for this or that Barracuda 7200.7 model suit for comparison with competitors’ products, or if we have to retest it anew.
And now let me describe what changes have been introduced in the new Barracuda 7200.7 family.
The case construction which has already stood the test of time in two generations of Barracuda drives, has been inherited by the new 7200.7 family:
It is worth pointing out that Seagate is little by little replacing the polished top cover of the HDD case with the matt one, which seems to be more suitable for better heat take off.
The ever growing data density now allows storing up to 80GB of data on a single platter, that is why the storage capacity of a dual-platter drive can no longer be addressed in a standard LBA mode. Therefore, Seagate followed in Maxtor’s, WD’s and IBM’s footsteps and introduced LBA48 mode support. You can read more about the differences of these two addressing modes in our article called Neptune: Missing Piece of the Maxtor HDDs Evolution.
For some unknown reasons the new Barracuda drives do not support the already traditional acoustics management, although the user’s guide for SerialATA models indicates this option. This way, you will not be able to make your drive work faster or quieter, than it was set to during the production stage. But which is even more surprising, Serial ATA Barracuda 7200.7 supports UDMA6 data transfer mode, i.e. it is capable of transferring over 100MB of data per second! Although our express testing showed that the today’s only SerialATA controller, which allows achieving such a high speed is the one integrated into Intel’s ICH5 South bridge.
I was very much excited about this great feature of the new product and decided to undertake a detailed investigation of the HDD’s electronic components, especially since the PCB is no longer hidden behind the metal plate.
The electronic components on the PCB of ATA and SerialATA Barracuda 7200.7 are completely different. The first one uses 10252820 processor from Agere or STMicroelectronics (it turned out that these two manufacturers produce similar processors for Seagate!), while the second one uses 10258581. Having taken a really close look I discovered that the electronics is very similar to what we saw on Barracuda ATA V.
Although the previous generation HDDs used 10226836 and 10238949 processors for ATA and Serial ATA models respectively. Both of them use 16bit SDRAM chips with 6ns cycle time, which indicates a potentially possible data transfer rate of up to 333MB/sec. I wonder how big will the difference be between Barracuda ATA V and the new Barracuda 7200.7?
Our testbed configuration looked as follows:
We used the following benchmarking software:
All drives that support “quiet seek / normal seek” operation modes were switched to the fast mode by means of Hitachi Feature Tool. For WinBench tests, the drives were formatted in FAT32 and NTFS as a single partition with a default cluster size. We used Paragon Partition Manager for FAT32 formatting. The benchmarks were run seven times each; the maximum result was taken for further analysis. The HDDs didn’t cool down between the tests. The tests in Intel IOMeter were run in SequentialRead, SequentialWrite, DataBase, WorkStation, FileServer and WebServer patterns. If you are looking for the detailed description of these patterns, please, see our previous articles.
Let’s start with the easiest things: the ability to read data sequences. The funny thing is that even this unsophisticated test
Prepared a few surprises for us. Have a look here:
The first thing that catches your eye is the significant lag of Barracuda 7200.7 Serial ATA when it works with small data blocks. Interestingly enough, but Barracuda Serial ATA V didn’t have any problems like that and copes with thie type of read tasks not any worse than its ATA fellows.
The second surprise is the slower performance of 160GB ATA models compared with that of HDDs with lower storage capacity. They run almost evenly fast up to 4KB data blocks and then the elder models with ATA interface slow down (this graph is marked in red on the diagram above) while all the bothers retain the same pace. The only explanation I can think of here is the slower performance in case of 48bit addressing. Actually, Barracuda 7200.7 Serial ATA doesn’t have any problems like that, and the models with different storage capacity perform almost equally fast here.
During sequential writing the HDDs with different firmware show different results. And again we witness barracuda 7200.7 Serial ATA being strictly individual: its graph is absolutely unlike the other ones. It boasts better performance with the data blocks over 8KB and with smaller data blocks the leadership belongs to Barracuda Serial ATA V. The red color now indicates the results for firmware version 3.04: they are considerably lower than they are supposed to be. The firmware version 3.06 and 3.54 is already free from this bug.
The maximum performance of Barracuda 7200.7 is more than 30% higher than Barracuda ATA V, which totally corresponds to the data density increase. However, if we look into the specs list, we will see a very strange thing there. The claimed bit density has become only 9% higher according to the specs, and the remaining increase in the storage capacity has been achieved only due to higher per track density! The data in the specs is definitely wrong. By the way, all this reminded me of the times of Barracuda ATA III, it was also much faster than the previous generation models in “raw” read/write speeds.
Now let’s compare the average access times. The reading and writing are performed in 8KB blocks with random addresses (these are the particular cases for the DataBase pattern).
What do we see here? And we see a TENDENCY. The average access time demonstrated by the hard disk drives is directly dependent on their “coolness”. The cheapest models show the worst access time, the 8MB buffer helps improve the situation a little bit, but Barracuda 7200.7 Serial ATA drives simply work wonders here! Their access time appeared better than that of the previous generation model with “cut-down” platters (you can read more about the influence of this “cut-down” design in our article called Maxtor DiamondMax Plus 8 HDD Family Review. Part I). The difference between ATA and SerialATA Barracuda 7200.7 makes about 2ms, which is more than 15%!
I was trying real hard to find a reasonable explanation to this fact. As you probably remember, we have already investigated a similar phenomenon by Western Digital drives and arrived at a very interesting conclusion (read our article called Western Digital WD2500JB HDD: More than Drivezilla?!). However, the surface graphs of the today’s testing participants are very similar and the differences lie within the allowed measuring error. Tat is why the assumptions about varying amount of servomarks (to be more exact, sector indicators aka Sid) or different spindle rotation speed do not count. There appear to be not too many variants left:
We checked the first hypothesis by ear. You can hear the noise produced by SerialATA Barracuda 7200.7 during seek operations much better than the noise of its ATA fellow. Moreover, the latter generates some “unstable” and “clear” sound while the SerialATA drive sounds are very “rich” and “persistent”, which is very similar to the noise of Quantum drives, so dear to me :) This fact may be a proof of the second point. As for the supposition about the influence of the firmware version, it didn’t get any proof this time.
Which of the three points do you like most? I assume that both above described ways contributed to the positioning speed of Seagate Barracuda 7200.7 Serial ATA, because the average access time they demonstrated is beyond all praise and yields a little bit only to some IBM/Hitachi products.
During random writes the gap between ATA and Serial ATA Barracuda 7200.7 drives gets really huge, and Barracuda Serial ATA V proves even faster than all the others. Once again the comparative analysis shows that barracuda 7200.7 have a lot of room for further improvement.
Since all contemporary hard drives cache write requests and perform “background” writing, the average write access time gives us to understand how efficient the lazy write algorithms are. If we divide the average access time during reads by the average access time during writes we will obtain the efficiency coefficient:
Again we observe a very clear system: the HDDs split into several groups. The previous generation of Barracuda drives is notably ahead, having demonstrated relatively efficient lazy writing (compared with the competitors’ products). And Seagate Barracuda 7200.7 can hardly boast anything here: if the SerialATA versions still manage to show some acceptable results, the classical modifications suffered a total fiasco, and even a large buffer doesn’t save them from failure.
In our Samsung SpinPoint PL40 HDD Review we have already mentioned a strange peculiarity of Seagate Barracuda 7200.7 when some test results contradict each other. Despite pretty low results obtained in this IOMeter benchmark they still perform quite fast in SoundForge test from WinBench 99 test package, which also requires efficient lazy writing. Where does this contradiction come from?
Of course, we don’t have any definite answer, but dare suppose that it has to do with the type of oncoming requests. Our experience shows that the latest Seagate Barracuda generations are not optimized for multiple concurrent reads and do not cache the data already read once. In other words, having read a certain data block and transferred it to the PC, Seagate drives immediately forget about it. As a result, any non-sequential data requests interfere with the read ahead algorithms referring to the previous address. No multiple segmentation (although the specs claim it is there), the HDD copes well with only one data stream. If we suppose that the situation is absolutely the same during writing (we didn’t have the chnace yet to check this out), then everything becomes absolutely clear.
In SoundForge test the HDD is addressed more or less linearly, and in IOMeter the requests are random. As a result, lazy writing works fine in the first case, and in the second case the buffer is constantly re-associated. Do you know what the consequences of this situation are? The more fragmented gets the data, the lower gets the HDD performance…
The analysis of the HDD performance with different firmware versions by ATA Barracuda 7200.7 drives indicates that the differences between them are minimal.
A special Database pattern allows to study the peculiarities of HDDs firmware in greater detail. The requests size is 8KB. To evaluate the ability of different firmware versions to combine the reads and writes the reads/writes ratio has been changed from 0 to 100%. The measurements are all taken in operations per second.
All Barracuda drives show very smooth graphs, without any splashes or drops, although they still behave very differently. If Serial ATA drives are pretty loyal to the writes, the ATA modifications are completely indifferent to them. Barracuda SerialATA V manages to combine reads and writes very efficiently, while the new generation of Barracuda 7200.7 drives still needs to catch up with it on that.
Now take a look at the results of DataBase pattern from a bit different viewpoint:
Seagate doesn’t have any problems with scalability according to the workload, all the drives perform identically close.
As I have already said, the new HDD models are not very skillful in handling the mixed modes, unlike their predecessor. SerialATA Barracuda is over 20% faster than its Parallel ATA counterparts.
With 100% writes the picture turned completely different: Barracuda Serial ATA V is steadily ahead of its younger fellows.
Well, the first benchmarks results appeared pretty unexpected. Although the electronics of Barracuda 7200.7 is very similar to that of Barracuda ATA V, the latter outperforms the newcomer almost in all parameters. Having boosted up the linear speed (it is the first case lately when the read/write speed grew up as much as the data density) and improved the average access time, Seagate had to sacrifice other HDD features. It’s a great pity, by the way…
Now that we have already taken a close look at the HDDs performance in DataBase pattern, all other patterns shouldn’t offer us any new surprises.
In the WorkStation pattern the leadership belongs to SerialATA models, which dash far ahead of the others. The 8MB buffer of Barracuda 7200.7 Plus hardly tells on the results here.
SATA models are again more preferable for FileServer type of workload, and the large cache-buffer doesn’t affect the performance at all.
Since there are almost no write requests in a classical WebServer pattern, everything depends on the data seek speed. The slight fluctuations in the lower part of the diagram are most likely to be caused by some chance factors.
With the help of ZD WinBench99 we can evaluate the performance of our testing participants in a few typical applications. Even though some of these applications have already become outdated and the data volumes involved into the testing process are too small for today, the results obtained in WinBench tests are still pretty illustrative and useful.
At first let’s compare the results obtained in FAT32 file system.
In Business Winmark the HDDs ran very close to one another and their results fully correspond to their positioning in the market. Larger buffer allows improving the performance quite tangibly. Barracuda 7200.7 Plus most probably managed to defeat Barracuda 7200.,7 SerialATA because of the differences in controller drivers for Promise Ultra133/TX2 and SATA150/TX2.
High-End WinMark showed a slightly different picture. The previous generation represented by Barracuda SATA V rolls back to almost the very last position, and the performance boost granted by a larger cache-buffer makes 9%, which is not as high as we expected, actually. Let’s analyze some discrete results to find answers to all these questions.
Hm, this is nice. In AVS the results of all HDDs look like a bunch of neat stairs, though you still can distinguish between the groups. It looks a sif a lot of factors are involved in the game here. ATA HDDs with 8MB buffer are ahead of all, outperforming the models with 2MB buffer by up to 18% in some cases. SerialATA Barracuda drives of both generations perform very closely, which once again proves how little the “raw” speed (the data transfer rate to and from the platters) matters for the actual performance of the drives.
In FrontPage the buffer size doesn’t matter at all. In fact, I do not see any tendencies here: it seems to be more of a controller test rather than a HDD test.
Microstation indicated that ATA models with 8MB buffer do much better here. They outpaced their 2MB counterparts by more than 22%. SerialATA models turned out to be in between.
And for Photoshop the physical features of the drives are much more important. Therefore, the laurels were won by the models with the shortest average access time, and Barracuda ATA V lost because of its relatively low data transfer rate to and from the platters.
Another Adobe product is also very dependent on the read and write speeds, although the buffer size is also important here.
Well, it seems to me that the SoundForge results are primarily dependent on the stars. At least I have no reasonable explanation of the 20% performance difference between almost identical Barracuda 7200.7 Serial ATA drives. Nevertheless, I have every right to state that Serial ATA drives are the winners here. I would also like to stress the almost 20% advantage of the Barracuda 7200.7 Plus solution over the models with 2MB buffer.
And in conclusion we will look at Visual C++ test, which clearly indicated: Barracuda 7200.7 SerialATA is a little faster than Barracuda 7200.7 Plus. At the same time, these two hard drives outperformed their counterparts with smaller cache-buffer by about 12%. The previous generation hard disk drives are again in the very end of this race.
Now let’s pass over to the tests in a more advanced file system: NTFS.
In NTFS the results of High-End WinMark appeared even more systematic, and SerialATA HDDs managed to surpass their ATA fellows. I would like ton especially praise the previous Barracuda model, which managed not only to improve its performance in all applications, but also to win in two of them: Business WinMark and FrontPage.
Here the indisputable victory belongs to SerialATA. Judging by the closeness of all SerialATA HDDs results, I dare assume that they owe this victory to the controller. Anyway, only the models with 2MB buffer appeared slower here.
Another victory of SerialATA drives.
ATA hard disk drives manage to retain leadership here with very much effort. If it hadn’t been for the really heroic performance of the ST120026A model, Barracuda Serial ATA V would have got one more point :)
The situation has hardly changed in Photoshop compared to what we have already seen in FAT32 tests. Serial ATA is still ahead. Though on the whole, the results look better organized this time.
The same situation can be observed in Premiere. Serial ATA drives are only a few percents slower than their ATA analogues with 8MB buffer. Just like in FAT32 Barracuda ATA V loses to Barracuda 7200.7 in both Adobe applications. It is probably because of the low read/write speed…
In NTFS the SoundForge results are no longer so complicated. It is interesting that all the drives retained the ranks they won in FAT32 except for Barracuda 7200.7 Serial ATA 80GB. Well, it seems that SoundForge results do have a certain system behind them, but I am still unable to perceive the idea. :)
Well, the situation here remained the same as in FAT32. Only barracuda 7200.7 Plus 80GB failed to show good results. SerialATA HDDs demonstrate very clear advantage here.
Let’s sum up a few things:
The HDD makers pay more and more attention to streaming operations lately. Copying files is one of them. With the help of our specially designed FC-Test utility we test the HDDs performance during file copying for different file sets. The results of FC-Test program are always very illustrative. So, let’s have a look at our today’s testing participants and their performance in FC-Test:
First of all we create files on the HDDs. These results allow us to evaluate the write speed of the testing participants. Since the results appeared the same for both file systems, we will offer you only one diagram:
Well, this is very interesting, I should say! It is not at all surprising that a model with 8MB buffer outperformed a model with 2MB buffer, however Barracuda 7200.7 Serial ATA turned out much faster than even Barracuda 7200.7 Plus! And it hardly yields at all to the last Barracuda Serial ATA V even though its data transfer rate to and from the platter is not that high! FC-Test clearly illustrates the advantage of Serial ATA HDDs, which we saw in Intel IOMeter lazy write benchmarks.
During regular reading the firmware peculiarities and buffer size usually do not affect the result. The determinative role in this case play the physical read speed for large files and fast electronics for small files.
This is exactly what happened this time: all the HDDs are based on the same platters that is why we see the difference only in mixed patterns, where the average file size is not too big or too small. In these patterns (namely, Install and MP3), Serial ATA HDDs are defeated by the classical ATA ones. However, there is nothing to be surprised at, if you remember the results of Intel IOMeter test, namely for SequentialRead on small data blocks.
However, in NTFS file system Barracuda 7200.7 Serial ATA managed to regain their credibility. Again we see that the controller or/and the HDD are optimized for professional file system.
Now let’s pass over to the tastiest bit: to actual copying. We will start with copying the files within one partition.
As we have expected, Serial ATA HDDs managed to use their trump: more efficient lazy write algorithms. However, Barracuda 7200.7 Serial ATA is not that immensely ahead here, while the larger cache-buffer of Barracuda 7200.7 Plus ensured an almost 30% copy speed boost. Despite the lower physical speed, Barracuda Serial ATA V easily outperformed Barracuda 7200.7 with a 2MB buffer.
During file copying into another partition, the data seek speed becomes more important that is why the advantage of Barracuda Serial ATA grows bigger. However, this success is nothing compared with the results obtained by HDDs with larger buffer. This time the copy speed of HDDs with 8MB buffer grew 50% higher!
If you remember, the results of Intel IOMeter tests showed that the results of Barracuda 7200.7 are not always better than those of Barracuda ATA V. In some cases the Barracuda Serial ATA drive got slightly ahead of Barracuda 7200.7 Serial ATA despite the faster read/write speeds and heads positioning speed of the latter. Therefore, I dare conclude that Barracuda 7200.7 didn’t acquire any of the firmware enhancements made to the Barracuda ATA V during the product “fine tuning” stage.
This is probably the reason why the newcomer didn’t manage to significantly outperform the predecessor in some real tasks, such as copying, for instance. I sincerely hope that the situation can be improved on the software level by reflashing the firmware version. By the way, ATA Barracuda 7200.7 models achieved the best results with firmware version 3.06. firmware version 3.54 also proved very efficient.
Now that you took a closer look at the features and performance peculiarities of the new Seagate Barracuda 7200.7 HDD family, I would like to share my subjective opinion about the product and draw some conclusions.
I would like to point out that this Seagate family doesn’t boast that revolutionary quietness, which was a big advantage of Barracuda ATA IV hard drives. The rotation noises as well as the heads positioning noises have become much more audible and I can’t find a suitable explanation for that. Nevertheless, Barracuda 7200.7 don’t yield to any of the competitors in the noise level. Only its Serial ATA modifications are somewhat louder. But this is not at all surprising: fast seek cannot be noiseless. I was a little bit upset about no acoustics management options, though.
As for the heat dissipation of contemporary high-speed HDDs, there is not much to say here. If the single-platter Barracuda 7200.7 HDDs remain a little warm during work, then the larger models should better get some extra cooling. Of course, they are not as hot as irons, but even a little extra case ventilation will significantly reduce the risks of losing an expensive HDD and priceless data.
At first, when I just got acquainted with the first Barracuda 7200.7 models, I didn’t like them that much. Their performance left much to be desired, their reliability was very doubtful, and noise level - far from perfect. As I was working on this review, I couldn’t help wondering at what I saw. I was mostly impressed by the average access time of barracuda 7200.7 Serial ATA. Moreover, their more advanced firmware (featuring pretty efficient lazy writing) makes me suppose that there are two different Seagate developer teams working on ATA and Serial ATA drives, because these drives appeared so different. The higher price of Barracuda 7200.7 Serial ATA is absolutely justified, and should be considered a good option for a serious system.
As for classical Barracuda 7200.7 models, these are just good products. Since the launch Seagate has been working hard on the firmware, so that it could be competitive. They managed to do it more or less well, but the drive still has no tidbit about it.
P.S.: When we were working on this article, we managed to get hold of Seagate Barracuda 7200.7 Serial ATA with a new firmware version. The results we obtained were really surprising. Stay tuned! :)