Terabyte Hard Disk Drives Roundup: Round Three

We are going to talk about the new 1 TB HDD models that appeared in the market over the past year. And since one of these solutions is Western Digital Caviar Green WD10EARS with 4 KB sectors, we will also discuss this type of HDD internal structure, its pros and cons.

by Aleksey Meyev , Nikita Nikolaichev
03/08/2010 | 12:52 PM

We seem to have established a tradition of writing an article about 1-terabyte hard disk drives each ten months or so. In the first roundup, written almost two years ago, we just acquainted you with such products because that was the peak storage capacity available then and only few makers had reached it with three platters. Other manufacturers needed four or even five platters for that. About one year ago, we published another roundup in which most of the HDDs were based on three platters only, and this factor had a positive effect on their sequential speeds. And now we are ready to give you one more snapshot of the 1-terabyte sector of the HDD market.

 

Over the last year 1-terabyte HDDs have almost ceased to be viewed as large because all the makers have introduced HDDs two times that capacity at both 5400 RPM and 7200 RPM. These giants have grown a larger cache which is now as big as 64 megabytes. Some of them have even transitioned to the new SATA 6 Gbps interface even though it does not offer any tangible benefits as yet.

1-terabyte HDDs have also lost their superiority in terms of cost of storage to 1.5-terabyte products but still remain highly popular as representing an optimal combination of price, capacity and performance. What have they acquired over this time? Well, most of them are already based on only two 500GB platters. Balancing a pack of two platters proved to be easier, so 1-terabyte HDDs were actually the first among 7200RPM products to adopt such platters, being a nice addition to the higher-capacity 5400RPM drives. Seagate’s Barracuda 7200.12 was the very first. Hitachi’s Deskstar 7K1000.C and Samsung’s SpinPoint F3 joined in later on. Western Digital seems to be the only company to ignore this combination although we may have just not yet seen such a model (alas, the marking of this company’s products does not tell definitely what differentiates models within the same series; the main part of the product name is often left the same for multiple models that have the same storage capacity at a different number of platters).

The combination of two platters and new electronics should be most beneficial for 5400RPM HDDs which are usually selected for passive data storage or low-noise computers. Such products are going to combine good performance with low power consumption (and, consequently, low heat dissipation).

And finally, the most exciting thing was the release of Western Digital’s WD10EARS drives which have a larger sector size (4096 instead of 512 bytes). We’ll dwell upon this issue later on. Right now, let’s have a look at the products we are about to test.

Testing Participants

Hitachi

Deskstar 7K1000.B: HDT721010SLA360, 1 TB

 

Deskstar 7K1000.C: HDS721010CLA332, 1 TB

 

Hitachi is represented by two products in this review: a relatively old 3-platter model from the 7K1000.B series and a 2-platter disk from the new 7K1000.C series. The photographs make it clear that the tin itself – the metallic casing – has not changed while the electronic parts differ. The 1000.B series used to look rather humble against its opponents with its 16 megabytes of cache memory (not only those opponents, but even its predecessor Hitachi 7K1000 has two times as much cache!), but the new series has a decent 32 megabytes. The new electronics is a promise of higher performance, too.

Samsung

SpinPoint F1: HD103UJ, 1 TB

 

EcoGreen F2: HD103SI, 1 TB

 

SpinPoint F3: HD103SJ, 1 TB

 

There are three models from Samsung in this review because we could not help comparing the newer products with the SpinPoint F1 which once surprised us with its 333GB platters and excellent performance (for its time). The newer products are an EcoGreen F2, which is an update of the power-efficient 5400rpm series, and a SpinPoint F3, a representative of Samsung’s new 7200RPM series. By the way, like Hitachi’s HDDs, Samsung’s new-generation products have acquired new and, hopefully, faster electronics.

Western Digital

Caviar Black: WD1001FALS-00J7B0, 1 TB

Caviar Black: WD1001FALS-00E8B0, 1 TB

 

Caviar Green: WD10EADS-00M2B0, 1 TB

Caviar Green: WD10EADS-00P8B0, 1 TB

 

Caviar Green: WD10EARS-00Y5B1, 1 TB

 

Western Digital is trying to overwhelm us with sheer numbers. This company is constantly releasing new variants of drives, and it is indeed fascinating to watch this evolution. There are as many as five WD-branded HDDs in this review.

The two Caviar Black models, 00J7B0 and 00E8B0, come from the sector of fast 7200RPM HDDs. The former earned our recognition in our previous review of 1TB drives for its superb performance. As for the latter, we had hoped it would prove to have a 2-platter design, but were disappointed. The difference between them is quite noticeable, though, so we will be able to watch the progress of one HDD model from one revision to another.

The other three drives from WD belong to the power-efficient Caviar Green series which has a reduced spindle rotation speed. This series has been a headache for hardware testers since the WDxxEACS models with 16MB cache because it includes too many variations differing in additional markings only whereas the actual difference may vary from negligible to dramatic. The 00M2B0 model is one of the few 5400RPM Caviar Green drives available today in shops.

We were especially curious about the 00P8B0 model. Western Digital adopts an odd policy with respect to its power-efficient products and does not specify the spindle rotation speed for them. Some time ago this provoked a lot of discussions and some people even ventured a supposition that the spindle might vary its speed at work. The tests were absolutely unambiguous, however. The WDxxEACS/WDxxEADS disks had a constant speed of 5400 RPM. But when we ran our tests for the 00P8B0, we found it to have a spindle speed of 5000 RPM. This may have helped ensure the necessary recording density on an “unlucky” head/platter pair or Western Digital just wanted to lower the drive’s power consumption some more, but these are only our suppositions. Anyway, we’ve got this 5000RPM drive (by the way, there are 5000rpm HDDs of other storage capacities, too) and we are very curious as to how it compares with its opponents.

The WD10EARS is a third representative of the power-efficient series. The change of one letter in the model name means a lot: this model has a large cache, being the first power-efficient HDD with 64 megabytes of cache memory. And its platters are divided into sectors 4 KB rather than 512 bytes large. Western Digital calls this Advanced Format and we will discuss it in the next section.

Advanced Format

A little bit of history first. The idea of transitioning to 4KB sectors rose up back in 1998 but it is only in 2007 that the International Disk Drive Equipment and Materials Association released a document that summarized the 7-year work on the Long Data Sector technology (it is sometimes referred to as Long Data Block, too). A variation of this technology is implemented by Western Digital under the name of Advanced Format.

The idea of long sectors is simple: the disk is divided into 4KB rather than 512-byte sectors. Thus, a 4-kilobyte chunk of data will require one sector to be stored instead of eight sectors as in today’s HDDs. The eightfold reduction in the number of sectors per track ensures a significant saving in terms of auxiliary data such as the Sync/DAM and ECC blocks accompanying each sector and in terms of intervals between sectors. The ECC block becomes larger with the larger sector, but the total space occupied on the disk by these auxiliary data is indeed reduced.

What is the point of all this? First, the larger ECC block increases the probability of recovering data read errors which is highly important for today’s HDDs with high recording density (and, consequently, with a very low signal-to-noise ratio). If ECC cannot help recover the data, the disk has to read the sector once more, which requires a full rotation of the platter. Thus, the higher ECC recover rate means that there will be fewer idle rotations and a lower probability of an irrecoverable error. Western Digital claims a 50% improvement in ECC-based data recovery.

Another good thing about the 4KB sectors is that the auxiliary information takes less space and this saved space can be used for user data. From an end-user’s point of view, it means that the developer can produce a larger-capacity HDD, the other technologies being the same. Although this advantage (about 7-11%) is not as high as we see each time when the developers move on to higher-density platters, it is good all the same.

And finally, this technology increases the disk’s linear recording density. It helps increase the amount of data fitting into each track and we can expect an appropriate increase in the drive’s linear speed.

What are the downsides of this technology? It puts a higher load on the drive’s electronics responsible for ECC but that’s not a big problem today. More importantly, there is a lot of software, from BIOS code to applications, that has been developed with 512-byte sectors in mind and it is impossible to switch to the new sector size in one move. Something has been done already. Microsoft’s new operating systems starting from Windows Vista (and including Windows 7) support 4KB hard disks. LDS technology is also supported by many RAID controllers. Mainboards are acquiring this support in their BIOS code, too. However, there is no full compatibility as yet.

It is because of this possible incompatibility that Western Digital chose the following strategy for its new HDDs. Physically, the HDD’s platters are divided into 4KB sectors. However, it pretends that it has 512-byte sectors. Each physical sector contains eight logical sectors. All the required address translations are done inside the HDD, ensuring maximum compatibility but at the expense of performance.

There is one biggest downside to this emulation that made Western Digital develop a special solution. The still popular Windows XP (and all of its predecessors) uses a curious way of formatting the hard disk. When creating a partition, the first 63 sectors (from sector 0 to sector 62) are reserved while the partition itself begins at sector 63. With 4KB physical sectors, this sector 63 proves to be the eighth logical sector in the eighth physical sector. Then, the NTFS file system accesses the hard disk in 4KB data blocks, producing a very bad effect. Due to the 512-byte difference (sector 63), each cluster of the file system resides in two physical 4KB sectors, although a cluster is the same size as a sector. When reading data from the disk, more information than necessary has to be read and the required data has to be extracted from it. When writing a cluster, the disk has to read two physical sectors, change the eight logical sectors corresponding to the cluster’s address, and write them back. Writing a 512-byte sector (system data are changed by means of such requests), the hard disk has to read a physical 4KB sector, change the required logical 512-byte sector, and write the modified physical 4KB sector back.

It resembles RAID5, doesn’t it? :)

Of course, this has a negative effect on the drive’s performance as you will see shortly. We don’t know why Microsoft chose the offset size of 63 sectors but it used to show up with RAID arrays (where disk stripes would be shifted relative to the OS clusters) and SSDs (due to block-based access) and has become a real problem by now. The good news is that Windows Vista and its successors (including Windows 7) are free from this problem, taking a different approach to formatting hard disks.

If you still want to use new disks under old OSes, Western Digital offers two solutions. First, you can set the HDD’s jumper to pins 7 and 8 and the HDD will move its logical structure by one logical sector so that sector 63 from the OS’s point of view is actually sector 64, i.e. the beginning of a physical sector. This shift can be easily implemented on the logical level: each incoming sector address is just augmented by one when translated into the disk. The disk’s capacity does not suffer because the last sector becomes sector 0 due to the cyclic shift.

This solution is recommended to create one partition on the disk. If you want to create multiple partitions, Western Digital recommends another method which is also used to transfer disk images.

You take your “unaligned” disk and run the WD Align tool (downloadable from the Western Digital website). This tool shifts the existing disk partitions to align them. This method is universal but takes more time because data are actually transferred from one sector to another. Empty sections will be transferred quickly but full ones will take some time. You must make sure that there won’t be power outages during the process. Again, the disk capacity does not change after this procedure but we wonder what WD Align would do with an unaligned and absolutely full partition the size of the whole disk?

WD Align won’t allow you to align an ordinary (one that doesn’t support 4KB sectors) or an already aligned disk. This protection can be easily bypassed if you insert the jumper after having aligned the disk. We hope you won’t do that, though. And we also hope you won’t insert the jumper into a disk formatted in Windows Vista 7 because the sectors will be additionally shifted by one 512-byte sector and the previously created partition won’t be accessible. So, don’t do that!

And now, let’s move on to our tests.

Testbed and Methods

The following testing utilities were used:

Testbed configuration:

The HDDs were tested with the generic OS drivers. We formatted them in FAT32 and NTFS as one partition with the default cluster size. For some tests 32GB partitions were created on the drives and formatted in FAT32 and NTFS with the default cluster size, too. The HDDs were connected to a mainboard port and worked with enabled AHCI.

The 4KB sectors of the WD10EARS interact with the test load in a specific way. It does not matter for synthetic benchmarks like IOMark and IOMeter whether this HDD is aligned or not because they work with the HDD without creating partitions on it. It must be noted, however, that random loads in IOMeter can begin with any sector address in 512-byte addressing. Easy to guess, only one out of eight requests will be lucky to coincide with the beginning of a 4KB sector. This is in fact the same as working with an unaligned disk.

The other tests work with file structures on the disk, so you will see two results in them, with the disk aligned and unaligned. We aligned our HDD with the WD Align tool to make sure the second partition in FC-Test was always aligned.

Performance in IOMark

We use our internal IOMark tool for low-level tests. Let’s begin with sequential reading.

Let’s compare the drives according to the speed at the beginning and end of the full-capacity partitions created on them.

You can see that the new 7200RPM drives with 500GB platters are faster than their predecessors. Funnily enough, the new 5400RPM drives have the same linear speed as the old 7200RPM models with 333GB platters. Take note of the 5000RPM drive from Western Digital. Its top speed is somewhat lower than that of its 5400RPM cousin but its bottom speed is much higher. This top to bottom speed ratio differs from the typical 2:1 and is usually observed with HDDs that have reduced-width platters. Is it the result of the developer’s attempt to transition to a higher recording density? Otherwise, it wouldn’t be possible to have 500 gigabytes of storage space on a reduced-width platter. Well, let’s just wait for the results of our response time measurements.

The WD10EARS with its unusual sectors does not deliver any advantages.

Now, what about the cache access speed? We’ve got rather interesting results here.

Hitachi Deskstar 7K1000.B: work with the buffer

Hitachi Deskstar 7K1000.C: work with the buffer

Hitachi’s new Deskstar 1000.C shows us a terrible picture that we already saw with the 7K2000. It has inexplicable slumps at reading and its writing speed goes up but temporarily. The overall performance is very low. This HDD seems to have got the new firmware which cannot draw a nice-looking cache access graph.

Samsung Spinpoint F1: work with the buffer

Samsung EcoGreen F2: work with the buffer

Samsung Spinpoint F3: work with the buffer

Samsung’s HDDs are steadily getting better and better. The F1 series was good. The EcoGreen F2 draws a smooth beginning of the graph but has some problems with writing large data chunks, and the F3 can be regarded as a perfect example.

Western Digital Caviar Green WD10EADS-00M2B0: work with the buffer

Western Digital Caviar Green WD10EADS-00P8B0: work with the buffer

Western Digital Caviar Green WD10EARS-00Y5B1: work with the buffer

Western Digital Caviar Black WD1001FALS-00J7B0: work with the buffer

Western Digital Caviar Black WD1001FALS-00E8B0: work with the buffer

Western Digital tries to use unified firmware for all its products, and its HDDs all show the same behavior. It is hard to see any difference in this test: all of the models deliver an excellent speed of reading and have some fluctuations of speed when writing large data chunks.

The Hitachi 7K1000.B is the first in terms of top speeds. It is closely followed by Samsung’s HDDs.

Performance in Intel IOMeter

Sequential Read & Write Patterns

IOMeter is sending a stream of read and write requests with a request queue depth of 4. The size of the requested data block is changed each minute, so that we could see the dependence of the drive’s sequential read/write speed on the size of the data block. This test is indicative of the maximum speed the drive can achieve.

The numeric data can be viewed in tables. We will be discussing graphs and diagrams.

The Samsung F3 is brilliant at sequential reading and writing, being much better than its predecessor from the F1 series. The Hitachi 7K1000.C, yet another 7200RPM HDD with 500GB platters, competes with the leader when processing large data blocks but fails to do so with small data blocks. Moreover, this Hitachi is actually the slowest among all, including its predecessor and the 5400RPM models, with small data blocks. The Samsung F2 is somewhat slower than the other power-efficient drives whereas the 5400RPM HDDs from Western Digital are very close to each other.

Disk Response Time

For 10 minutes IOMeter is sending a stream of requests to read and write 512-byte data blocks with a request queue of 1. The total of requests processed by each HDD is much larger than its cache, so we get a sustained response time that doesn’t depend on the HDD’s buffer size.

Western Digital’s Caviar Black series drives are still the best when it comes to response time at both reading and writing. The newer model has even improved a little more.

The second team place goes to Hitachi, the older 1000.B being somewhat better at reading and the newer 1000.C much faster at writing.

Samsung’s products are generally not very good in this test, and this time around its 7200RPM HDDs are just a little faster than Western Digital’s 5400RPM ones. The EcoGreen F2 has the highest response time at reading among all these drives. The only good news for Samsung is that the F3 is somewhat better than the F1.

Take note of the P8-indexed WD10EADS. Despite its lower spindle speed (5000 RPM), it is about 0.5 milliseconds better than the M2 model. So, it has reduced-diameter platters indeed.

The WD10EARS has to emulate 512-byte sectors and is a little slower than same-class HDDs at reading, but its writing result is downright poor. It has to read a full physical sector to change one eighth of its data, and this is the outcome of such inefficiency.

Next goes the test of average positioning speed. The drive is being bombarded with read requests like in the response time test, and we calculate the difference between the LBA addresses of the previous and next requests and divide it by the time it took to perform the request. In other words, we have the distance (in gigabytes) the drive’s heads can run through in 1 second.

This test is sensitive to recording density, but Western Digital’s Caviar Black are superior nonetheless. Their swift read/write heads make up for the lower recording density. The Hitachi 1000.C is rather too bad here. It is a shame for a 7200RPM drive with 500GB platters to have last place in the competition. The WD10EADS P8 feels much better. Its smaller-diameter and high-density platters make up for its low spindle rotation speed.

Random Read & Write Patterns

Now we will see the dependence between the drives’ performance in random read and write modes on the size of the data block.

We will discuss the results in two ways. For small-size data chunks we will draw graphs showing the dependence of the amount of operations per second on the data chunk size. For large chunks we will compare performance depending on data-transfer rate in megabytes per second.

An HDD’s performance at random-address operations is proportional to its response time until the requested data block is as large as to make the speed of sequential reading or writing more important. You can see this moment easily: staring from 2MB data blocks, the HDDs with 500GB platters (the Samsung F3 and the Hitachi 1000.C) go ahead and leave WD’s Black Caviar series behind.

Of course, this general rule applies only if there are no special factors like the 4KB sectors of the WD10EARS. Its writing graphs go much lower than the other HDDs’ because, when processing small data blocks (smaller than 4 KB), it has to write back the unrequested part of a 4KB sector (to remind you, the HDD itself does not know if there is any valuable information at this address or not). When it comes to large data blocks, the requested address block coincides with the disk sectors in one out of eight cases only. In the other seven cases, data in two sectors has to be rewritten. It is hard to tell why but this HDD is much slower than the same-class opponents until very large requests when the load becomes sequential.

Database Patterns

In the Database pattern the drive is processing a stream of requests to read and write 8KB random-address data blocks. The ratio of read to write requests is changing from 0% to 100% with a step of 10% throughout the test while the request queue depth varies from 1 to 256.

You can click this link to view the tabled results for the IOMeter: Database pattern.

We will build diagrams for request queue depths of 1, 16 and 256.

Western Digital’s Caviar Black drives leave no chance to their opponents in this test, too. The newer E8 version is even an improvement on the already impressive performance of the older J7. Funnily enough, even the power-efficient products of the WD Caviar Green series are as good as the 7200RPM drives from their opponents at the queue depth of 1 request. Samsung’s F3 is the best drive among those opponents whereas its power-efficient cousin EcoGreen F2 is, on the contrary, the slowest one in this test.

The non-aligned random access has a terrible effect on the performance of the WD10EARS because it has to do more write operations than the other HDDs. Judging by the shape of the graph, this disk cannot make up for this deficiency with its large cache.

When the queue is as long as 16 requests, Western Digital comes out the winner again while the best of its opponents, the Samsung F3 and the Hitachi 1000.C, are only competing with the power-efficient series from WD. The Caviar Black drives are unrivalled, the E8 being better than its predecessor, especially at writing. We should acknowledge the progress WD’s competitors have made: the Hitachi 7K1000.C and the Samsung F3 are faster than their respective predecessors. The Samsung EcoGreen F2 is inferior to the power-efficient drives from the other brands, though.

The WD10EARS still has a very low performance (due to the reasons we’ve explained above) if there are any writes to be processed.

There are no changes in the standings when the queue grows even longer. We can only see the efficiency of NCQ in Western Digital’s drives. They are especially fast at low percentages of writes and, judging by the slump in the right part of the graphs, these HDDs prefer to use their cache for reordering read requests rather than for deferred writing. It is only at pure writing that the cache is optimized for deferred writing with the ensuing increase in its efficiency.

To sum up this part of our tests, we will show you diagrams with each drive’s performance at five different queue depths.

Hitachi’s new-generation HDDs obviously have new firmware algorithms as the overall shape of the graphs suggests. The Hitachi 7K1000.C has become better than its predecessor under mixed loads and has more effective deferred writing. Unfortunately, its NCQ is still far from perfect and its performance scalability at longer queue depths might be better.

The Samsung EcoGreen F2 has inherited its predecessor’s firmware. And it shows a rather passive behavior and does not do well under server loads. The SpinPoint F3 is different. It looks like Samsung finally dared to introduce considerable changes into the firmware algorithms in that HDD. The drive’s NCQ has become somewhat better at short queue depths and its deferred writing has improved, too. On the other hand, the HDD recognizes long queue depths but does not process them confidently: there are occasional slowdowns when the queue gets longer.

It is simple with the Western Digital Caviar Black: the company has long been using the same unified firmware for all its products. And this firmware is currently among the best we have seen in a long, long time. The developer has even managed to make deferred writing even more efficient in the E8 version of the Caviar Black.

The power-efficient HDDs have the same but less aggressive firmware. It is clear that these HDDs have a slower request processor and their read/write heads are not so fast (but quieter).

The WD10EARS shows us NCQ but no deferred writing. It is a shame we cannot load the HDD with aligned requests: the result would be interesting considering the 64 megabytes of cache.

Web-Server, File-Server and Workstation Patterns

The drives are tested under loads typical of servers and workstations.

The names of the patterns are self-explanatory. The Workstation pattern is used with the full capacity of the drive as well as with a 32GB partition. The request queue is limited to 32 requests in the Workstation pattern.

The results are presented as performance ratings. For the File-Server and Web-Server patterns the performance rating is the average speed of the drive under every load. For the Workstation pattern we use the following formula:

Rating (Workstation) = Total I/O (queue=1)/1 + Total I/O (queue=2)/2 + Total I/O (queue=4)/4 + Total I/O (queue=8)/8 + Total I/O (queue=16)/16.

This test depends on the HDD’s response time, so the Caviar Black meet no competition here. The Hitachi are only as fast as the power-efficient models from Western Digital whereas the last three places go to the Samsung drives which don’t like this load.

Take note how good the WD10EARS is when there is no writing to be done. It is just as good as its same-class opponents.

The picture changes as soon as we add write requests into the load. The leading pair of Caviar Black drives is split up by the Samsung F3. Yes, the F3 has indeed become better with the new firmware because its predecessors are at the bottom of the diagram, being only ahead of the WD10EARS that has problems with writing. On the other hand, the Samsung F3 is not so good under low loads, which explains its sixth performance rating. We should also note the excellent results of the power-efficient P8. With its small-diameter platters and highly efficient firmware algorithms, it beats both 7200RPM drives from Hitachi despite the difference in spindle rotation speed.

The load becomes more complex and the Samsung F3 falls behind the two leaders. The rest of the results are overall the same as in the previous test. We can only note the sagging part of the Hitachi 7K1000.C’s graph which is indicative of a firmware flaw. Fortunately, this flaw resulted in but a minor performance reduction.

When the test zone is greatly limited (32 gigabytes is less than 5% of the total storage capacity of a 1-terabyte drive), the result often depends on the drive’s recording density. However, WD’s Caviar Black models remain in the lead thanks to their superb electronics. A drive with denser platters can only be seen in third place. It is the Samsung F3. If you compare the graphs or the ratings of the Hitachi drives, you can see that the new dual-platter 7K1000.C has firmware flaws that make it slower than its 3-platter predecessor.

Multithreaded Read & Write Patterns

The multithreaded tests simulate a situation when there are one to four clients accessing the hard disk at the same time – the clients’ address zones do not overlap. We will discuss diagrams for a request queue of 1 as the most illustrative ones. When the queue is 2 or more requests long, the speed doesn’t depend much on the number of applications. You can also click the following links for the full results:

Multithreaded reading tests often produce unexpected results. Here, it is the Hitachi 7K1000.C that beats the others with two data threads. Its performance is almost the same as when reading one thread only. Finally, we’ve found something good in its firmware! The 7K1000.B is not so fast.

Samsung can also be praised for the work they have done on firmware. The Samsung F3 has been cured from the terrible performance hit we’ve seen with other Samsung drives over the last two years (you can still see it with the EcoGreen F2 series).

Western Digital’s drives are stable and cope with the multithreaded load well enough, yet their performance is not high here.

There are a couple of things we want to note about these results. First, the P8 version WD10EADS has much better results than the other models of the same series although we can’t tell you why it is better. And second, the Samsung F3 slows down at four threads almost to the level of its predecessors. So, while the company is on the right way, there is still a lot of work to be done with the firmware.

None of the HDDs seems to have any problems with multithreaded writing but we must name the winner. It is the Samsung F3 that copes best with writing multiple data threads.

Performance in FC-Test

For this test two 32GB partitions are created on the drive and formatted in NTFS and then in FAT32. A file-set is then created, read from the drive, copied within the same partition and copied into another partition. The time taken to perform these operations is measured and the speed of the drive is calculated. The Windows and Programs file-sets consist of a large number of small files whereas the other three patterns (ISO, MP3, and Install) include a few large files each.

Like in the previous tests, the WD10EARS is tested twice: with its partitions aligned and not aligned.

You should be aware that the copying test not only indicates the speed of copying within the same HDD but is also indicative of the latter’s behavior under complex load. In fact, the HDD is processing two data threads then, one for reading and another for writing.

This test produces too much data, so we will only discuss the results achieved in NTFS in the Install, ISO and Programs file-sets. You can use the link below to view full results:

Well, the Samsung F3 is ahead at writing irrespective of the file-set. It is followed by Western Digital’s Caviar Black drives that only give way to the Hitachi 7K1000.B in the ISO pattern. Interestingly, the newer and higher-density Hitachi is slower with each file-set, especially with small files. So, the firmware of the 7K1000.C is not really good after all.

As for WD’s power-efficient drives, the M2 version WD10EADS is ahead of the P8 whereas the aligned WD10EARS is good at processing large files, worse in the Install pattern and downright slow with the small files of the Programs pattern. But if you don’t align it, it will be very poor at writing whatever files you have to offer.

There is nothing interesting at reading. The standings are exactly as we might expect: the Samsung F3 and the Hitachi 7K1000.C are in the lead thanks to their high-density platters and 7200RPM spindle speed. They are followed by the Caviar Black series which are slightly faster than the other 3-platter products. And the power-efficient models go next. Take note that the WD10EARS is again no different from same-class drives at any loads when it comes to reading.

The Samsung F3 is unrivalled at copying files. The two Caviar Black drives take the other places on the podium, being a little faster than their pursuers. The Hitachi 7K1000.C shows an interesting behavior. It is good at copying large files but slow with small files. The WD10EARS has a hard time due to its emulation of 512 sectors. The aligned disk is as fast as its opponents with large files but slows down with small ones. The same goes for the unaligned disk, but it is always slower.

Performance in PCMark 2004/2005

PCMark 2005 has the same tests as the 2004 version (not only in names, but also in results as we have seen a lot of times), so we only discuss one test from PCMark 2004 which is not available in the 2005 version. It is called File Copying and measures the speed of copying some set of files. The other results can be learned from the table. The PCMark 2005 tests are:

The final result is the average of ten runs of each test. See the detailed table here.

PCMark04 agrees with FC-Test that the Samsung F3 and the two Caviar Black drives from WD are the fastest of all. The other HDDs have similar results except for the P8 version Caviar Green and the WD10EARS which performs badly whether aligned or not.

This group of tests produce similar results. And they are explicit enough: the newcomers cannot throw the WD Caviar Black off the top. But we can note that Samsung’s HDDs have progressed: the F3 is more appealing than its predecessor. Unfortunately, we cannot say the same about Hitachi’s products: the newer Hitachi is no better than the older one.

The WD10EARS is just as fast as its same-class opponents when aligned and very slow when unaligned.

Interestingly, the WD10EARS performs somewhat worse when unaligned even in this test which is highly sensitive to caching. It’s hard to tell anything about the leaders because they deliver similar results, yet the WD Caviar Black and Hitachi 7K1000.C are just a little faster than the others.

The Samsung F3 seems to be fond of writing files. Its performance is very good even for an HDD with 500GB platters. Otherwise, there are no surprises. The standings are exactly what we might expect after the previous tests.

Judging by the overall scores, the WD Caviar Black series remains in the lead although the Samsung F3 is a strong opponent to them when it comes to home applications. The Hitachi 1000.C is far from brilliant. It has not failed in this group of tests, but we’d like to see some progress from it.

The WD10EARS needs aligning. Otherwise, you’ll have a cripple barely coping with its job instead of a normal hard disk.

Performance in PCMark Vantage

To make this part of our test session complete, we are going to run the latest version of PCMark called Vantage. Compared with the previous versions, the benchmark has become more up-to-date and advanced in its selection of subtests as well as Windows Vista orientation. Each subtest is run ten times and the results of the ten runs are averaged.

Here is a brief description of each subtest:

Basing on these subtests, the drive’s overall performance rating is calculated.

We won’t analyze the results because they are generally the same as in the previous version of PCMark. The newer version is more favorable to the Hitachi 7K1000.C, yet this HDD is anyway inferior to the leading pair of WD’s Caviar Black drives as well as to both 7200RPM drives from Samsung, including the old F1.

The EcoGreen F2 is surprisingly good. It is somewhat better than WD’s power-efficient models under Windows Vista. Take note that the P8 version is the best power-efficient drive from WD here. Its small and high-density platters are very good even at a reduced rotation speed. We can’t find any fault with the WD10EARS when it is aligned.

Defragmentation

Next goes our homemade test of defragmentation speed. We created a very defragmented file system on a 32GB partition of a disk by loading it with music, video, games and applications. Then we saved a per-sector copy of the disk and now copy it to the disk we want to test. Next we run a script that evokes the console version of the Perfect Disk 8.0 defragmenter and marks the time of the beginning and end of the defragmentation process. We run this test with AHCI enabled. For more information about it, you can refer to this article.

After copying the partition image to the WD10EARS, we aligned it with the WD Align tool.

Well, we’ve got some interesting results here. The Samsung F3 is first. Samsung has something to be proud about until Western Digital transitions its 1-terabyte Caviar Black drives to 500GB platters. The Hitachi 7K1000.C has dense platters, too, but cannot overtake WD’s Caviar Black series.

The WD10EARS is not competitive even when aligned. The sector simulation overhead is too high at both reading and, especially, writing. The unaligned disk is just awful as it took over an hour to perform the task. We’ve only seen such poor performance with server-oriented SAS drives.

Performance in WinRAR

Now we are going to show you one more interesting test in which we use WinRAR version 3.8 to compress and then uncompress a 1.13GB folder with 8118 files in 671 subfolders. The files are documents and images in various formats. These operations are done on the tested drive. This test depends heavily on CPU performance, but the storage device affects its speed, too.

It’s the first time we test 1-terabyte HDDs with WinRAR. Surprisingly enough, the Hitachi 1000.B proves to be the best when creating the archive. Next goes the updated Caviar Black, Samsung F1 and, amazingly, the M2 version WD10EADS. The 7200RPM newcomers with high-density platters are no better than the rest of the HDDs. The unaligned WD10EARS does not fall too far behind because this task is mostly about reading.

The unaligned disk needs over eight minutes while the others unpack the archive in only one. Well, the WD10EARS is not really good even when aligned: it needs three times as much time as the others because the files are small. The rest of the HDDs split up into two groups: the leading group includes the Caviar Black drives, the two new models from Samsung (the F3 and the F2) and the old Hitachi 7K1000.B.

Power Consumption

You can refer to our article called Hard Disk Drive Power Consumption Measurements: X-bit’s Methodology in Depth for details on this test. We will just list the specific modes we measure the power consumption in:

Let’s check out each mode one by one.

The 7200RPM drives are all similar in terms of startup current except that the 7K1000.B needs somewhat less than the others on the 5V line. It’s funny with the power-efficient HDDs: the Samsung F2 needs as much power on the 5V line as the “fast” HDDs whereas the others want more! But all of them require less power from the 12V line. Oddly enough, the WD10EADS version P8 has the highest startup current although it has to spin its platters up to 5000 rather than 5400 RPM.

It is good that the HDDs require less and less power in idle mode. WD’s Caviar Black are the only drives to be always ready to work. The others switch into sleep mode. Hitachi’s HDDs have made some great progress. They have always been good in terms of electronics but now their mechanics has become economical as well. As a result, the Hitachi 7K1000.C is the most economical among all the 7200RPM drives. The difference of 2 watts cannot be only due to the loss of one platter. The actuator must have been modernized, too. Samsung’s HDDs show a similar behavior, the F3 being much more economical than its predecessor.

As for the power-efficient products, the Samsung F2 is slower than Western Digital’s HDDs among which the P8 version has the highest consumption despite its 5000 RPM.

When the HDDs have to move their heads about, most of them consume the same amount of power. The Samsung F3 is better than its opponents, which is a very good result considering that its access time is quite competitive among the 7200RPM models. The dual-platter Hitachi 7K1000.C needs as much power as the 3-platter WD Caviar Black, but cannot deliver as low a response time as them.

The power-efficient HDDs take top places. The P8 version WD10EADS consumes somewhat more than the others, again.

The drive’s electronics has to do some work at random writing. As a result, Samsung’s HDDs take top places, being very economical.

The electronics contributes even more to the overall consumption of an HDD at sequential operations: it consumes as much power as the mechanics in the power-efficient models. Take note of the excellent results of the Hitachi 7K1000.C which is as good as the power-efficient HDDs. It seems to have a very effective drive of the platters and also a rather voracious heads actuator. But the heads do not move and consume much at sequential operations.

Conclusion

Here is the summary of our third test session with 1-terabyte hard disk drives.

Western Digital’s Caviar Black are still the best if you are looking for the fastest HDD. These 1TB drives have not yet transitioned to 500GB platters and are slightly inferior to their opponents in terms of top speeds under sequential loads, but they are really very, very good at everything else. The most enjoyable fact is that new revisions of Caviar Black drives prove to be a little bit faster than their predecessors.

Samsung should also get some of our praise. The SpinPoint F3 does not have the advantage of ultra-dense platters like the SpinPoint F1 had in its time, but looks good against strong competition. Although it does not feel confident under server loads, it does offer a perfect balance of performance, high speed of processing files and low power consumption which is so necessary for home applications.

Hitachi has tried to revise its firmware but there is still a lot of work to do. The results of the Hitachi 7K1000.C were often poor due to firmware flaws and this perspective product is only as fast as its predecessor while having higher potential.

Competition is tough in the sector of power-efficient HDDs. The WD10EADS are not unrivalled anymore. Their opponents have come up very close. The Samsung EcoGreen F2 is no worse when it comes to processing files or as a quiet and cold system disk. WD’s products should be given credit for being still beyond competition under server loads: they can even challenge 7200RPM models from other brands there! Even the P8 with its lower spindle speed is just as good as the others and occasionally faster due to high track density. The different versions of WD’s power-efficient drives do not really differ much.

And finally, we want to say a few words about the WD10EARS, the first widely available HDD with 4KB sectors. First of all, this HDD vitally needs that the file system partitions be aligned with the sectors on the platter. Otherwise, it suffers a catastrophic performance hit, any write operation being a huge problem. The aligned disk differs from same-class products in one thing only: it has a small performance hit when processing small data blocks (or small files) because the emulation of 512-byte sectors makes an additional read of data necessary. This emulation ensures excellent compatibility with all existing software, though.

Hopefully, such HDDs will be cheaper than ordinary ones. Otherwise, we cannot recommend them as a replacement for the WD10EADS. On the other hand, the WD10EARS is going to be a good choice as a passive data storage device. The following table shows what alignment is needed for it in different situations.