by Aleksey Meyev , Nikita Nikolaichev
03/16/2009 | 03:01 PM
Recently we have taken a look at our HDD section and found that our previous roundup of 1-terabyte models is almost 1 year old. It is an eternity by the standards of the IT industry. Of course, we have been busy writing reviews of 750GB drives and 640GB drives but the pause is rather too long anyway.
In fact, we could have written a review (or even several reviews) of 1-terabyte drives a long time ago. But being perfectionists, we wanted to accumulate a full collection of drives with 333GB platters. It is Hitachi’s drives that made us wait the most. Announced in July 2008, the Deskstar 7K1000.B series reached us in January 2009 only.
This wait is over now and we can exhaustively cover the topic of hard disk drives with 333GB platters. It’s high time we did this since 500GB-platter models are already coming up (e.g. Seagate ST31000528AS).
Before we describe the drives we are going to test, we’ll remind you of the events that have happened in the 1-terabyte sector recently.
First of all, the increased activity of Western Digital must be noted. This company perplexed us in our earlier test session with its HDD that claimed to have a variable spindle rotation speed. When they introduced Caviar Blue and Caviar Black, the brand’s product line-up seemed to get logical and complete. But then Western Digital began to improve its products… First, the Green Power series transitioned to 333GB platters (to get cheaper to make), the older and newer drives getting mixed in storehouses and shops as the consequence. Fortunately, there was a way to tell them apart: the newer drives had two magical letters in their name (WD10EACS-00D6B0) that guaranteed the 3-platter design.
Then, the Green Power series was modernized once again (this may remind you of CPUs with their core steppings). Their cache was increased in double, obviously to improve performance, and the marking was changed into WD10EADS.
Thus, there are three HDD models in the Green Power series only. But Western Digital also released enterprise HDDs based on the GP series products…
Realizing the importance of 3-platter design, Seagate too made the life more complex for the user by adding a new drive into the aging Barracuda 7200.11 series. Now it includes two models with a capacity of 1 terabyte.
There is also the SV35.3 series which is a modification of the Barracuda 7200.11 for video-oriented systems (digital recorders, video surveillance systems, etc). Interestingly, SV35.3 drives came out at the same prices as the Barracuda 7200.11 drives. Specialized HDDs are usually more expensive than their desktop counterparts (you have to pay more for the exclusiveness of the product) or cheaper (if an OEM throws a batch of such drives to the secondary market). But here, the prices were incomprehensibly identical.
Getting hold of the much desired 333GB platters, Hitachi thought that releasing only one 1-terabyte drive with such platters would be too trivial. And they introduced two drives at once: Deskstar 7K1000.B (an allusion to the 7K1000 model) and Deskstar E7K1000.
Hitachi was not at all ashamed that the only difference between those models was the amount of cache memory. Their targeting differed, too. The 7K1000.B was meant for desktop PCs while the E7K1000 was to work in data storage systems (not in corporate data storage systems because the word corporate is so far reserved for 1.5 and 2GB models).
Yet another piece of news from the last year is the total lack of any news from Samsung, the main sensation-maker of one year ago. Samsung shocked everyone then, being the first to show hard disk drives with a recording density of 333 gigabytes per platter. Its opponents have just been catching up with it over this year.
However, it is now clear that Samsung has not been active enough. The company has not prepared for a new breakthrough. The announcement of HDDs with a recording density of 500GB per platter is no breakthrough because Samsung has to reduce the spindle rotation speed to achieve this density (go Western Digital’s way, so to say). Moreover, Western Digital has already started to ship its 2-terabyte drive based on 500GB platters. Hopefully, Samsung has just been preparing some surprise for us. J
So, there are three interesting questions this review may answer.
First, we can find out what HDD is the fastest today. Second, we are interested to see the influence of higher recording density and larger cache on the performance of hard disk drives because the complicated line-up of Western Digital’s drives and a couple of Hitachi’s models allows us doing that. And third, we will solve the mystery of the SV35.3 drive and see its difference from the Barracuda 7200.11.
We will be introducing the participating devices in alphabetic order. Coincidentally, first goes the Hitachi Deskstar 7K1000, the first 1-terabyte drive to hit the market. This company had used a five-platter HDD design before and it is thanks to this design solution that Hitachi managed to be the first developer to create a hard disk drive with such a pretty-looking, round value of capacity (while the other developers were waiting for denser platters). We took a note of this HDD earlier, having compared it against its predecessors.
Hitachi offers an enterprise version called Ultrastar A7K1000 but it proved to be so similar to the 7K1000 in our previous test session that we have decided not to include it into this review.
But when the competing brands quickly introduced 4- and even 3-platter drives of 1-terabyte capacity, the five-platter 7K1000 was far from looking good in comparison. It just had lower recording density and worse power consumption parameters. So, Hitachi needed to find higher-density platters and replace the electronics. The company did the job and introduced the Deskstar 7K1000.B series which had a recording density of 333 gigabytes per platter. Unfortunately, that series took quite a long time to reach shop shelves. Interestingly, Hitachi’s HDDs lost some buffer memory on the transition to the denser platters: the 7K1000 had 32 megabytes while the senior models of the 7K1000.B series are equipped with 16 megabytes and junior models, with 8 megabytes only. As for the frequency of unrecoverable read errors, it is 1 to 1014 operations again. Hitachi seems to have returned from enterprise to desktop HDD design in this series.
Still, the company lacked a flagship model. Therefore they introduced the E7K1000 series. Equipped with the same platters and, as the photos show clearly enough, with the same electronics as the drives of the 7K1000.B series, the E7K1000 series models are endowed with 32 megabytes of cache. Frankly speaking, this HDD looks more like an enterprise model and the word Deskstar in its name looks like a typo (Hitachi’s enterprise models are referred to as Ultrastar). Anyway, we can benchmark it whatever it is called.
By the way, chips are always in open view in Hitachi drives, and we can see a new chip and new memory in both new series (7K1000.B and E7K1000).
The 7K1000 had a different selection of chips:
Well, there is nothing extraordinary, actually. The change of logo on the chip is due to the fact that Infineon sold its HDD chip manufacture to LSI in March 2008. So, the logo is different but the chip marking is the same. According to it, Hitachi’s new HDD series have a revised controller that is expected to deliver higher performance.
The memory is different, too. There are now 400MHz chips from Samsung instead of Hynix’s 333MHz memory.
Let’s move on to the other firms, though.
Thanks to Showa Denko platters Samsung introduced the 1-terabyte model of the SpinPoint F1 series in a very short time. It even outdid Seagate with these 3-platter drives because Seagate’s first 1-terabyte drive was based on four platters.
We have this model’s results from our previous test of 1-terabyte drives but we will test a drive from a newer batch and with different firmware in this review.
In a little while Samsung realized that HDDs for desktop computers were good but it would be even better to add enterprise models to them – every other brand, save for Samsung, offered such models! As a result, Samsung’s desktop drives acquired the suffix DT and the SpinPoint F1 RAID series was introduced. Judging by the specifications, the only difference between the desktop and enterprise series is that the latter consumes more power and has an MTBF of 1.2 million hours. We wonder if the higher power consumption is due to a stronger and, accordingly, faster actuator? We’ll see…
The Barracuda 7200.11 series is very variegated. Its disks range in capacity from 160 gigabytes to 1.5 terabytes and are based on 250GB, 333GB and 375GB platters. The only thing we can say confidently about this particular model is that it is based on four platters (as is indicated by the second last digit in the model name).
Most of the models have enterprise counterparts in the ES.2 series, and we benchmarked its ST31000340NS model in our previous test session. Its results are so similar to those of the desktop drive that we don’t want to include it into this test session.
Higher recording density helps achieve the same capacity using fewer platters. This provides an advantage in speed as well as in manufacturing cost. So, as soon as Seagate had denser platters, it introduced another 1-terabyte model in the 7200.11 series. This model uses three platters and is called ST31000333AS.
The ST31000340SV is yet another 1-terabyte drive from Seagate. It comes from the SV35.3 series. Judging by the marking, it is based on the four-platter model of the desktop series.
SV35.3 series drives are intended for installation into video surveillance systems and various household appliances. The manufacturer claims them to be able to work at a case temperature of 75°C and deliver an increased data-transfer rate. Well, we won’t fry the drive up, but watch for its performance very closely.
Western Digital was reaching the capacity of 1 terabyte in a calm and steady manner. First it introduced the “green” WD10EACS which had four platters rotating at 5400rpm and only then the 7200rpm models came out. A “green” 1TB model is quite a competitive product, by the way. Although its performance is somewhat lower, it needs far less power and is quieter than its opponents. If your hard disk is used only for storing rarely used information, you don’t need it to show record-breaking performance while such properties as low noise, reduced power consumption and heat dissipation will come in handy.
People at Western Digital can count their money, too. So, they transitioned the Green drive to 3-platter design as soon as they had got the opportunity. This reduced the manufacturing cost and increased the speed of the product. Its power consumption should be lower according to the laws of physics, but there is a difference between the drives’ electronics cards. The newer model’s electronics seems to have been revised as well. The only problem is that the newer model can only be differentiated from the older one by the full model name (by the letters behind the dash, to be exact), which may be confusing for the customer.
Western Digital went further in improving its green drive and increased the amount of memory from 16 to 32 megabytes in the three-platter model. This is a reasonable solution. If the cache is utilized cleverly, it can make up for the performance loss provoked by the reduced spindle rotation speed. Good for the customer, these drives can be easily identified by the letters EADS in their name. By the way, it is in this series that Western Digital is going to introduce a 2-terabyte model with four platters. The company seems to like the idea of achieving new capacity milestones by means of 5400rpm models.
Besides desktop drives, the green technologies have spread to enterprise series. Low power consumption can be a called-for property if such drives are used in data storage systems. Judging by the 16MB cache, the RE2-GP series (GP stands for Green Power) is based on EACS disks. The difference between them can be found in the specs: the enterprise series has a higher guaranteed number of start/stop cycles (600 rather than 300 thousand), has StableTrac and TLER technologies. StableTrac means that the electric motor’s shaft is fastened at one end only in order to reduce the effect of external vibrations. TLER limits the time of error recovery so that RAID controller didn’t “lose” drives in case when a drive takes too long to correct an error and doesn’t respond.
If you need both high capacity and high speed for server applications, you can try Western Digital’s RE3 series, which has replaced the older RE2. The series features 333GB platters, 32MB of cache memory, increased processing power of the electronics, a set of enterprise technologies of the previous series, and increased vibration tolerance. The latter thing needs some clarification. There is a third vibration sensor in the RE3 for sensing high-frequency vibrations and fourth-generation RAFF technology that corrects the drive’s operation under strong angular or linear vibration.
Western Digital also caters to ordinary users who want to get the maximum from their 3.5-inch drive by offering the Caviar Black series. These drives have lower-performance Caviar Blue counterparts in lower capacities, but the capacity of 1 terabyte is only available in the Green and Caviar Black series. Like in the RE3 series, this one uses a faster controller and 32 megabytes of buffer memory, but lacks the special technologies to fight vibrations.
The following table lists the specifications and firmware versions of the drives:
So, we have a unique opportunity to see the influence of large cache memory and higher recording density on the performance of hard disk drives. We can see the former thing by comparing Hitachi’s Deskstar 7K1000.B and E7K1000 and Western Digital’s Caviar Green WD10EACS-00D6B0 and WD10EADS. In both pairs the drives have the same firmware version. It doesn’t guarantee that the firmware is identical, but there is a high chance that the drives in each pair are equal from this aspect. We also have two pairs to evaluate the effect of increased recording density: Seagate’s Barracuda 7200.11 ST31000340AS and ST31000333AS and Western Digital’s Caviar Green WD10EACS-00ZJBO and WD10EACS-00D6B0.
The following testing utilities were used:
Testbed configuration:
We installed the generic OS drivers for the drives and 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. In every test, save for IOMark and the Defragmentation test, the drives were connected to a Promise SATA300 TX4302 controller installed into a PCI-X slot and had their NCQ support enabled.
Since we’ve got drives belonging to the same series but differing in specifications, we have to use additional designations for Seagate’s Barracuda 7200.11 and Western Digital’s Caviar Green products. The designations are based on the number of platters (three or four) and, for the WD Caviar Green series, on the amount of cache memory (16 or 32 megabytes). We hope this will make the review easier to read and comprehend.
We use our internal IOMark tool for low-level tests. Let’s check out the sequential read speed of the drives first.
Now we can compare the HDDs by the read speed at the beginning and end of the partitions created on them.
There is one preliminary comment about this test: with modern adaptive formatting of platters, the speed at the beginning of the disk (i.e. the maximum speed) depends greatly on whether the disk has a lucky combination of a head and platter surface there. If there is a lucky pair, the speed is excellent. If there are two or more lucky pairs, there will be more peaks of speed and the graph will be smoother. As a matter of fact, the HDDs have very different pairs that vary in speed up to 10MBps, and even up to 15MBps with the Samsung F1 DT. As a result, our tool reports the latter’s top speed wrongly, showing a lower value.
Let’s now see what the numbers have to tell us.
First, we can see the effect of higher recording density: the 4-platter WD Caviar Green drives with a spindle rotation speed of 5400rpm are as fast as the 5-platter Hitachi. The 3-platter Green models are competitive to the 4-platter drives from Seagate, the best of them (in terms of head/surface pairs) even reach the speed of the worst of 7200rpm 3-platter drives (it is the Samsung F1 DT due to its unlucky head/surface pairs). This effect is especially conspicuous with the two WD10EACS models: the name is the same but the speed differs by 15MBps.
Then, it is good that the 3-platter 7200rpm models all are faster than 100MBps. Seagate’s drives have always had very good sequential speeds and now notch 120MBps in the 3-platter design. And if you take a look at the graph, you can see that this speed is achieved not on one particular head/surface pair, but on all pairs. Hitachi’s 3-platter drives can be commended for good results, too. The 7K1000.B and E7K1000 have the best speed at the end of the partition. They also have good speed at the beginning and smooth graphs, thus showing very stable overall performance.
And what about buffer memory?
The maximum speed of working with the buffer data depends directly on the HDD’s processor which is responsible for data transfers over the interface and on the speed of buffer memory proper (the controller the HDD is attached to influences this speed too, but we connect all the drives to the same controller in this test). Thus, this test helps compare the HDDs’ electronics.
It is clear that HDD generations share the same electronics. WD’s 4-platter Green models from both desktop and enterprise series are the worst in terms of electronics – the developer must have tried to reduce power consumption too much when designing them. Hitachi’s 5-platter model is not much faster: its result must be influenced by the “time-tested” design it has inherited from the previous series. Western Digital’s 3-platter drives obviously have new electronics and have become faster, but not fast enough to catch up with the HDDs from Samsung and Seagate. Hitachi’s new series are in the lead thanks to the new LSI-Infineon controllers – they boast a data-transfer rate of 250MBps when reading data from the buffer.
The graphs are both interesting and not. On one hand, you can easily name the manufacturer of the particular drive without seeing its name. On the other hand, you want some progress and variegation. Samsung’s and Hitachi’s products have the smoothest graphs of reading and writing. The old 5-platter Hitachi has minor problems with writing large data blocks but the company’s new models deliver almost perfectly flat graphs (we wonder if the new processor is equipped with a larger input buffer). WD’s six models all behave in the same way here, producing excellent read graphs but having some fluctuations of speed at writing. Seagate’s drives are traditionally disappointing. Their read and write speeds drop by 25% and 50%, respectively, at data blocks larger than 256KB. We wish Seagate followed Hitachi’s example.
From the low-level IOMark to the synthetic IOMeter. 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’ll discuss graphs and diagrams. It is no good drawing 14 graphs in a single diagram (it is hard to make out anything in the resulting mess), so we will divide the graphs into two diagrams: one for 5400rpm drives and 4- and 5-platter models, and the other for the fastest HDDs (i.e. with 3 platters and a spindle rotation speed of 7200rpm). Thus, we have two groups of drives: slow and fast ones.
Yes, every 3-platter 7200rpm model has a top speed higher than 100MBps indeed. And the Seagate even notches 123MBps. However, the HDDs reach this speed on different data blocks. Being the best with small data blocks, the Hitachi 7K1000.B accelerates to 100MBps on 4KB data blocks. The Hitachi E7K1000 and WD RE3 reach that speed on 8KB data blocks. The Seagate and Caviar Black need at least 32KB data chunks for that. And Samsung’s models both reach the speed of 100MBps on 64KB data blocks.
The HDDs differ with small data chunks, too. Hitachi’s drives are good then, the 7K1000.B being ahead of the E7K1000. The WD RE3 is fast, too. The Seagate and WD Caviar Black start out at a good speed but slow down afterwards. Samsung’s drives are a disappointment as they are far slower than the others at processing small data chunks.
Seagate’s 4-platter drives notch 100MBps, too. Western Digital’s 3-platter 5400rpm models are only 5MBps behind them. Hitachi’s 5-platter model and the 4-platter Green drives from WD are slow at 80MBps.
The video applications oriented SV35.3 model is different from Seagate’s ordinary drives: it achieves its top speed on 4KB data blocks like the best of its opponents. The WD Green with 3 platters and 32MB buffer is the only “slow” drive that can do the same.
The drives are all worse at sequential writing than at sequential reading but most of the characteristic features are the same. The “fast” Seagate has the best top speed while the two drives from Samsung are worse than any other 3-platter 7200rpm model in terms of top speed and in speed of processing small data chunks. The most notable change over the read speed results is that the WD RE3 is now unrivalled when processing small data chunks while the Hitachi drives are not faster than their opponents. The Hitachi E7K1000 is slower than the 7K1000.B again while the WD Caviar Black falls behind the WD RE3 on large data blocks.
The second group of drives has the same features at both reading and writing, too. The Seagate SV35.3 differs for the better from the desktop model while the 3-platter WD Caviar Green with 32MB buffer is excellent on small data blocks.
Thus, WD’s drives with 32 megabytes of cache memory deliver excellent speed when writing small data blocks. The Caviar Black is somewhat slower than them. These drives are good at reading, too, but Hitachi’s new models are competitive to them at reading. The new drive from Seagate (with 3 platters) is unrivalled in terms of top sequential speeds.
In this test IOMeter is sending a stream of requests to read and write 512-byte data blocks with a request queue of 1 for 10 minutes. The total number of requests processed by the HDD is over 60 thousand, so we get a sustained response time that doesn’t depend on the HDD’s buffer size.
Western Digital’s enterprise RE3 and desktop Caviar Black models feature the best read response time among all the drives. The result of lower than 12.5 milliseconds is very good for 7200rpm drives. Hitachi’s new models are just a third of a millisecond slower. They have improved by 0.5 milliseconds over their predecessor (but that’s not a record for this brand’s products because we have seen even better results from them). The 4-platter Seagate is also good with a response time of lower than 13 milliseconds.
What is the most interesting thing, the 3-platter WD Caviar Green with 32MB buffer has a response time comparable to that of the Samsung drives although its spindle rotation speed is only 5400rpm. The other 3-platter Green drive from WD is worse by 0.5 milliseconds, and the 4-platter models are worse by 1 millisecond. The 3-platter Seagate is quite a disappointment: 15 milliseconds is too much for a 7200rpm drive. Well, if the drive’s platters are moving slower, its noise is lower when processing a large number of random requests. But why did they stifle the drive like that? And the Seagate SV35.3 is even worse at 18.9 milliseconds!
We decided to check out this high value in more detail and see how that drive’s read access time was distributed – IOMark provides this opportunity. You should keep it in mind that we are talking about average values. The real access time depends on what exactly sectors the heads have to move between.
First, here is the access time distribution diagram for a similar product from a desktop series, the 4-platter Seagate Barracuda 7200.11.
So, everything is neat and fits within the standard distribution. You don’t have to be an expert in statistics to say that this HDD has an average response time of over 12 but less than 13 milliseconds. The top of the “hump” is exactly at that point.
Now let’s take a look at the same diagram built for the SV35.3:
The difference is notable indeed. There are few rather quick transitions (lucky situations when the previous and next sectors are in nearby cylinders) and a large number of very long seeks. This drive’s heads are moving very, very sluggishly. That’s a huge sacrifice for the sake of silence.
The write response results are interesting, too. We’ve got the same leaders: the two 7200rpm drives from WD and the Hitachi E7K1000. The 7K1000.B cannot keep up with the leaders, probably because of its 16MB buffer (the leaders are equipped with 32 megabytes of cache) that is not large enough for effective deferred writing. The comparison of the Green drives indicates this fact, too, because the model with 32MB buffer is faster. The difference between the 3-platter and 4-platter WD10EACS models is small because recording density has almost no effect on the response time (take note of the good result of the 5-platter Hitachi).
The three drives from Seagate produce shocking results. The SV35.3 has inefficient deferred writing that makes it far slower than the worst of WD’s Green models. Perhaps the difference would be small if this drive had quick heads, but the silence-oriented optimization affected its writing speed badly. The other two HDDs from Seagate have a higher response at writing than at reading. This might set us aback if we hadn’t seen the same thing in our earlier reviews. We don’t know if it is a problem of the electronics or firmware, but comparing the results of the 3- and 4-platter models it seems that Seagate has tried to solve the problem, but not quite successfully. Hopefully, the upcoming 7200.12 series will be free from this drawback.
We will do one more test in this section. It is similar to measuring the read response time but takes into account the number of platters and recording density.
So, it is 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 can run through in 1 second. The results are averaged and compared.
There are no surprises here. We’ve got the same leaders: the two 7200rpm drives from WD and two new Hitachi models. The old Hitachi 7K1000 is fifth thanks to its five platters. Then the standings repeat those of the response time diagram but Seagate’s drives take two last places. The 3-platter Seagate Barracuda 7200.11 cannot do anything even with its highest recording density: its slow heads kill its performance. The SV35.3 is only half as fast as the leader.
Now we’ll see the dependence between the drives’ performance in random read and write modes on the size of the data block size. We will only discuss the processing of small data blocks measured in operations per second. With large data blocks, the performance depends on the drive’s sequential speeds.
Random reading in small data blocks agrees with the results of the response time test. It shows the drives’ standings clearly. When the data block size changes, the speeds also change, but the relative standings remain the same. So, we can once again compliment the Western Digital RE3 and Caviar Black as well as the Hitachi 7K1000.B and E7K1000. The low results of the 3-platter Seagate Barracuda 7200.11 and the lower-still performance of the SV35.3 can be observed with any size of the data chunk.
The standings at random writing are the same as in the response time test, too. But it is interesting to note the performance of the Seagate drives here. First of all, it is the 4-platter Barracuda 7200.11 which cannot cope with writing in very small data blocks (we use 512-byte blocks in the response time test) and has very low speed with them. The 3-platter model from the same series is free from that problem. It just writes small blocks slowly just like the SV35.3. It looks like Seagate’s HDD had problems caching small write requests and the developer solved them by just turning this caching off. It’s time to move on to Database where we may see this thing more clearly.
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 the following link to view the tabled results:
We will build diagrams for request queue depths of 1, 16 and 256.
The 3-platter Seagate 7200.11 really has no deferred writing at low loads. If its writing speed were even lower, we’d even suspect it to do write verification, which can produce the same effect as the lack of deferred writing. But we can see no double performance hit when we compare this drive’s writing and reading, so this is just a lack of deferred writing. A very odd behavior for a desktop product, we should say!
We can see the effect of the 32MB cache of the E7K1000 drive in comparison with the 16MB cache of the 7K1000.B. Its deferred writing is more efficient and it is almost as fast as the WD drives that take top places. Interestingly, among the WD drives the RE3 is somewhat better at writing than the Caviar Black although enterprise HDDs usually have worse deferred writing. Samsung’s team is an example: the F1 RAID version is slower than the desktop F1 TD at high percentages of writes.
Western Digital’s Green drives are not too bad, especially the 3-platter model with 32MB cache. The first model in the series, the 4-platter EACS, is the only one that is noticeably slower than the others. It is even slower than the similar RE2-GP model.
The Seagate SV35.3 is poor: it is not meant for server loads.
When the queue is increased to 16 requests, the HDDs can use request reordering algorithms. Western Digital’s pair is beyond competition still – they are going to be leaders of this review. The Hitachi E7K1000 is good but slower than the leaders. The 7K1000.B with 16MB buffer is far from brilliant: its NCQ is all right but it lacks memory for deferred writing.
Samsung’s drives show interesting behavior. The enterprise model has more efficient NCQ algorithms, but the desktop one has better deferred writing.
The 3-platter Seagate 7200.11 has a curious graph. It is comparable to most of its opponents at high percentages of reads, but slows down to 60 operations per second as soon as there are writes in the queue.
When the queue is longer, the 5400rpm drives find it difficult to compete with the 7200rpm models. The best of them (with 3 platters and 32MB of cache) cannot catch up with the 4-platter Barracuda.
The SV35.3 has poor performance again. It seems to have been deprived of both deferred writing and NCQ.
Western Digital’s drives deliver superb performance at the maximum load: 230 operations per second! By the way, it means that the drive performs two reads per each rotation of the platters. The Hitachi E7K1000, the best of the opponents, can only catch up with the leaders at pure writing.
Otherwise, there are no changes in the standings.
The huge load is good for the 5400rpm drives. Thanks to effective NCQ they get close to the 7200rpm models and look competitive to them. The WD10EADS with 32MB cache is especially good.
Winding up this section of the review we will offer you diagrams with graphs for five different queue depths. They show what firmware the HDDs have very clearly.
Among the Hitachi drives the old 7K1000 is good enough but has a slump in the middle of the graphs. The picture changed dramatically on the transition to the 7K1000.B series: NCQ is less efficient at long queue depths but more efficient at short queue depths. Deferred writing got worse, too. When the HDD was again equipped with 32MB of cache in the E7K1000 series, deferred writing improved, but there was no progress in terms of reordering numerous write requests. It looks like the drive has got a shorter maximum length of the reordering queue.
Samsung’s products are odd, too. The desktop F1 DT either has problems with NCQ algorithms or its firmware is optimized for lower noise: a HDD just should not be slower at 100% reads than at 90% reads. The enterprise version is all right when it comes to reading but its deferred writing is suppressed. The firmware doesn’t seem to be perfect because ideal firmware doesn’t have such sudden fluctuations of performance.
Seagate’s 4-platter 7200.11 model looks good. It has excellent NCQ algorithms and effective deferred writing. But we see quite a different thing with 3 platters: the reduced performance at reading may be explained by the drive’s having too slow actuator in order to produce less noise. But what about writing? Why is there this barrier at 60 operations per second? The performance only grows up somewhat at a queue depth of 256 requests when the controller driver’s reordering algorithms come into play.
The Seagate SV35.3 is quite an incomprehensible thing. It totally lacks NCQ and has almost no deferred writing. Its performance is as low as to resemble some prehistoric HDD. Seagate offers very odd drives for video-oriented applications.
It looks like the change of the 4-platter into 3-platter design in the EACS model was also accompanied with changes in the electronics and firmware. The newer drive has stronger deferred writing and somewhat more efficient NCQ algorithms. And the addition of an extra 16MB of buffer memory came in handy: we can see both efficient NCQ and deferred writing.
The WD RE3 and Caviar Black are undoubtedly the best drives in this review for server loads. They combine excellent algorithms and fast heads. The WD RE2-GP is also good, but it lacks more buffer memory, even though it is better than the 4-platter WD10EACS due to more perfect firmware.
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 (we’ll only show ratings for the Workstation pattern in order to reduce the number of diagrams). 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.
Quite expectedly, the pair of Western Digital’s drives is unrivalled. Third place goes to the Hitachi E7K1000. The Samsung F1 DT is the worst of the 7200rpm 3-platter drives.
Well, the 5400rpm drives find it hard to compete with the 7200rpm products at reading (the Seagate SV35.3 is an exception).
Compare Hitachi’s drives: the higher-density 7K1000.B is worse than the old 7K1000 due to the smaller amount of cache memory. As for the Green drives from WD, the model with large buffer outperforms the enterprise model from Samsung and the 3-platter slow-headed Seagate thanks to effective NCQ algorithms.
It’s easy to see the leaders and losers among the “fast” drives: WD’s team are in the lead, and the 3-platter Seagate has last place. Third place goes to the Hitachi E7K1000 again. This drive is better than its mate with 16MB cache under server loads. The enterprise model from Samsung is good, too, but its firmware needs improvement as it has problems at long queue depths.
The 4-platter Seagate is beyond competition at low loads among the “slow” drives. The 5-platter Hitachi outperforms it at high loads (thanks to effective deferred writing) and the Green drive from WD with 32MB cache joins the fight then. The other Green series drives are slower.
The final ratings are interesting, too. Third place goes to the 4-platter Seagate 7200.11 rather than to the Hitachi E7K1000. The WD Caviar Green with 32MB cache takes fourth place, being somewhat ahead of the two drives from Hitachi with the same amount of cache memory. The other HDDs have similar ratings, but the two drives from Seagate with slow heads and the 4-platter Caviar Green with 16MB cache are slower.
Summing up the server tests, good firmware and a larger cache buffer are important factors for success. And if the drive combines them with quick heads, it will be unbeatable.
The change of load affects the standings: the enterprise Western Digital RE3 is better than the desktop Caviar Black. And third place goes to the Samsung F1 DT which has not been brilliant so far.
The Caviar Green with 32MB buffer outperforms every other 5400rpm drive as well as the Hitachi 7K1000.B.
When the test zone is limited, the Caviar Black rolls back to third place, giving way to the Hitachi E7K1000. The Samsung F1 DT is worse now while the 3-platter Seagate 7200.11 is closer to the leaders. The Green drives can only beat the Seagate SV35.3 in this test.
The multithreaded tests simulate a situation when there are one to four clients accessing the virtual disk at the same time, the request queue depth varying from 1 to 8. The clients’ address zones do not overlap. We’ll 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:
There are no surprises at one thread: the results are the same as in the sequential read test, the 3-platter Seagate having top place.
It is very different at two threads: Western Digital’s drives occupy three top positions, enjoying a large lead over the others. The Caviar Green with 16MB buffer is second, which is a sensational result for a 5400rpm drive. The desktop Caviar Black is better at reading two threads than the enterprise RE3. It is also interesting that the WD Green with 32MB buffer is slower than every Green series drive with 16MB buffer, including 4-platter models. The Seagate SV35.3 outperforms the 4-platter 7200.11 series model and is almost as fast as the 3-platter one. It seems to be optimized for multithreaded reading.
Hitachi’s drives with 16MB buffer are slow at multithreaded reading but the E7K1000 delivers average performance.
Samsung’s HDDs are the losers of this test. Like Seagate’s 7200.10 series products, they show very poor performance. They need just a little to show normal speed: a queue depth of 2 requests at least (you can see this if you take a look at the tables with numbers). Samsung’s new firmware must have been unable to identify two threads and worked with them as with one random read thread. It is sad since the desktop Samsung F1 model (even without the suffix DT) with older firmware had been among the leaders of this test in our previous review.
It’s all roughly the same at three threads but the Seagate SV35.3 takes fourth place now. This drive is not totally hopeless after all.
WD’s Green drive with 3 platters and 16 megabytes of cache takes first place when processing four threads, ousting its Black mate. That’s a truly impressive result.
When writing one thread, the drives repeat the results of the sequential test. The 3-platter Seagate, WD RE3 and Hitachi 7K1000.B are in the lead. The 4-platter WD Caviar Green and the 5-platter Hitachi occupy last places.
The picture changes dramatically when a second thread is added. The Hitachi 7K1000.B manages to keep its high speed but is outperformed by the Hitachi E7K1000 and the enterprise model from Samsung. The WD Caviar Black is slow. It is slower than the enterprise RE3 as well as the two 3-platter 5400rpm models. Seagate’s products have last places in this test. They are no good at multithreaded writing.
When there are three write threads, Samsung’s HDDs take top places. Third place goes to the WD RE3 while the Hitachi E7K1000 is fourth. Seagate’s drives are the losers again.
There are more changes with the addition of a fourth thread but we have the same leading trio. The WD Caviar Black is fast now. The 3-platter Seagate shows average performance while the Hitachi 7K1000.B is among the losers.
For this test two 32GB partitions are created on the disk and formatted in NTFS and then in FAT32. After that a file-set is created. It is then read from the disk, copied within the same partition and then copied into another partition. The time taken to perform these operations is measured and the speed of the disk 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, the ISO pattern having the largest files.
We’d like to note that the copying test is indicative of the drive’s behavior under complex load. In fact, the HDD is working with two threads (one for reading and one for writing) when copying files.
FC-Test produces too many numbers, so we will only discuss the NTFS data in the Install, ISO and Programs patterns. You can use the links below to view the full results:
We have the same leaders at creating files irrespective of the specific file-set. They just change their places a little. The Hitachi E7K1000 is always in the lead and is followed by the Hitachi 7K1000.B, Samsung F1 DT and WD RE3 in changing order. Seagate’s 4-platter drives take two last places. Seagate’s 3-platter model is somewhat faster but can only compete with the 4-platter Green drives from WD which have lower recording density and spindle rotating speed. The WD Caviar Green with 32MB of cache memory looks good in this test. It is almost as fast as the 7200rpm drives, outperforming the Caviar Black when processing small files.
The Hitachi E7K1000 and WD RE3 win the Install pattern. Most of the 3-platter 7200rpm drives go close to each other, except for the old Hitachi 7K1000 and the freaky 3-platter Seagate. The latter two drives even give way to the two 3-platter Caviar Green drives from WD. Seagate’s two 4-platter models are in last places again.
The same goes for the Programs pattern. Seagate’s HDDs don’t like small files at all. They are again slow, only competing with WD’s Green drives. But where is the difference due to the spindle rotation speed?
It can be seen when the drives are reading the large files of the ISO pattern. Seagate’s 3-platter model takes first place then, making the previous leaders move one step lower in the standings. Seagate’s 4-platter models feel at ease in that test, too. They are ahead of the 5400rpm drives under this load and almost as fast as the 3-platter 7200rpm models.
When copying within the same partition, the Western Digital RE3 never leaves first place. It is accompanied by the Hitachi E7K1000 with small files and closely followed by the pair of Samsung drives with large files. The WD Caviar Green with 32MB buffer looks good again, being as fast as 7200rpm models and occasionally outpacing some of them. We don’t mean Seagate’s drives, though. They are slow again, competing with the 4-platter models from WD’s Green series.
Copying from one partition into another changes the top standings: the Hitachi E7K1000 shows but modest performance here, so the top three positions are occupied by the two drives from Samsung plus WD RE3. And there are no changes among the losers.
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 use 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 tests are: Windows XP Startup is the typical disk subsystem load at system startup; Application Loading is the disk activity at sequential starting-up and closing of six popular applications; General Usage reflects the disk activity in a number of popular applications; File Write is about the speed of writing files; and Virus Scan benchmarks the disk’s performance at scanning the system for viruses. The final result is the average of ten runs of each test.
The File Copying test produces about the same results as FC-Test. The WD RE3 and Hitachi E7K1000 are in the lead. Third place goes to the WD Caviar Green with 32MB cache – this sprightly fellow leaves a number of 7200rpm models behind, including the WD Caviar Black. It is interesting to watch the progress of the Green drives: the transition from four to three platters gives them about 11MBps in this test (the increased recording density doesn’t have such a strong effect with Seagate’s drives, but they are no good for making comparisons because the two 7200.11 series models have very different firmware) and the enlarged buffer adds 16MBps more.
The Hitachi E7K1000 is somewhat better than the others at booting up Windows XP although the 3-platter 7200rpm drives all go close to each other, and even the Seagate nearly joins their group. The 5400rpm models find it more difficult to boot the OS up: the increased recording density doesn’t provide much advantage here but the model with larger buffer delivers somewhat higher performance. The SV35.3 is poor again. It is no competitor to the other 1TB drives.
WD’s RE3 and Caviar Black are in the lead when loading applications, but the Hitachi E7K1000 has nestled in between them. The latter is again much better than the Hitachi 7K1000.B which even falls behind its 5-platter ancestor in this test. Its 16MB buffer is a limitation after all. The WD Caviar Green with 32MB cache performs superbly. It joins the group of 7200rpm drives, outpacing both models from Samsung. Seagate’s three models can only compete with the a-priori slower Green models from WD.
The standings don’t change much in the General Usage test. The Hitachi E7K1000 is one step lower while the 16MB Hitachi has rolled back in the group of 7200rpm drives. The WD Caviar Green 32MB is still in that group, too. This model delivers an outstanding performance really.
Hitachi’s two new drives have no rivals when scanning files for viruses – they put their cache memory to a very good use. Next go three more models: the good result of the WD RE3 might be expected while the high performance of the WD Caviar Green and Seagate SV35.3 is only due to their firmware. As for the latter drive, it is the second time in this review that it leaves last places.
The pair of new drives from Hitachi is in the lead when writing files. Seagate’s 3-platter model is third. This brand’s HDDs find it a shame to lose in sequential operations even though they are far from brilliant in other tests. The Caviar Black is surprisingly a loser. Something is wrong with its writing. Take note of the WD Green series: the increased recording density provides an obvious effect here.
Judging by the overall HDD scores of PCMark 2005, the Hitachi E7K1000 is the best 1-terabyte drive for a desktop PC. The WD RE3 and Hitachi 7K1000.B are better than the other 3-platter drives, too.
And once again we want to single out the WD Caviar Green with 32MB buffer. Notwithstanding its lower spindle rotation speed, it is as fast as 3-platter 7200rpm drives.
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.
Do you remember that Western Digital’s drives were good at multithreaded load? Yes, they deliver high performance again, but the standings look familiar overall.
Gamers will like the two drives from Western Digital. The Hitachi E7K1000 is good, too. This trio enjoys a nice lead over the others.
The WD RE3 and Caviar Black cope better than the others with loading photos into the gallery. Funnily enough, the enterprise RE3 is ahead of the desktop Caviar Black through many tests. That was logical under server loads, but not quite under desktop loads.
The 3-platter Seagate rises up from last places to third position. This HDD seems to be asleep most of the time, waking up but occasionally. The Green series model with 32MB cache is good again. You would never tell by its results that its platters are rotting at a speed of only 5400rpm.
We’ve got the same leaders at booting the OS up (this time it is Windows Vista). The new drives from Hitachi have woken up, climbing up to places right behind the leaders’ backs. The 3-platter Seagate has returned to its hibernation again. It is slower than the less dense Hitachi 7K1000.
The pair of fast drives from Western Digital is going to win nearly every subtest here.
This subtest depends on the efficient interact of the drive’s firmware with buffer memory. This time WD’s RE3 and Caviar Black are only third and fourth, being ousted by Samsung’s team. The 5400rpm trickster is fifth: the combination of good firmware and 32 megabytes of cache memory conceal its lower spindle rotation speed well enough.
Take note that the leaders are more than two times as fast as the losers. This awesome gap makes it clear enough how important firmware algorithms are.
Everything is normal under this load: Western Digital’s drives take two top places and are followed by the Hitachi E7K1000. Seagate’s HDDs have last places, but it is clear that they won’t be heroes of this review.
Seagate’s drives leave last places under the averaged load, but the 3-platter model is the only one of them to look competitive to the Caviar Green with 32MB buffer. There are no changes among the leaders.
The overall scores of this version of PCMark differ from those of the previous version. The Western Digital RE3 and Caviar Black take first and second places and enjoy a large lead over the Hitachi E7K1000 which is third.
If you are still watching the competition within WD’s Green series, you can note that the enterprise RE2-GP model is faster than its 4-platter desktop counterpart but loses to the 3-platter WD10EACS. And all of them are far slower than the WD10EADS. Thanks to its 32 megabytes of cache memory and good firmware the latter model easily beats its mates and even competes with many 7200rpm drives.
Next goes our homemade test of defragmentation speed. We created a very defragmented file system on a 32GB partition of a hard 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 HDD we want to test. The tested HDD is connected to the mainboard’s SATA controller whose operation mode (AHCI/Standard IDE) is controlled from the mainboard’s BIOS. 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. Thus, each drive is tested twice – with AHCI support turned on and off on the controller. You can refer to this article for details about this test.
This test is yet another nail into the… Well, it’s just yet another loss for the Seagate drives. Defragmentation takes much more time of them than on any other drive, and SV35.3 spending twice the time of any 7200rpm drive to do the job.
This test is won by the two fast drives from Western Digital and the Hitachi E7K1000. Samsung’s models are somewhat slower than the leaders. Note that the 32MB cache gives no edge to the Green drive whereas the higher density of the 3-platter models helps them outperform the 4-platter versions by over 2 minutes.
You can refer to our article called Hard Disk Drive Power Consumption Measurements: X-bit’s Methodology Indepth for details on this test. We’ll just list the specific modes we measure the power consumption in:
Let’s check out each mode one by one. We are especially interested in how economical the WD Caviar Green drives are in comparison with the others.
Let’s turn them on now.
We’ve got three leaders in terms of 12V consumption. And while we could expect the 3-platter WD Caviar Green drives to be in the lead (they have three platters to spin up to only 5400rpm), it is quite a surprise to see the Seagate SV35.3 in third position. Perhaps it just starts up the way it words, i.e. very slowly, and does not require high current for that. Next go a group of medium-consumption devices. As for the losers, they are all of the 4-platter drives (save for the SV35.3), the WD RE3 and the 3-platter Seagate 7200.11. The latter two drives must have required more power in order to get ready to work in the shortest time possible.
As for 5V consumption, we have a large group of medium-consumption drives, too. The two new drives from Hitachi can be singled out among them thanks to their new electronics. The older Hitachi is the most voracious model in this test. The 4-platter Green drives from WD need rather too much power from the 5V line but their 3-platter mates are less hungry because Western Digital promptly replaced their electronics.
Well, it is hard to question the power efficiency of Western Digital’s Green products. Their 12V consumption is only half that of the other drives. Their electronics (the 5V line) has a modest appetite, too. Only Hitachi’s new models can compare with the Green series in this respect. By the way, Hitachi’s new drives are also very economical in comparison with their predecessor 7K1000 in terms of 5V consumption. Samsung’s HDDs have economical electronics, too.
When the drive is not accessed, its mechanics just keeps the platters rotating. The 3-platter models are unrivalled here: the older drives consume noticeably more. The only exceptions are the Hitachi 7K1000.B that has surprisingly high power consumption and the Seagate SV35.3 which proves to be economical.
When it comes to WD’s Green series, the reduction of platters from 4 to 3 helped lower their power draw by half a watt. This is a difference considering that the total power draw is below 4W.
And of course the Green drives have no rivals in total consumption. The Hitachi E7K1000 is the most economical model among the 7200rpm drives while the earliest 1-terabyte models, the Hitachi 7K1000 and the 4-platter Seagate 7200.11, have the highest power consumption.
Random reading is the most power-consuming mode because the HDD cannot save power. It has to move the heads about at maximum speed, looking for necessary tracks. The electronics can work not at its full capacity, however, because there is no stream of accesses to buffer memory here.
Hitachi’s new models must be noted. The old 7K1000 looks like a voracious monster while the new models are much more modest. Their electronics is very economical again while the mechanics is still demanding, even though being 2W more economical than the mechanics of the ten-headed ancestor.
It is no wonder that WD’s Caviar Green and RE2-GP drives are the best ones in this test. Interestingly, the 4-platter Green models are not much better than the 7200rpm drives whereas the 3-platter models are 50% better due to the 0.5W reduction of the power consumption of their mechanics (thanks to the lack of one platter) and the same reduction in the consumption of their new electronics. The model with 32GB buffer is especially impressive now. Its performance is just brilliant when you know that it consumes so little power.
The SV35.3 is the best 7200rpm drive, but we don’t think this is a good price for its low performance. Perhaps Seagate should have released the drive with a lower spindle rotation speed, like in WD’s products.
Deferred write algorithms came into play at random writing, and the HDD’s heads do not have to move about so quickly. But the HDD has to access its buffer memory a lot.
Western Digital’s Green drives are still in the lead thanks to their low 12V consumption but the new models’ electronics is not the most economical. The older models’ electronics is almost the most voracious among all the drives. As a result, the old 4-platter WD Caviar Green and RE2-GP consume as much power as the 3-platter Seagate (its electronics proves to be economical here). Every drive from WD has higher consumption than same-class drives from other brands – this must be the tradeoff of their effective firmware algorithms.
The Hitachi 7K1000 is the obvious loser here. It consumes very much from both 5V and 12V lines and its total power draw is 50% higher than that of any other HDD.
The HDD’s electronics contributes more to its power consumption at sequential reading because the heads only have to stay above the track.
The 3-platter 5400rpm drives are still in the lead while their 4-platter predecessors are not that good. They have low 12V consumption but consume a lot from the 5V line. As a result, Hitachi’s new drives catch up with them, showing very economical electronics. WD’s 7200rpm drives look good, too. The RE3 and Caviar Black are only 0.5W slower than Hitachi’s products and are almost as good as the RE2-GP.
The losers are the 4-platter Seagate 7200.11, Samsung F1 DT and Hitachi 7K1000. They have the highest 12V consumption, and the old Hitachi also has the most voracious electronics.
The standings are overall the same at sequential writing but Western Digital’s drives, save for the 3-platter Caviar Green, have lower places. The 5400rpm drives consume as much (or even more) power from the 5V line as from the 12V line. That’s the tradeoff of efficient caching algorithms at a reduced spindle rotation speed.
Summing up this section of the review, we’d like to note the results of the new drives from Hitachi. Their new electronics is very power efficient, and their mechanics has become more economical, too. It’s time for the 5-platter ancestor to retire now.
We will do some summarizing now. First of all, we can name the fastest 1-terabyte drive in all our tests. It is the WD RE3 (WD1002FBYS). This is the best buy if you prefer maximum performance. But if you don’t like its high price, you can take the WD Caviar Black (WD1001FALS) which is somewhat slower but costs much less. The Hitachi Deskstar E7K1000 (HDE721010SLA330) is third in terms of sheer speed.
Then, we also wanted to see the effect of increased recording density and larger cache on HDDs’ performance. It is not easy to answer what is better, actually. Taking the three drives from Western Digital’s Green Power series, we can see that it’s better to have both: the WD10EADS with higher-density platters and maximum cache is the fastest of them.
If we consider those tests that do not reveal the HDD’s maximum capabilities but check its performance under real-life conditions, we can see that a larger cache is better than higher recording density. But it should be noted that this doesn’t refer to products from all brands. Western Digital’s HDDs can utilize a large cache most efficiently, though.
And finally, is the mysterious Seagate SV35.3 worth the money? We guess not. For example, the WD10EADS has lower power consumption and lower noise but delivers higher performance. It is actually much faster than the SV35.3. Yes, the SV35.3 did well in the multithreaded tests but has last places in the others.
When it comes to choosing an optimal hard drive for home, we can recommend the Hitachi E7K1000 and Western Digital Caviar Black. Both can do well as a system disk. The WD is somewhat faster while the Hitachi is quieter and consumes less power.
The WD Green Power WD10EADS will be the best and only choice for a quiet and cold home multimedia center.
Accoridng to the results of our today's test session we decided to award Hitachi E7K1000 and Western Digital Caviar Black with our Editor's Choice title as the best HDD choice for a performance home computer system and Western Digital Green Power WD10EADS - with our Editor's Choice title as the best home theater PC hard drive:
