by Alexey Volkov , Nikita Nikolaichev
05/09/2004 | 10:05 AM
After receiving your great feedback on the mega roundup of 80GB hard disk drives we unhesitatingly moved on to test devices with the most attractive price-to-performance measurements. As you probably know or guess, this time we are going to check out HDDs with 120GB storage capacity.
<%BANNER[article]%>The number of participating drives has increased over the previous roundup – we have compiled files on 22 devices! The qualitative aspect of the today’s roundup should hopefully be higher along with the quantitative one. We will offer you testing results of a SerialATA drive which Samsung had equipped with an 8MB cache buffer as well as the results of the 7200rpm SATA drive from Western Digital. Speed matters as we learned in our previous roundup: the WD740GD model nearly spoiled all the fun in our first article, having won in about all tests and completely discredited all other devices. Today we don’t have such an unquestionable leader – we will be seeking for the best among peers.
Like in our previous test session, we took same-capacity drives, so some makers are represented by devices from different generations, but this fact only contributes to our desire to present a thorough picture of this class of devices.
So let’s first take a look at the technical characteristics of the drives as specified by their manufacturers and cast a glance over each of them.
Hitachi supplied us four drives of three generations (120GXP, 180GXP and 7K250) which is an indication of Hitachi/IBM’s having long passed over the 120GB capacity barrier. To our bitter regret, we couldn’t get a 120GB drive from the Deskstar 7K250 series with the SATA interface and an ATA driver of the same generation with an 8MB cache.
Maxtor, on the contrary, is represented by drives of one and the same generation, DiamondMax Plus 9, but they are so different!
We wrote it numerous times that you can find drives with platters of difference capacity among DiamondMax Plus 9 products. Considering that the drives may have different amount of cache memory, we have 4 models and that’s not all yet! The series includes drives with ATA as well as SerialATA interfaces! Don’t get dizzy, there are only 6 drives in total (there are no SATA drives with a 2MB buffer in this series)!
So our roundup includes all four modifications of Maxtor’s 120GB ATA hard disk drive and one SATA device.
It’s simpler and at the same time more complex with drives from Samsung. Thanks to the adaptive formatting technology, each drive from Samsung is a unique device with its own zone map (we are preparing a separate article concerning this technology).
Our today’s test session includes three HDDs from Samsung: an ATA drive with a 2MB buffer, a SATA drive and a 5400rpm drive (the only drive with such low spindle rotation speed).
The Seagate team consists of three drives of the last generation (Barracuda 7200.7) plus an older device (Barracuda ATA V). Well, the “oldie” is equipped with an 8MB buffer – maybe this barracuda will be swimming about for a while?
Western Digital has the largest presence in this roundup. Don’t think we favor this company over the others, it wouldn’t be true at all. It’s just because there is absolute uncertainly about the capacity of platters in WD drives. According to the specs of the Caviar and the Caviar SE drives of 120GB capacity, as published on the Western Digital web-site, these drives have three platters. Considering that the number of read/write heads is 6, we may infer that the platter capacity is 40GB.
Back in early 2002 we had no doubts about the specs – 40GB per platter was the norm then. But now it’s hard to believe that WD1200BB and WD1200JB have three platters.
We have test results for different WD1200BB and WD1200JB drives: some were produced in early 2003 and others are nearly fresh (December of 2003). Judging by the linear read speed, they are not made on 40GB platters! We will return to this topic later, right now the table contains our assumptions about the real platter capacity.
We had to use two controllers as we had hard disk drives that connect across two different interfaces. So we chose two controllers from the same manufacturer:
The testbed was configured as follows:
The drives had the following firmware versions:
We used the following benchmarking software:
We wrote the Maxtor drives “through” to avoid the forced write checking (click here for details).
We formatted the drives in FAT32 and NTFS as one partition with the default cluster size (FAT32 formatting was performed by Paragon Partition Manager). We ran the tests seven times each, taking the best result for further analysis. The HDDs didn’t cool down between the tests. For FC-Test we partitioned the drive into two logical volumes, 32GB each. For Intel IOMeter tests, we used Sequential Read, Sequential Write, Database, Workstation, Fileserver and Webserver patterns. You can refer to our previous reviews for details about the patterns.
The Database test opens up the show. This pattern reveals the drives capability to process a mixed stream of requests for reading and writing random-address 8KB data blocks. By changing the ratio of write and read requests we can estimate how well the drive is sorting the requests out.
The drives are grouped according to their manufacturer to make the table smaller:
For better readability, we draw diagrams for different workloads.
The drives from Hitachi behave much like each other under a small load (one request). When there are many write requests (more than 50%), the HDS 722512VLAT20 starts lagging behind the “older” mates. This situation lasts until 90% writes, when the drive “realizes” that there are so few read requests that it’s not necessary to allot much cache memory for them, but better reserve it for lazy writing.
We increase the workload to 16 requests.
The story repeats: Hitachi HDS722512VLAT20 is quite indifferent to the increase of the writes share until there are 100% of them in the queue. In this mode, however, it is the fastest of all. It seems like it has more segments in the buffer, although a small overall buffer size (compared to the IC35L120AVV207-1).
Now, the workload is maximum, 256 requests.
The drives are faster in this test and keep on their speeds throughout the test, save for the HDS722512VLAT20.
Next goes the Maxtor team.
And the graphs:
All the ATA drives are nearly identical under the linear load, while the SATA drive is somewhat slower.
The lag of the SATA drive is more noticeable under a workload of 16 requests. The rest of the drives show similar speeds.
Under the maximum load, the SATA drive caught up with its ATA mates.
Now, let’s watch the Samsung team getting through this test:
The two drives of the new SpinPoint P80 series are practically identical, although connect across different interfaces. The SV1203N is a bit slower than them, because it has a lower spindle rotation speed, 5400rpm. At the same time, you see that drives from Samsung with different rotational speeds similarly react to the increase of the writes share. It’s clear we have no artificial “slowing-down” of the junior product here.
The situation remains the same under a higher load. The drive with the lower rotational speed quite predictably loses in performance to the pair of 7200rpm drives. In the leading duo, the SP1213C is a little faster than the SP1203N in modes with a high percentage of writes. This is no wonder as the SP1213C has an 8MB buffer, while the SP1203N has four times less of cache memory.
It’s again the same under the maximum workload. The SP1203N is a little faster in modes with a small percentage of writes, while the SATA drives speed ups when there are more writes in the queue.
The next manufacturer is Seagate:
And the graphs:
Unlike the drives from other manufacturers, we can single out a leader among Seagate’s products. It is the ST3120026AS model!
As you remember, Seagate sharply segmented the Barracuda 7200.7 family into three sub-families:
Products of the first two sub-families differ between themselves in the amount of cache memory. The 8MB buffer of Barracuda 7200.7 Plus drives really boosted their performance across a number of tasks.
Meanwhile, the drives of the third sub-family had a much better access time, although were loud at work… Well, you had to choose between speed and silence. Products of the Barracuda 7200.7 SATA series were intended for powerful workstations and entry-level servers, so the quietness was sacrificed for performance.
Among other drives, I’d like to single out the ST3120024A model (Barracuda ATA V), which was fast enough in modes with a high writes percentage.
The drives remained on their respective positions under this workload. Curiously enough, the ST3120024A lost to others at 100% reading, but took its second place after that to never leave it.
So, let’s push the load to the maximum.
The drives split into two pairs under the 256 requests workload: the ex-outsiders ST3120026A and the ST3120022A were the leaders at 100% reading, but lost to the other pair of drives at 20% writes.
The alphabetically last, but not least in importance, Western Digital comes.
All drives from WD show nearly the same performance under this workload, only the WD1200JD breaks away from the others, showing the same dislike for writes as the Hitachi HDS722512VLAT20. However, like the Hitachi drive, the WD1200JD catches up with the group at high percentages of writes. Note the low efficiency of lazy write algorithms of the WD1200BB/60 – it seriously lags behind the others when there are many writes in the queue.
Now, the workload becomes higher.
The graphs have the same shapes as under the smaller workload.
IOMeter is sending a stream of read/write requests with a request queue depth of 4 to the hard disk drive. Every minute, the data block size changes, so we get the dependence of the linear read (write) speed on the size of the data block.
Click to enlarge
The worst results are highlighted red; the best results are highlighted with blue.
The table suggests that the drives from Hitachi show the highest speed, although on different block sizes. The Seagate ST3120024A was the slowest on big blocks (it is the oldest drive after all), while the Seagate ST3120026AS was surprisingly slow on small blocks.
As you understand, it’s unreasonable to put all the results into one diagram, so we again divide the drives into groups according to their manufacturer.
You can see the graphs by clicking the following links:
Now, let’s see how these results change in the Sequential Write pattern.
Click to enlarge
The outsider on large blocks remained the same, while the WD1200BB/60 was the slowest on small data blocks. Curiously enough, we have three best drives in this test that come from three different manufacturers: Maxtor 6Y120P0/80, Samsung SP1213C and Hitachi HDS722512VLAT20.
You can view the diagrams below:
Now we’re going to check out patterns that simulate the work of a typical server disk subsystem.
Let’s compare the drives by their averaged performance (the average of the drive’s speeds at four loads – 1, 4, 16 and 64 requests).
The Hitachi team occupies the top of the table, followed by drives from Western Digital and two Samsung HDDs (SP1213C and SP1203N). Two drives from Seagate are at the bottom of the table, surpassed even by the SV1203N model from Samsung, which has smaller spindle rotation speed!
The Hitachi team goes unrivalled, like in the previous test. Two Samsung drives outperformed the group of WD devices and the rest of the table is practically the same. Well, there is a change that should certainly be emphasized. The Samsung SV1203N is the last here. It means that the “failure” of the two Seagate drives in the File Server pattern is not because of their high access time (we’ll see it later), but rather because of their disdainful attitude towards lazy writing. This is perfectly seen in the graphs of the drives which they drew in the Database pattern.
This pattern abounds in write requests which will probably bring some changes into the ranks. We also calculate a performance rating for each drive in the Workstation pattern, but with a different formula.
Yes, the leaders have changed: now drives from Western Digital take top positions (WD1200BB/80 and WD1200JB/90*). They are closely followed by the Samsung SP1213C, while the Hitachi drives are now only on the fourth and fifth places. Note, though, that the drives, save for a few outsiders, form a dense group, with very similar results.
Now we reduce the address space to 32GB and run the test once again.
Speeds have grown after the reduction of the working area to 32GB. The Samsung SP1213C drive has been emerging in the leading trio from time to time and now it has reached the top position. The Seagate ST3120026AS, previously unnoticed among leaders, puts on a surprisingly good performance here.
This is the last of synthetic Intel IOMeter tests, we can now go over to WinBench 99.
We use the WinBench test to check out the hard disk drives in the “desktop PC” mode. We format the disk into the NTFS file system with the system tools (the default cluster size is 4GB) and into FAT32 using Paragon Partition Manager (the cluster size is 32KB). We also perform our tests on the 32GB capacity in NTFS and FAT32 file systems (partitioning the drives with the standard Windows 2000 Disk Manager).
First, let’s check out the results that don’t depend on the file system as they refer to physical parameters of the devices. Average access time comes first:
Four best times go to Hitachi! Now it’s clear why they are so good in File Server and Web Server patterns! The drives from Seagate have the highest access times – this is the price you have to pay for their quietness. On the other hand, the drives from Samsung have better access times, but are anyway no louder than the Seagate Barracuda 7200.7 (subjectively, they are even much quieter!)
Now, let’s compare linear read speeds:
The Samsung SP1213C was the best in this test, being the only drive to overcome the 60,000KB/s barrier. Of course, the adaptive formatting technology employed by Samsung allows increasing the data density per track, thus increasing the linear speed, if the combination of the platter and the head permits. At the same time, no one can guarantee that the linear speed will be, say, 60MB/s with a randomly-taken sample – it may be better or worse…
You see that drives with 80GB platters are at the top of the table, while the 5400rpm drive from Samsung and the Seagate Barracuda ATA V are the worst of all.
Traditionally, we offer you transfer-rate graphs we got with the help of WinBench 99.
The tables were very clumsy, so you can watch them one by one by clicking the following links:
Hitachi:
Maxtor:
Samsung:
Seagate:
Western Digital:
Let’s now examine the FAT32 results in two integral tests: Business Disk Winmark and High-End Disk Winmark.
Two Maxtor 6Y120P0 drives on 60GB and 80GB platters both show the best results in High-End Disk Winmark. The Samsung SP1312C is again among the leaders – stability is a sign of class?
The WD1200JB/60 won the Business Disk Winmark test, making up for the lack of Western Digital’s drives in the upper part of the table. The Hitachi IC35L120AVV207-1 looks appealing, too.
Well, we won’t sing again the well-known tune about the influence of the cache size on results in WinBench.
Now, let’s deal with NTFS:
The Maxtor 6Y120P0 on 80GB platters suddenly fell down to the sixth position, but the other Maxtor, with the SATA interface, substituted it. Interestingly, the Samsung SP1213C couldn’t hold its third place, yielding it to the Hitachi IC35L120AVV207-1.
Now, the same tests, but on a capacity of 32GB.
The reduction of the disk size in FAT32 didn’t practically affect the results. Only the Samsung SP1213C lost two positions and found itself on the fifth place. The Maxtor drives with an 8 MB buffer have no rivals in High-End Disk Winmark, while WD drives rule in Business Disk Winmark.
Like in the previous case, there are no surprises. Drives from Maxtor cap the table, while the Samsung SP1213C regained one position…
That’s the most exciting part of our roundup. We run FC-Test according to our standard methodology: we created two logical volumes, 32GB each, on the HDDs and formatted them in NTFS and FAT32. We created a set of files on the first logical volume and read it, then copied it into a folder on the same volume (i.e. within one partition) and to another partition. We use five sets of files:
NTFS starts the tests.
Due to the number of reviewed drives, we discuss the results in detail pattern by pattern. So we start out by creating a set of files on the drive. Dividing the total size of the pattern by the time spent for this operation we get the writing speed of the particular drive:
The drives from Maxtor are the best of all at creating files of the Install pattern. The Samsung SP1213C could only show some competition to them and this is no surprise – this drive has been fast and confident in other tests, too. The result of the Samsung SP1203N is interesting as it shows a very high speed, notwithstanding its small cache buffer.
The test devoted to creation of ISO-like files goes next:
The Maxtor team ousted the Samsung SP1213C and occupied all top positions of the table. Otherwise, there are no changes since the first pattern. Interestingly, the Maxtor with a 2MB cache buffer found itself in the “premier leaguer” of hard disk drives: size alone doesn’t matter if you can’t apply it with skill.
On the other hand, it’s also true that skill and size should go hand in hand to work wonders…
The Samsung SP1213C alone is challenging the Maxtor team for top places at creating MP3-like files. The slowest of the Maxtors (6Y120L0/60) has stepped up one step closer to the main group.
The next patterns – Programs and Windows – contain numerous small files.
The Maxtor team loses its ground on small files and there are only two 6Y120P0 models on different platters in the top part of the diagram. The Samsung SP1213C found itself on top at last, while the Hitachi IC32L120AVV207-1 and Samsung1203N were good enough, too.
Now, let’s see the drives doing reading.
Strangely enough, the Hitachi HDS752412VLAT20 wins the Install pattern (you may recall its being obsessed with reading), followed by the Samsung SP1203N. Three drives from Seagate are good in this test: ST3120026A, ST3120022A and ST3120026AS.
The Samsung SP1213C and Maxtor 6Y120P0/80 have got closer to the leader at reading the ISO pattern. WD1200LB and WD1200JD from Western Digital also look good in this test. In this case we see a clear correlation between the areal density and the reading speed. Well, with files that large we are in fact measuring the linear read speed. Only anticipatory reading influences the results.
The Samsung SP1213C was the fastest at reading files of the MP3 pattern. The Seagate ST3120022A took the second place, replacing the ex-leader Hitachi HDS752512VLAT20.
Hard disk drives from Hitachi take the lead at reading small files.
The Samsung SP1203N and Seagate ST3120022A are trying to compete with them in the Programs pattern. In the Windows pattern, the Hitachi team takes the whole podium.
Now, let’s deal with copying within one and the same partition (copy near).
Again, we see the drives from Samsung and Maxtor coming to blows. The Samsung SP1213C was the absolute winner, though, as it is the only drive to overcome the 20MB/s barrier.
Copying of large files suggests that large cache doesn’t necessarily mean success at copying – the Samsung SP1203N (with 2MB of cache memory) competes with more advanced participants. And really, the size of the copied file exceeds that of the buffer here…
The drives from Hitachi are surprisingly poor in this test – probably the firmware doesn’t allow “enlarging” cache lines to the maximum.
The Samsung SP1213C likes copying MP3-like files – it left other drives far behind. The IC35L120AVV207-1 and HDS752512VLAT20 models from Hitachi make amends after the disaster at copying the ISO pattern. The Samsung SP1203N seems like stuck to two drives from Maxtor, 6Y120P0/80 and 6Y120M0. Notwithstanding its small cache, the Samsung never loses to the Maxtors for a few tests already.
The drives from Samsung are good at copying both small and large files: you can see it by the results of the Programs and Windows patterns. Two drives, the SP1213C and the SP1203N, were the fastest in Programs, while the slow 5400rpm SV1203N outperformed the fast Maxtor 6Y120P0 in the Windows pattern! The Hitachi team seems to wake up on small files, showing decent results.
Now we will be copying files into another logical volume.
The Samsung SP1213C is good at copying near or far. Curiously, the Samsung SP1203N and the SV1203N show similar results. It is an indication of good caching algorithms that allow models with a low rotational speed provide good performance.
The leaders remained the same in the ISO pattern, but three drives from Western Digital have got closer to them. Note that the WD1200JB/80* loses considerably to the same model on 60GB platters. It seems like firmware means more here than the difference in the areal density.
The Hitachi IC35L120AVV207-1 speeds up on small files, while other drives remain on their respective places.
The results of copying small files to another partition practically repeat the results of copying within one partition. The drives from Hitachi are good here again, surpassing the competitors. The Maxtor drives are poor at handling small files.
The results for the NTFS file system don’t differ greatly from what we saw in FAT32. So there’s no need for a close examination of the results, just a few remarks: firstly, the speeds of the drives are slightly higher in FAT32 than in NTFS. Secondly, the Maxtor drives are somewhat better in NTFS than in FAT32 and seldom lose to the Hitachi drives in tests with small files.
It is in very rare cases that a comparison of so many hard disk drives can reveal an undisputed leader. There’s always a test where the seemingly best drive shows poor results. So let’s try to determine winners in every particular test we used today.
The drives from Hitachi were good in the Intel IOMeter tests, especially the IC35L120AVV207-1 model. The WD1200BB and WD1200JD from Western Digital also handle server workloads well.
The Maxtor 6Y120P0 drives on any platters and the 6Y120M0 were the best in WinBench 99. The Samsung SP1213C showed the fastest linear read speed.
In our favorite test, FileCopy Test, the Samsung team was preferable with their flagship, the SATA-interfaced SP1213C drive being the best of them.
The SV1203N, the only 5400rpm drive in this review, made a good impression, too. It may suit to people who care about money and/or silence.
The drives from Maxtor easily handle large files, which makes them a good choice for capturing video and such applications. The drives from Hitachi, in our opinion, suit best for standard Windows applications.