Seagate Momentus XT: Hybrid Drive in Action

When they launch new drive series, it is especially interesting to find out what innovations they made to justify the appearance of the entire new series. With Seagate Momentus XT we knew right from the start: this drive combines magnetic platters and flash memory in one package. How cool is that? Find out from our new article.

by Aleksey Meyev
08/17/2010 | 12:34 PM

Mixing together two technologies which are developing in parallel may often produce the compromise that users are looking for. When it comes to hard disk drives, we can recall Western Digital’s Raptor as a product that combined a spindle speed of 10,000 RPM typical of server HDDs and a SATA interface available on ordinary desktop PCs. The resulting super-fast desktop HDD enjoyed high demand.

 

Today, we can also see two technologies developing in parallel: traditional HDDs offer huge storage capacities at low prices whereas solid state drives deliver high performance but have a high cost of storage per gigabyte. Could there be a compromise? The simplest way is to take one HDD and one SDD and use them together, each for storing certain types of data. But what do you do if your notebook has only one disk bay or if you just don’t know beforehand what data will require quick access? The solution is offered by Seagate in the way of the hybrid Momentus XT drive. We will compare it with ordinary HDDs to see what benefits it brings us.

Testing Participants

Seagate Momentus XT: ST35005620AS, 500 GB

 

The new series is no different from ordinary 2.5-inch drives externally. The continuity is also emphasized by the use of the Momentus brand. The XT suffix must have been borrowed from the desktop Barracuda XT series. The HDD part of this hybrid device is interesting in itself, even though not exceptional. It is a 2.5-inch HDD with one or two 250GB platters rotating at 7200 RPM. It has 32 megabytes of cache memory (the previous Momentus 7200.4 series used to have only 16 GB). We’ve got the highest-capacity (500 GB) model for our tests.

The SSD part of the Momentus XT consists of 4 gigabytes of SLC flash memory. You can see the chip in the bottom right of the photo: it is a Micron 0FB12 MW167.

What can this flash memory do? You don’t have to read Seagate’s presentations to understand that if you know just some basic facts. What is the main advantage of flash memory over HDDs? Yes, it has a lower read access time. It’s hard to improve this parameter with HDDs because it is determined by such slow (in comparison with the speed of an electronic chip) operations as moving the read/write head to the necessary track and rotating the platter, but it is often the case that the user accesses the same and rather small amount of data rather than all data stored on the HDD. So, if such data is cached in flash memory, it can be retrieved very quickly during the next access, without waiting those 12-15 milliseconds for the head to meet the necessary sector.

What else can flash memory do in an HDD? Theoretically, it can be used to cache write requests but 32 megabytes of traditional cache installed in the Momentus XT are going to do that just fine. This cache memory doesn’t cost much and write caching doesn’t require too much memory, anyway. The flash part of the HDD can also store some auxiliary information which is being regularly updated, but that’s a trifle. The biggest effect is going to be produced by caching read requests. But as noted above, it will only work when the user is accessing the same piece of data over and over again. Many benchmarks may not even notice it. Well, that only makes our test session the more interesting.

As for opponents, we took two 500GB/7200RPM drives from our latest comparative review of 2.5-inch HDDs: a Seagate Momentus 7200.4 and a Hitachi Travelstar 7K500.

And there are also a couple of HDDs we have not tested before.

Toshiba MKxx56GSY: MK5056GSY, 500 GB

 

This is a representative of the fast HDD series from Toshiba. It has two 250GB platters, a spindle speed of 7200 RPM, and 16 megabytes of cache memory. That’s everything it may need to be competitive in our tests. By the way, Toshiba has announced a new 61 series of HDDs based on 320GB platters but we have not got them yet. Hopefully, they will have arrived to us by this fall.

Seagate Constellation ES: ST3500514NS, 500 GB

 

An ordinary 3.5-inch drive with a spindle speed of 7200 RPM is going to serve as another reference point in our comparison. This is a 500GB model from Seagate’s new Constellation ES series. Its 1-terabyte cousin did very well in our recent HDD roundup, restoring Seagate’s reputation somewhat. Perhaps we are not quite correct in taking a server series drive, but such HDDs are popular among end-users. So, why shouldn’t we take it if it is among the fastest?

The following table shows the firmware versions of the HDDs:

You should keep it in mind that the same models of HDDs may perform differently with other firmware.

Testbed and Methods

The following testing utilities were used:

Testbed configuration:

The HDDs were tested with the generic OS drivers and formatted in NTFS (wherever formatting was required) as one partition with the default cluster size. 64-gigabyte NTFS partitions with the default cluster size were created for FC-Test. The HDDs were connected to a mainboard port and worked with enabled AHCI.

Performance in IOMark

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

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

The 2.5-inch drives are all close to each other in this test since they have the same recording density and the same spindle rotation speed. The differences that you can see are only because one HDD is luckier than another in terms of the head/surface pair at the beginning and end of the partition. The 3.5-inch HDD looks much better than its 2.5-inch cousins. Having larger platters rotating at the same speed, it can read more sectors in a given period of time and thus deliver higher speed. The recording density being roughly the same, the 3.5-inch drive is faster by a third.

You should keep it in mind that the Momentus XT is going to behave just like an ordinary HDD in most tests. For its flash memory to become useful, the same data should be read again, but synthetic benchmarks do not do that.

Now what about reading from the cache buffer and writing into it?

Hitachi Travelstar 5K500.B, 500 GB

Seagate Momentus 7200.4, 500 GB

Seagate Momentus XT, 500 GB

Toshiba MKxx56GSY, 500 GB

Seagate Constellation ES, 500 GB

Seagate’s HDDs have progressed in terms of working with the cache buffer. We have been criticizing them for their low speed of working with large data blocks, especially at writing. You can see that clear enough in the diagram of the Momentus 7200.4. 3.5-inch HDDs from Seagate usually behave in the same way, but the Constellation ES is faster whereas the Momentus XT is even trying to catch up with its opponents. As for the opponents, the Hitachi 7K500 is excellent in this test while the Toshiba is just good.

Working with the cache buffer may seem to be the first use of the integrated flash memory. Well, this is not so. The performance of an HDD is limited by its firmware and external interface here. Each HDD delivers its maximum available speed in this test, the drives without flash memory just processing small amounts of data stored in their ordinary cache memory.

The peak speeds do not matter much in this test although this time around the highest speeds are indeed shown by the overall leader Hitachi 7K500.

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 results can be viewed in tables by clocking the links below. We will be discussing graphs and diagrams.

It is only at sequential reading that the 2.5-inch drives differ: the Momentus XT is ahead of its opponents on small data blocks. The 3.5-inch drive from Seagate is unrivalled, though. The Toshiba is somewhat disappointing as it reaches its top speed on data blocks of 32 KB only.

The Momentus XT is better than its predecessor at sequential writing: the Momentus 7200.4 is somewhat better than the others on small data blocks but falls behind them on large data blocks.

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.

The Momentus XT is somewhat worse than its opponents in terms of read response time. The HDDs are all close to each other in terms of write response time.

Take note how much better the 3.5-inch HDD is in comparison with its 2.5-inch counterparts. Yes, it has to move its heads by a larger angle, but its actuator is more powerful. Moreover, its server orientation allows it to move its heads as fast as possible, disregarding the noise factor.

Random Read & Write Patterns

Now we will see how the performance of the drives in random read and write modes depends on the size of the requested 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 basing on data-transfer rate in megabytes per second.

We don’t have any unexpected results at random reading. It is good that the Momentus XT doesn’t produce the odd-looking hump which you can see in its predecessor’s graph. The difference in performance is small while predictable results are far more important.

The two 2.5-inch drives from Seagate behave completely differently at random writing. The Momentus 7200.4 isn’t good at writing small data blocks as if it has too few cache lines to store a lot of small write requests. The Momentus XT is free from that downside. Moreover, it performs faster than its opponents on some data blocks, so we can suppose that its flash memory takes part in caching write requests. We can’t be sure about that after only one test, though.

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 IOMeter: Database pattern.

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

 

There is nothing unexpected at the minimum request queue depth. The three modern 2.5-inch drives are close to each other. The Constellation ES is ahead due to its lower response time and more effective deferred writing whereas the 7200.4 doesn’t do much deferred writing and has low results as the consequence.

The graphs become more complicated when the queue grows longer. The Constellation ES is still unrivalled at pure reading and writing but is not that good under mixed loads. The 7200.4 is very good at reading but poor at writing. The new XT delivers stable performance, being always ahead of the Toshiba as well as the Hitachi. The latter is poor at mixed loads, too.

At the maximum queue depth the Momentus XT takes second place, behind the unrivalled Constellation ES.

The next diagrams show the performance of each HDD at five different queue depths.

The Constellation ES behaves exactly like the 1-terabyte model from the same series that we tested before. It has aggressive read reordering algorithms and effective deferred writing. But it doesn’t (or cannot) do both at the same time. The Momentus XT and the Toshiba look more balanced. They have all the optimizations, too, but apply them in such a way as to deliver high speed (especially for their not-very-fast read/write heads) at any loads. Well, it is good that the firmware flaws we could see with the Momentus 7200.4 (it has too weak deferred writing) seem to have been left in the past.

Web-Server, File-Server Patterns

The drives are tested under loads typical of servers. The names of the patterns are self-explanatory. The results are presented as performance ratings which are calculated as the average speed of the drive at every load.

The read-only load doesn’t reveal anything new. The Constellation ES is ahead, and the 7200.4 is second due to its somewhat higher performance at short queue depths.

The addition of write requests changes the picture somewhat. The Constellation ES seems to be asleep until a request queue depth of 32, but then wakes up and accelerates quickly, winning top place in the performance ratings. The Momentus XT is better than the other 2.5-inch drives at any queue depths whereas its predecessor 7200.4 is only good at short ones and then falls behind the Hitachi and Toshiba.

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. You can also click the following links for the full results:

Well, our new test method with the read zones being far apart seems to be difficult for HDDs. The second read thread changes the standings greatly: the Constellation ES loses its leading position while the 7200.4 falls far behind the rest of the HDDs. The third thread slows the Hitachi down, the Momentus XT taking the lead.

Multithreaded writing is not as difficult as multithreaded reading, yet some HDDs do not do well here. The Toshiba and the Momentus 7200.4 lose half their speed at two data threads, for example. The Hitachi slows down when writing three threads. The HDDs from Seagate’s two new series win this test.

Performance in FC-Test

For this test two 64GB partitions are created on the drive and formatted in NTFS. 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.

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 obtained in the Install, ISO and Programs file-sets.

Each drive shows its peculiar character when creating files. The new drives from Seagate, especially the Constellation ES, prefer to write large files whereas the Hitachi and the Toshiba are faster with small files. Take note that the Hitachi is even ahead of the 3.5-inch HDD, despite the latter’s higher sequential speeds, in the Windows and Programs patterns that contain small files. This is the effect of firmware optimizations.

It’s somewhat different at reading: the HDDs are closer to each other, although we can see that the Constellation ES is very good (and the 7200.4 is inexplicably poor) with large files whereas the Momentus XT is better than the others at reading small files.

When copying files, the Constellation ES is ahead with large files while the Momentus XT, with small ones.

Performance in PCMark Vantage

Compared with the previous versions, the Vantage version of PCMark is more up-to-date and advanced in its selection of subtests as well as Windows Vista orientation. Each subtest runs 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.

Well, this is the test that reminds us of the special feature of the Momentus XT. It is up to three times as fast as its opponents here. Why? Because, as we wrote in the description of this test, “each subtest runs ten times and the results of the ten runs are averaged”. This HDD just put most of the test data into its flash memory during the first run and produced them from there during the next runs. Let’s check out what we had during each run of PCMark Vantage:


Click to enlarge

As you can see, the results improve considerably between the first and second run of the benchmark, the overall score growing almost twofold. Then, they grow somewhat more during the third and fourth runs and then stabilize. The Media Player subtest is the only one in which the Momentus XT doesn’t improve its performance because the performance is already higher than the speed of reading from the platters. That is, this subtest is processed in the HDD’s cache memory. Photo Gallery and Movie Maker are two more subtests in which the HDD’s performance doesn’t improve much. That’s because most of the load in these subtests consists of write rather than read requests.

By the way, this raises the question whether these results should be considered correct or not? On one hand, we stick to our methodology, but on the other hand, you don’t often read the same data over and over again on your computer. Thus, it is the results of the first run of the benchmark that are closer to real-life applications. And these results suggest that the Momentus XT is as good as the Hitachi 7K500 but not exceptional.

In fact, everything will depend on your usage scenarios. If you come home each day and launch the same programs, why shouldn’t the hard disk put them into its cache?

By the way, how does the Momentus XT react to system reboots? Does it clean up its cache? Yes, flash memory is nonvolatile, but we should check this out. So, we repeated the ten runs of PCMark Vantage, but rebooted the system before each run.


Click to enlarge

So, data remains in the flash memory when we turn the computer off but you can see that the results have become lower. Why?

Well, each reboot means reading OS data. For the hard disk, there is no difference what data are read from it and it puts every data into its flash memory. 4 gigabytes of flash memory is not so much for today’s applications and OSes and new data replace old ones in that cache.

It is easy to check this out. Let’s run the benchmark ten times more, rebooting the system and inserting a full cycle of reading in our FC-Test before each run.


Click to enlarge

The results have become lower because a large part of the 4GB flash memory is taken by the data we read during FC-Test. So, if a user starts his computer up each evening and then, besides loading the same set of applications, performs such activities as listening to music, watching movies or using the hard disk actively in some other way, the situation is going to be the same as in this test, the HDD’s flash memory having to load new data all the time.

You can also note that the Media Player results are now much lower. It looks like our experiments affect not only the flash memory caching but also other caching mechanisms.

Finally, we do the same test but use FC-Test to read the Programs pattern and to write the MP3 pattern.


Click to enlarge

There are but minor changes compared to the previous test variant. Writing seems to be done without using the flash memory.

Now that we’ve found out the peculiarities of the Momentus XT, we can proceed with our traditional tests.

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 integrated defragmenter of Windows 7 and marks the time of the beginning and end of the defragmentation process. For more information about this test, you can refer to this article.

The more we test HDDs with this application, the more unpredictable the results get. Although we can note some general trends, it is hard to predict the winner in a particular test session. Here, the Hitachi 7K500 is the winner, followed by the Momentus XT and Constellation ES which spend over a minute more.

Performance in WinRAR

Now we are going to show you one more interesting test in which we use WinRAR version 3.91 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.

The Momentus XT performs this task faster than its opponents. This HDD seems to have found a way to make use of its flash memory.

The Momentus XT cannot repeat the same trick when unpacking the archive, so the top places are taken by the Hitachi and the Constellation ES.

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 Constellation ES is not included here as it consumes electric current from two lines, making the comparison unfair. The Momentus XT is right in between the two HDDs we already tested before whereas the Toshiba requires a lot of power, over 1 ampere. Roughly speaking, it will not be powered up even by two USB ports!

The Momentus XT shows itself very economical in idle mode, requiring no more than half a watt. If you read our latest 2.5-inch HDD report, you already know that the HDD just turns off its motor if not accessed for a certain period of time. Considering that it has 4 gigabytes of data in flash memory, this is a very reasonable power-saving solution.

Take note that the 3.5-inch HDD is much more voracious than its smaller counterparts. It needs about five times as much power when idle. And where does this power go? Yes, it is dissipated as heat. This difference may be crucial in some situations.

At random-address operations the 2.5-inch drives are very close to each other both at reading and writing. The difference between the models is a mere 10%. Compared to them, the 3.5-inch Constellation ES is a glutton as it eats three times more power!

When processing sequential requests, the HDDs of different form-factors get somewhat closer to each other, yet the 3.5-inch one still consumes more than two times as much power as its opponents. The Hitachi 7K500 wins this test but by a very small margin.

Conclusion

Representing a new generation of hybrid storage devices, the Seagate Momentus XT is quite an interesting product. Seagate has got rid of older problems that used to pester its previous series and now its 2.5-inch drive is indeed competitive to rather strong opponents. The Hitachi Travelstar 7K500 and Toshiba MKxx56GSY both showed high performance, so we cannot name an overall winner among them. You should base your choice on your usage scenarios.

The Seagate Momentus XT is a special case since it is not only a fast 2.5-inch HDD but also a hybrid device with 4 gigabyte of integrated flash memory. We guess this new generation of hybrid HDDs is better than their predecessors (we had a chance to check them out some time ago) because the effect of flash memory can be easily seen. But will this effect be called for? That’s a difficult question to answer. Modern OSes have advanced caching mechanisms while system memory doesn’t cost too much money. It is sad that the hybrid drive’s resources cannot be fully controlled by the OS. We guess it would be more useful than the total caching of everything as we can see now. Still, we are not inclined to disregard this technology as useless. The 4GB cache may be quite enough to make things go faster for some users, especially for notebook users with one HDD bay who just don’t have any other means to speed their disk subsystem up.

As for the comparison between full-size 3.5-inch and small 2.5-inch HDDs, there have been no fundamental changes. The large HDDs are still faster, both at sequential and random-address operations. Their advantage is not too large, however. So, if the dimensions or heat dissipation matter to you, you can consider 2.5-inch 7200RPM drives as a viable option. No one will dare call them slow.