by Aleksey Meyev
06/29/2010 | 03:36 PM
The market of solid state drives is growing up at a rapid rate, the manufacturers regularly announcing new models. Most importantly, the competition has led to the long-anticipated decline in prices. Although flash memory itself does not get cheap too fast, the relatively large production volumes, especially compared with early SSDs that used to be nearly unique products, leaves some elbowroom for price maneuvering for the manufacturers. It is due to the competition, too, that they have to look for new customer-oriented options and extend their product range. For example, one of the SSDs to be tested today, the Intel X25-V, is the result of such extension into the low-end market sector. The rest of the products are quite exciting as well. We’ve got an updated V series model from Kingston whereas the other two SSDs represent the first steps of Western Digital, the famous HDD maker, in this field.
It is also the first time we will apply our new testing methodology to solid state drives. Let’s now take a look at each of the products we are about to test.
Products from Intel should be well known to everyone who is interested in solid state drives. It was Intel that, with its X25-M and X25-E series, dramatically changed our notions about SSDs based on MLC flash memory as those series were absolutely free from the previously common problem of low random writing performance. This dramatic improvement was achieved by means of a 10-channel controller that was very effective at caching requests as well as at distributing them among the memory cells, which is not an easy matter when it comes to flash memory. The tradeoff for such an impressive performance was that those SSDs had a lower sequential write speed than their opponents.
Later on, the X25-M series was revised. It acquired flash memory manufactured on thinner tech process (which helped double the top storage capacity), a larger cache (32 instead of 16 megabytes) and an updated controller. The second-generation controller features TRIM, a special command that helps prevent performance degradation of write operations (you can learn more about it from our earlier review).
And now we’ve got the new models. The letter E in the name of the X25-E series is for Extreme whereas the letter M in “X25-M” is for Mainstream. Thus, the new X25-V series is obviously a Value one. In fact, it was made out of the 80GB second-generation X25-M by cutting the number of flash memory modules in half. So, this SSD has only five memory access channels but the rest of its features have been inherited from its progenitor including the excellent controller, 32 megabytes of cache and TRIM (as well as NCQ support). The flash memory did not change. It is the same MLC flash memory of the joint Intel and Micron manufacture. The reduced number of memory access channels has an expected effect on the product specifications: the declared read and write speeds are 170 and 32 MBps, respectively. The series includes only one model with a capacity of 40 gigabytes but an 80GB model is expected in Q4 2010.
The price of this model is affordable indeed, so we are going to carry out an interesting experiment. We will benchmark the performance of a RAID0 array built out of two such SSDs. On one hand, this is but slightly more expensive than one 80GB X25-M drive, but on the other hand, this RAID array built using the driver for Intel’s South Bridge is free for the user. Many people may want to increase the capacity or performance of their disk subsystem, so we will see if you can do that by simply adding a second SSD to your existing one.
Simple rails for installing the SSD into a standard 3.5-inch disk bay are included into the kit. This is a good example for all other manufacturers as such rails do not cost much considering the price of the SSD itself. Users will be glad to find a means to install the SSD into their computers easily.
The firmware version of this SSD is 2CV102HD.
A model from this series participated in our tests once, but we now want to check out the 32GB one, especially as we’ve changed our testbed and testing methods. We also want to see how this small-capacity server-oriented product based on SLC flash memory compares with the newer products. It uses Intel’s first-generation controller and does not support TRIM. It has specified read and write speeds of 250 and 170 MBps, respectively.
We wonder what disk subsystem will be faster: this rather expensive SSD or a couple of affordable X-25V drives combined into a RAID array.
The firmware version of this SSD is 045C8860.
It is but recently that we reviewed the first-generation V series from Kingston and now we have a second-generation model in our hands. The differences are dramatic. While its predecessor was based on a JMicron 602B controller, the second-generation V series features the newer Toshiba TC58NCF618G3T, which is a slightly modified version of the JM618 controller. This SSD has MLC flash memory and 64 megabytes of cache. It is declared to support TRIM. The 128GB model is specified to have a read speed of 200 MBps and a write speed of 160 MBps. The lower-capacity models (30 and 64 gigabytes) are somewhat slower at writing.
The firmware version of this SSD is C091126a.
Major HDD makers couldn’t neglect such a rapidly developing and potentially dangerous market of solid state drives but were not among the pioneers of it. It is only recently that they have begun to produce SSDs of their own. Western Digital is at least ahead of its main rival Seagate in providing SSDs for us to test.
The first thing you can note about these products is the word Blue in the series name. Western Digital’s HDDs fall into three main series and the Blue series is not the fastest among them. Thus, we can expect this manufacturer to produce a Black series in the future which has to be even faster. As for the Blue series, it currently includes three models with capacities of 64, 128 and 256 gigabytes. They are all based on MLC memory and a JMicron 612 controller, possibly modified by Western Digital, which is backed up by 64 megabytes of cache. The series is declared to support NCQ and TRIM and deliver peak read and write speeds of 250 and 170 MBps, respectively.
These SSDs both have the same firmware version: 5.12.
The following testing utilities were used:
The SSDs were tested with the generic OS drivers and formatted in NTFS (wherever formatting was required) as one partition with the default cluster size. 32-gigabyte NTFS partitions with the default cluster size were created for FC-Test (if the drive is smaller than 64 gigabytes, it is partitioned in two halves). The SSDs were connected to a mainboard port and worked with enabled AHCI. The sequence of tests is absolutely identical for each SSD, so all of them are under the same conditions.
The most dramatic change in our new test method is the transition from the outdated Windows XP to Windows 7. Windows 7 is especially good for SSD tests as it supports the TRIM command. As for the hardware, our testbed now includes a mainboard with an Intel ICH7 South Bridge. This controller is widespread and does not depend on the peripheral bus bandwidth as standalone disk controllers do.
There are some changes in the list of our tests, too, although it is based on the old one. First, we have finally got rid of PCMark 2004 and 2005, leaving the Vantage version only. These tests largely duplicate each other or other tests and produce similar results. Besides, we have some suspicions that the next version of this benchmark is about to come out, 3DMark 2010 having been announced already. Then, we have abandoned the Workstation pattern because PCMark Vantage provides a better picture of a disk subsystem’s workstation performance. We now use WinRar version 3.91 and have replaced Perfect Disk with the Disk Defragmenter integrated into Windows 7.
Finally, we have adjusted some of the IOMeter tests, but this is only important for hard disk drives. For example, we now test our disks in more detail under random-address loads, using a step of 2 rather than 4. The maximum data block size is now 2 megabytes, the largest that modern Windows OSes employ. If the disk request is even larger, the performance becomes influenced by the HDD’s sequential speed. However, for our SSD tests we will use the older method for a while, although we do not compare SSDs’ performance on large data blocks anymore.
We will also use the old method of testing multithreaded performance with a distance of 8 gigabytes between the threads (we have increased this distance to 100 gigabytes in our HDD tests). The reason is simple: SSDs have not yet reached such capacities as to allow for four separate 100 GB data threads. As a matter of fact, the distance between the threads is unimportant for SSDs. While it affects the speed of an HDD (because the angle of turning of the read-write heads depends on it), an SSD only has to cleverly handle this load and read from all the controller channels in parallel.
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.
Well, the sequential read speeds of all the drives are close to the specified levels. The difference of 10 MBps can be written off for some variations in the measurement method. The manufacturers may have their own way of benchmarking their SSDs. The RAID array of two X25-V drives delivers an impressive 320 MBps which just wouldn’t be possible with any SATA 300 drive. It is clear that the SSDs will have a performance boost when transferred to SATA 600. The array is also faster than the single drive on small data blocks, but Intel SSDs are generally superior in this respect.
We’ve got an interesting picture at writing. The X25-E goes ahead although closely followed by the MLC-based opponents. The difference between the two models from Western Digital is conspicuous, the 256MB one being faster. The graph of the X25-V goes lower due to the reduced number of memory access channels. This series is going to be inferior not only to the other SSDs but also to 2.5-inch HDDs. The RAID0 array improves the situation, producing a nearly twofold performance boost.
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 SSD is much larger than its cache, so we get a sustained response time that doesn’t depend on the SSD’s buffer size.
Intel SSDs have always had an extremely low response time at both reading and writing, and the X25-V series is no exception. The rest of the SSDs are somewhat slower but without such fatal shortcomings as we could see with the earlier SSDs based on the JM602B controller. The highest response time at writing is actually no higher than 0.5 milliseconds.
Take note that the 256MB drive from Western Digital has a higher response time at writing than its smaller-capacity cousin although the latter should theoretically have the same or somewhat worse response. Comparing the controllers, the JM618 in Toshiba’s version is somewhat worse at reading but much better at writing than the JM612 in WD’s version. But again, none of the SSDs betrays any serious problems in this test.
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’ve got some interesting things at random reading. First, the pair of Intel X25-V drives in RAID0 are very good, competing and occasionally outperforming their SLC-based opponent. Second, the Intel controller loses its advantage on 32KB data blocks but is superior on smaller ones. Third, the JM612 controller in the WD drives is better than the JM618 in the Kingston V series.
Funnily enough, the graph of the 128GB model from WD is nowhere to be seen in the diagram although it is indeed there. At this scale, the results of the two SSDs from Western Digital are so similar that their graphs coincide.
We’ve got a lot of interesting things at writing. First, the RAID0 array with two X25-V is ahead of the X25-E on small data blocks. The single X25-V is good as well, although the flat stretch of its graph at 10,000 operations per second makes us suspect some architectural limitations.
Comparing the two models from Western Digital, the larger-capacity one delivers more stable results although falls behind its 128GB cousin on some data blocks. The JM618-based Kingston is faster than both drives from WD but cannot match the performance of the “value” X25-V.
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 patterns.
We will build diagrams for request queue depths of 1, 16 and 256.
Once again we see the two SSDs from Western Digital behave in the same manner. It is clear that the junior model is just based on smaller-capacity chips but has the same architecture as the senior one. We’ve actually got two groups of drives in this test. The first and faster group includes the SSDs from Intel: the X25-E is expectedly better than its cousin, especially at writing. What is more surprising for us, the RAID array of two X25-V drives can challenge the X25-E throughout a wide range of loads, being only inferior to it at very high percentages of writes.
The rest of the SSDs are in the second, slower, group. We can note that the WD drives are better at pure reading and at write percentages up to 20%. The Kingston, on its part, manages to keep its performance up at high percentages of writes. It is only at 60% writes that it does not cope and slows down to 100 operations per second. Anyway, these three SSDs (and their two controllers) look much more appealing than the notorious JM602B controller that used to be a total failure at high percentages of writes.
When the request queue is longer, the X25-V improves its standing. The RAID0 driver seems to have highly effective deferred writing algorithms which ensure a large advantage of the array over the X25-E everywhere but at extremely high percentages of writes.
Winding up this part of our tests, we want to show you diagrams that illustrate each SSD’s performance at five different request queue depths.
We can note that Intel’s SSD controllers can effectively work with long request queues. They can identify a long queue and increase the performance accordingly.
The effect from RAID0 can be observed easily. Just take a look at the performance growth in the left part of the diagram, which extends rightwards, towards the higher writes, as the request queue grows longer.
The JM612-based controller from Toshiba is very different from its predecessors and from the JM602B due to having cache memory or to some serious changes in the operation algorithms. First, this SSD can reorder requests, producing a performance growth at request queue depths other than 1. This is not too effective as the graphs for queue depths of 4 and 256 requests are almost identical. Anyway, it supports NCQ, which is good. Second, the controller copes successfully with low percentages of writes. This was an insurmountable task for the JM602B controller installed into the first-generation Kingston V series as well as for the Toshiba T6UG1XB installed into the Kingston V+.
The JMicron 612 controller in the Western Digital drives is better than its predecessor, too. It features NCQ (in small amounts, too) and its performance is higher at both reading (the read speed is higher than that of the JM618 and the difference cannot be due to their possibly having different flash memory) and writing (alas, the write speed is still much lower compared to the Intel and Indilinx controllers). We have an impression that JMicron developed this controller on the basis of the JM602 but took some ideas from the Samsung PB22-J because the graphs are shaped characteristically.
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. We’ve removed the Workstation load because PCMark has similar tests based on more modern applications.
This test can but rarely surprise us after what we see in the Database pattern. Here, the Kingston is expectedly last whereas the WD drives are inferior to the X-25V. The only unobvious thing is that the X-25V array is so much faster than the single SSDs from Intel.
The standings change somewhat when there are write requests to be performed. The X-25E is fast until long request queue depths where its graph swoops down (this is an interesting thing, perhaps indicative of the “fragmentation” of the SSD’s memory that occurs when the controller is constantly under load). The X25-V delivers stable performance and takes first place.
As for the slower products, the JM612 controller seems to find it hard to process write requests. It slows down after filling up its cache. This is the only explanation of its performance hit at long queue depths. The Kingston with the new Toshiba controller performs consistently and does not show such problems.
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:
The SSDs are rather good at multithreaded reading except for two products. The Kingston suffers a twofold performance hit and the X25-V suddenly feels bad at four data threads. This may be a problem of Intel’s drivers because the single SSD does not show such behavior.
These SSDs are indifferent to the number of write threads. They even try to speed up because the multiple threads are not unlike a requests queue.
For this test two 32GB 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 disk but is also indicative of the latter’s behavior under complex load. In fact, the disk is processing two data threads then, one for reading and another for writing.
The X25-E is unrivalled at writing, especially when it comes to processing large files which are hard to hide in the OS’s cache. As for the others, the Kingston prefers large files whereas the WD products prefer small ones of roughly the same size. The 256GB model from WD looks better than its junior cousin. The Intel X25-V delivers its specified speed, which is rather low by today’s standards, both when single and in the RAID array.
The X25-V based RAID is just excellent at reading. It is nearly as fast as 300 MBps when reading large files. No other SSD can reach such a high speed due to the limitations of the interface. Take note that the single X25-V performs poorly, often finding itself in last place. The Kingston is not good with small files. The WD drives are better in such patterns.
The X25-E wins the copying test, obviously thanks to its excellent speed of writing. The Western Digital team performs well, too. The Kingston again finds it hard to match its opponents’ performance with small files.
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 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.
The Intel SSDs are splendid again. Although the X25-V occasionally suffers from its low sequential write speed, its low response time and good read speed make up for that easily. If two such SSDs are combined into a RAID, they become simply brilliant and even beat the X25-E.
The WD drives come to the finish with nearly identical results, once again proving that the performance of this series does not depend on the storage capacity. Their speed is quite high overall. As opposed to the slower Kingston, they are roughly comparable to SSDs based on the Indilinx controller in this test.
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.
We must remind you that defragmentation is useless and even harmless for solid state drives due to their operation principles. However, we use this test as it allows to benchmark their performance at a rather peculiar load that involves both reading and writing of small data blocks.
The X25-E does not participate in this and next tests because our test image is exactly 32 gigabytes large and does not fit on that SSD.
This test proves how real-life loads may differ from theory. This test is won by the 256GB Western Digital whereas its 128GB cousin spends twice the time and takes last place. The X25-V array is 50% faster than the single such SSD.
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 RAID is surprisingly poor in this test. Its results are as low as if the array was doing anything but archiving most of the time. The rest of the SSDs fall into two groups: the faster group includes the 256GB WD and the Kingston. The X25-V and the 128GB WD are in the slower group.
It’s simple in the unzipping test: the two drives from Western Digital share top place, leaving their opponents far behind.
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:
We don’t publish the results of the X25-V RAID because our testbed only allows measuring the power draw of one drive at a time. You can just multiply the results of the single drive by 2, though.
Let’s check out each mode one by one.
The Intel SSDs need only half the current required by the other models. Interestingly, the 256GB Western Digital consumes much more power than its junior cousin which must be due to the effect of its larger-capacity flash memory chips. That model from WD and the Kingston V series require over 1 ampere, which is more than today’s 2.5-inch HDDs need.
The Kingston is the only SSD to act up in idle mode. It needs twice the power the other SSDs want. It either cannot switch into sleep mode with reduced power consumption or was busy minding some business of its own.
At random loads, the Kingston needs a lot of power at writing while being quiet economical at reading. It is the Intel SSDs that are the most economical at reading, though. They are also as economical as the WD drives at writing.
The Intel SSDs are also superior at sequential operations. The WD drives consume more than the others at reading just as the Kingston does at writing.
We’ll do some summarizing now, starting from the Intel X25-V. Cutting it short, we like this product. Intel has come up with an affordable but rather fast SSD. It was especially good in our RAID0 array, providing both a large storage capacity and a high speed of writing. As for reading and mixed loads, the RAID0 built out of two X25-V drives was just brilliant, outperforming its server-oriented X25-E with SLC memory. The latter is blameless, though. Its performance is still competitive in comparison with MLC-based SSDs with new controllers inside.
The first step of Western Digital in the world of solid state drives is not a failure, either. Although the JM612 controller is not as fast at writing as the Indilinx, the SiliconEdge Blue series offers a decent writing performance and will make a good choice for workstations or top-end gaming computers.
The updated Kingston V series looks good, too, especially compared with the extremely low performance of its JM602B-based predecessor. On the other hand, this model looks inferior to its opponents due to its low writing performance and high power consumption. However, we like this SSD’s performance under mixed loads (i.e. with both read and write requests).
The SSD market is on the rise, so stay tuned for our upcoming SSD reviews!