by Aleksey Meyev
10/15/2008 | 09:42 PM
About a year ago flash memory could only be used as the main storage by PC enthusiasts or in specialized industrial computers. There were many reasons for that: low capacity, low speed of linear operations, a very high cost of storage (in comparison with hard disk drives), and an inconvenient interface (Compact Flash cards were connected to a PC as PATA drives via adapters). However, the sudden reduction of prices for flash memory had led to an explosive production of solid state drives designed in the same form-factor as standard 2.5-inch HDDs and equipped with a SATA interface.
What is the reason for this enormous popularity? Flash memory based storage has very appealing advantages in comparison with ordinary hard disks that store data on magnetic platters. Here are some of them:
Of course, there are no ideal things in this imperfect world, and SSDs have drawbacks as well, particularly:
Half a year ago, when we were testing early SSDs from Samsung and talking about the highs and lows of this type of storage in comparison with other types, there was virtually no SSD market at all (for details see our article called SSD, i-RAM and Traditional Hard Disk Drives). There were but a few early models available. The choice got broader in the summer, but there were too many small makers and no big players. It was at the August IDF that Intel declared its far-reaching plans on producing solid state drives. The company declared both quick models for the server market and mass models for home/office applications. Voiced by such a reputable company as Intel, the plans had to be taken seriously especially as the industry giant had long had a joint venture with Micron on producing NAND-type flash memory. The declared specs were astonishing: a read speed of 250MBps and a write speed of 70MBps even for the cheaper and slower multi-level-cell memory! The arrival of big players with enticing offers into a sector of rapidly developing and perspective products might have been predicted. We can only feel for smaller companies as there is going to be a war for survival like many wars in the past of the computer world. You can recall the history of hard disk drives or the time of the active development of desktop processors. We guess only large companies with their own manufacture of flash memory (such as Intel, Samsung, Toshiba) will stay on the market while the others will be ousted into narrow segments such as entry-level or specialized products. The end-user will benefit from this war because it will surely lead to price cuts.
We didn’t have to wait long for Intel to fulfill its promise. And now we have a device called Intel X25-M in our hands. It is a mass-produced solid-state drive of 2.5-inch form-factor with a storage capacity of 80 gigabytes.
Intel’s SSD line-up includes three series right now. The X25-M and X18-M series are based on MLC memory, have a capacity of 80GB and differ in size: they are designed in 2.5-inch and 1.8-inch form-factors, respectively. The X25-E series is based on SLC flash memory and has a capacity of 32 gigabytes. Intel is going to extend all the three series at the end of the year by adding models with twice higher capacities, i.e. 160GB and 64GB for the mass and extreme series, respectively.
The X25-M looks like any other SSD. It has the same rectangular case of 2.5” form-factor, the same size and fastening, and the same position of the connectors. The case contains twenty 4GB chips manufactured on 50nm tech process and marked as 29F32G08 (we couldn’t find information about such chips but there are Micron chips that have a similar marking and specs; as we have noted in the Introduction, Micron and Intel have a joint manufacture). Besides the flash memory, there is a PC29AS21AA controller and a 16MB module of 166MHz SDRAM (Samsung K4S281632I). So, we didn’t have to wait long for SSDs to acquire buffer memory. We don’t know if it is used for the drive’s internal purposes (we’ll explain below what purposes these could be) or as a buffer for organizing a queue. Anyway, it is good that there is buffer memory here.
The SSD owes its high speed to the ten-channel controller (one channel for each two memory chips). Thus, we have a kind of RAID0 built out of ten flash memory chips, the controller distributing the load uniformly among all the chips. The resulting speed of the whole SSD is roughly equal to the speed of one chip multiplied by the number of channels. This is the explanation of how Intel managed to achieve such a high speed of linear operations out of rather slow MLC chips. That’s a logical solution. If simple USB flash drives come with dual-channel controllers now, an SSD just demands a multi-channel controller. Coupled with buffer memory, this should solve the problem of low sequential speeds of flash memory that was the main drawback of SSDs which we noted in our previous review. You will see in this review what speeds this new solution can offer.
The declared specifications promise superb speeds of writing and reading. The write speed is lower than that of last-generation 3.5” hard disk drives that have become faster than 100MBps, yet it is very good for flash memory, especially for multi-level cells. The read speed of 250MBps is impressively high. The operating temperature range is good, too. Flash memory used to work at temperatures not higher than 55°C but the X25-M is rated for temperatures up to 70°C. Such a high operating temperature used to be declared for industrial flash memory that differed considerably from the ordinary type in terms of pricing.
The service life parameter is most important, too. It is usually declared for flash memory as the number of rewrite cycles. For modern MLC chips it is declared to be 10,000 cycles (this parameter equals 100,000 cycles for SLC chips, which is the second most important difference between these two flash memory types, besides write speed). This is somewhat inconvenient because you can’t compare the number with the service life of hard disk drives. And it is overall unclear for how long your SSD is going to work. Of course, you can multiply the number of cells (or, to be exact, the number of pages because the minimum changeable unit of information with flash memory is a page; the X25-M has a page size of 4 kilobytes) by the capacity of each and then by the number of rewrite cycles and get the maximum amount of information you can write to the drive. This value is 800 terabytes for the X25-M. It seems to be a big number, but there are two problems about it.
First, this number includes all possible write operations including auxiliary operations. PC enthusiasts have already learned that the share of auxiliary operations is high when the OS is installed on a flash drive, and the service life of this drive proves to be surprisingly short. Recording the date of accessing a file is a write operation, for example, and affects the drive’s service life. There are quite a lot of such operations that may not be obvious for an inexperienced user because modern OSes and programs keep various log files, work with the file allocation table (located on the drive) and change their internal files (the page-file alone is being changed constantly). All of this involves write operations. As a result, 800 terabytes is not the total of files the user can write to the drive during the latter’s lifetime, removing old and writing new ones. The real number is far lower even if you try to reduce the amount of such unobvious writes by turning off the page-file and setting your OS up accordingly. We wonder if notebooks with SSDs which are becoming popular now undergo such optimization.
The second problem roots even deeper, in the very algorithm of operation of flash drives and their interaction with OSes. A page is the minimum size of a data chunk at writing whereas the smallest data chunk for erasing is a block consisting of several pages. This is determined by the mechanism of erasing data in flash memory. For the X25-M this block is as large as 128 pages or 512 kilobytes or half a megabyte. As a result, if there is a request to erase (or rewrite) one page, the drive has to erase 128 pages. If there is useful information in these pages (which is a likely situation), this information has to be first copied into the cache and then written back. As a result, the number of write operations is increased manifold, with a proportional reduction of the drive’s service life.
Another aspect of the problem is that the load is distributed unevenly among the flash memory cells. Some cells store the same information for very long without any changes because some files are only read but not modified. Other cells, on the contrary, are often rewritten. Thus, the drive degenerates unevenly. Some of its cells may still undergo a number of rewrite cycles when other cells have already reached the limit.
And the third aspect of the problem is the interaction of SSDs with the file system. To remind you, the OS uses the file system to access the drive by means of the LBA addresses of data blocks it wants to request. LBA addressing is natural for HDDs but flash memory doesn’t have rotating platters divided into tracks and sectors. So, flash memory is equipped with an address translator that knows which physical cell corresponds to which LBA address. What may be a problem here? The problem is that the modern protocols do not erase data from all the LBA addresses when removing a file. They just change the part of the allocation table that refers to the removed file. Coupled with the address translator, it means that data written into a memory cell can be considered present even when from the file system’s point of view the file containing this data is removed. And recalling the above-described peculiarities of erasing data on the physical level, it means that nearly every request to delete data necessitates the writing back of even those data that belong to the already removed file.
This may be somewhat hard to grasp, so let’s put it in a different way: the LBA address translator that is responsible for removing data from cells corresponding to specific LBA addresses does not know anything about information stored in the file system. When deleting a file, the file system does not send a request to remove data from all the LBA addresses corresponding to the file. As a result, information in memory cells remains effective for the SSD (from the SSD’s point of view, it must be stored still) even when those cells belong to files that are already removed from the OS’s point of view.
The job of reducing the overhead (i.e. the amount of rewrites for each cell) associated with the described problem must be done by the address translator. In fact, the table of correspondence between LBA addresses and physical cells is not fixed and constant. It is dynamic. At different moments different cells may correspond to the same LBA address. So, the translator’s algorithms try to change the table in order to distribute the load uniformly among all cells, to achieve the same level of wear for all of them. This is where the SDRAM module installed on the SSD’s card comes in handy.
The translator is also responsible for reducing the wear from the erasure of cells in blocks. As you may guess, the efficiency of these algorithms affects the SSD’s performance as well as service life. Particularly, the drive takes some time to adapt to a change of load (for example when switching from sequential to random-address reading).
But let’s emerge from the technical depths onto the end-user’s level. What does the service life of the drive mean to us? Intel has suggested an elegant solution. The service life of an SSD can be determined basing on the average amount of written data while the above-described technical problems that increase the effective amount of written data is accounted for with two weights. The write amplification weight reflects the above-described problem with the increased amount of rewritten data. The wear leveling weight makes up for the reduction of the service life due to the fact that some cells wear off faster than others.
Intel evaluates the efficiency of its drive’s algorithms quite high. The company uses a write amplification weight of 20 and a wear leveling weight of 3 for the competitors’ MLC-based products. For its own products the company sets both weights at 1.1. Of course, we can’t check this out, so we have to trust Intel. Intel guarantees that if you write 100 gigabytes of data onto its SSD, the latter will serve you for 5 years (but the company provides a 3-year warranty for this product). It is, however, unclear how an ordinary user can measure the amounts of writing happening on his computer. It seems like we have to rely on statistical data and on the manufacturers’ claims.
But what does Intel mean by the MTBF of 1.2 million hours (as long as 137 years)? What does this impressive number mean and how was it arrived at? Obviously, it is calculated basing on statistical data and means the probability of the device’s failure after a certain period of time. But we don’t know what loads this MTBF is calculated for. Perhaps we’ll find this out in the future.
What we can check out right now are the speed characteristics of the X25-M. We took the drives discussed in our previous SSD-related review as its opponents. These HDDs still remain among the best in their market sectors. Here they are:
We did not include 2.5-inch hard disk drives into this review because they would just clutter the tables and diagrams. Such HDDs wouldn’t have had a chance among such opponents.
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. The drives were connected to a Promise SATA300 TX4302 controller installed into a PCI-X slot. The HDDs were switched from the quiet mode (with Advanced Acoustic Management enabled) into the ordinary operating mode when necessary. The SAS interface was provided by an LSI SAS3041E-R controller. Of course, the drives are all tested with enabled AHCI (by the way, Intel strongly recommends turning it on).
Unfortunately, this review is just our “first date” with the new device. We didn’t have enough time to test our Intel X-25M in all of our traditional tests, particularly in IOMeter’s Database pattern, in PCMark Vantage, and in power consumption tests. We are going to make our amends for this omission in future reviews where we will test Intel’s SSDs in much more tests.
We started out with the simplest of tests. We recorded the drives’ data-transfer graphs in WinBench 99.
Read graph: Intel X25-M SSD on Promise SATA300 TX4302
Well, we got an ideal flat line, just like the theory suggested, but it went at 138,000KBps whereas the manufacturer declared a speed of 250MBps. We trusted Intel and suspected a problem with our testbed. So we recorded the data-transfer graph once again on a different computer. Our testbed for benchmarking with IOMark, our internal program for testing hard disks, is exactly alike, but the tested drive is connected to the ICH7 in it, which is necessary for IOMark to work.
Read graph: Intel X25-M SSD on ICH7
That was quite a different thing. The max speed was now 249,000KBps, which was almost exactly like the declared value. Take note of the considerable fluctuations of speed in the first half of the diagram that lasted for about two minutes and a half. We guess it is the consequence of the address translator adjustment (the SSD had previously undergone a full test cycle in WinBench 99 which was quite a variegated load) in order to ensure maximum performance in this operation mode. This is a normal thing for the translator to do and Intel speaks about that openly. On one hand, two minutes and a half is not too long, especially when the drive eventually delivers maximum performance and ensures a longer service life for the flash memory cells. On the other hand, it is somewhat unpleasant to know that there is something going on inside your SSD that affects its performance. It seems that we must get rid of our notions about the typical behavior of a storage device that we have got used to with hard disk drives.
For comparison, here are the data-transfer graphs of the other drives on the Promise controller:
Samsung SSD on Promise SATA300 TX4302
|
|
And a diagram for better readability:
So, even slowed down inexplicably on the Promise controller, the Intel X25-M proves to be faster than any existing HDD including the 15,000rpm model which belongs to the elite of the HDD world. The SSD has no rivals when working at its best on the ICH7 controller. You’d need to unite two HDDs into a RAID0 in order to achieve such a high sequential read speed. We guess this is a most illustrious example of the ten-channel controller’s efficiency. Just compare this to the performance of Samsung’s SSD!
To make sure the speed characteristics were all right we launched our IOMark. This program is intended for HDD tests but it can tell us something about SSDs, too. The X25-M confessed that it had a read buffer with a capacity of 3281KB and knew about look-ahead reading, performing it in 8-sector blocks (8 sectors equals 4KB which is the size of a page in flash memory). But of course we were more interested in speed.
It’s all right with sequential reading here. The neat graph goes at 248MBps even on very large data blocks. Take note that the write speed is even higher than specified by the manufacturer: most of the graph goes at 79MBps.
The speed of working with the buffer is somewhat surprising. Reading from the buffer is slower than the SSD’s sequential read speed (although higher than the speed of the SSD on the Promise controller). Writing from the buffer is performed at a varying speed which jumps very high and then drops to the speed of sequential writing. This must be due to the deferred writing algorithms and the address translator’s operation.
The results of our low-level tests are ambiguous. The SSD proved to be true to its own specifications. Its sequential speeds are indeed very high, the speed of sequential reading being far higher than that of any modern HDD (note that we are talking about sequential operations which were the sore spot of early SSDs, not about random-address operations). On the other hand, the SSD had lower performance on our Promise controller. Frankly speaking, it looks like our controller made the SSD work in SATA-150 rather than SATA-300 mode (the controller chooses the mode automatically without user’s intervention) or the interface bandwidth was too low (but that shouldn’t have been the case as we installed the controller into a PCI-X slot that had a frequency of 100MHz rather than 33MHz). We decided to perform our tests on the Promise controller due to time constraints but we promise that we will soon retest Intel’s SSD under better conditions.
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 highest speed the drive can achieve.
The numeric data can be viewed in tables. We’ll discuss diagrams.
IOMeter: Sequential Read results
The X25-M doesn’t deliver the promised 250MBps or even those 134MBps that we saw in the data-transfer graph. This doesn’t seem to be the controller’s fault only but we will only be sure of that when we retest Intel’s SSD. Anyway, Intel has something to be proud of. Its MLC-based drive is as fast as the best of 3.5-inch HDDs and only inferior to the 15,000rpm SAS drive even under our imperfect test conditions. This is a tremendous progress since early SSDs that used to be slower than 2.5-inch HDDs. Interestingly, the speed of the X25-M keeps on growing till the right edge of the diagram, contrary to the graphs of the HDDs. It seems that if we used even larger data chunks, Intel’s SSD would overtake the SAS drive as well.
IOMeter: Sequential Write results
The X25-M is somewhat worse at writing because such operations are the most difficult for MLC-based memory. It is not a failure, though. The 3.5-inch HDDs deliver higher top speeds but 70MBps of the X25-M is quite a high speed for many applications. This is especially conspicuous on small data chunks where the X25-M is ahead of the Samsung F1 and close to the Fujitsu which has a huge cache, optimized firmware and very fast platters. Yes, buffer memory and a good multi-channel controller make wonders to flash drives!
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.
Response time at reading is one of the main trumps of flash memory. Even early models of SSDs were far better than 15,000rpm HDDs in this parameter. Intel has managed to improve it threefold. Its SSD has a read response of 0.07 milliseconds in our test. This fantastic result must be due to the fast controller. Take note that the read response is most important for your user experience because it is this parameter, not the sequential speed, that determines the delay when you try to read small files. And this parameter is among the hardest to improve. In fact, there is only one way to reduce the read response of HDDs – to increase the spindle rotation speed. The speed of the read/write heads and of the electronics doesn’t have a big effect on this parameter. You can’t improve it even by uniting HDDs into a RAID array: this can increase the amount of operations per second at high loads as the requests will go to the different disks, but the access time of each specific disk will still depend on the spindle rotation speed of the disks you use.
Well, there is even more progress with the write response of the SSD. Early SSDs had a very high write response, many times higher than that of HDDs, but the write response of the X25-M is very low. We were indeed shocked at such a tremendous progress in this parameter, but it was confirmed throughout our tests (and our method of the response time test doesn’t allow the drive to cheat – it just can’t do anything else with such a large amount of requests). The X25-M not only overtakes but beats the best of HDDs thanks to its multi-channel controller, efficient algorithms and buffer memory!
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 discuss the results of the disk subsystems at processing random-address data in two versions. For small-size data chunks we will draw graphs showing the dependence of the amount of operations per second on the data chunk size. For large chunks we will compare performance depending on data-transfer rate in megabytes per second. This approach helps us evaluate the disk subsystem’s performance in two typical scenarios: working with small data chunks is typical for databases. The amount of operations per second is more important than sheer speed then. Working with large data blocks is nearly the same as working with small files, and the traditional measurement of speed in megabytes per second becomes more relevant.
IOMeter: Random Read, operations per second
Random reading in small blocks is where flash memory knows no rival unless you count in such specific devices as RAM-based drives. Even early SSDs were incomparable to HDDs in this test. The ten-channel controller employed in the X25-M improves performance even more. For example, the X25-M delivers almost 6,000 operations per second on 8KB data chunks (not the smallest, but frequently used size of a data block) whereas the best of HDDs only deliver 200 operations per second. As you can guess, even a multi-disk RAID cannot match the SSDs here.
The difference isn’t big on large data blocks but the X25-M is still in the lead. Samsung’s SSD is slower than the HDDs after a certain size of the data chunk due to its low sequential speed.
IOMeter: Random Write, operations per second
The dramatic reduction of response time at writing helps the X25-M win the test of random writing in small blocks whereas Samsung’s SSD is far slower than its opponents (it was slower even than 2.5-inch HDDs in our previous test session). The twofold advantage over the 15,000rpm HDD in terms of random-address write operations is a superb result. Take note of the sudden performance jump on 2KB data blocks. Our test program doesn’t include 4KB data blocks, so we’ve got a question we will try to answer in our next review: if the drive’s performance goes down on 512-byte data blocks or there is a sudden increase in performance on data chunks of 4KB or similar size? Anyway, this is a superb result for a MLC-based flash drive.
It is the speed of sequential writing that affects the results on large data blocks. Samsung’s SSD improves somewhat as the data chunks grows larger (its write access time was even more of a problem than its low sequential write speed) while the X25-M maintains the same speed throughout the test. Although the latter is somewhat slower than modern HDDs on very large data blocks, the gap isn’t large. The ten-channel controller helps this SSD again.
The multi-threaded tests simulate a situation when there are one to four clients accessing the hard disk at the same time. The depth of the outgoing request queue is varied from 1 to 8, and the address zones of the applications (called Workers in IOMeter) do not overlap.
You can follow the links below to see tables with results, but we’ll discuss diagrams for a request queue of 1 as the most illustrative ones. When the queue is longer, the speeds depend but little on the number of applications.
Multithreaded reading is a very difficult operation for HDDs because they have to move their read/write heads among multiple streams located in different parts of the platters, but for an SSD this load is hardly different from ordinary sequential reading. An SSD doesn’t care about the order of reading. As a result, the X25-M does not slow down but even accelerates in this test, winning it due to its high sequential speed.
It’s different with multithreaded writing. The drives can now collect data in their cache and write in large blocks. Writing is more difficult for an SSD than reading, and it has a lower speed of sequential writing even though it doesn’t have to move anything among the test zones. Thus, an SSD’s speed of multithreaded writing is similar to its sequential write speed. The latter is quite high with the X25-M, so it is not far slower than the HDDs in this test.
The drives are tested under loads typical of servers and workstations.
The names of the patterns are self-explanatory. The Workstation pattern is used with the full capacity of the drive as well as with a 32GB partition. The request queue is limited to 32 requests in the Workstation pattern.
The results are presented as performance ratings. For the File-Server and Web-Server patterns the performance rating is the average speed of the drive under every load. For the Workstation pattern we use the following formula:
Rating (Workstation) = Total I/O (queue=1)/1 + Total I/O (queue=2)/2 + Total I/O (queue=4)/4 + Total I/O (queue=8)/8 + Total I/O (queue=16)/16.
This is the consequence of the increased speeds: early SSDs are as fast as ordinary HDDs in this test and inferior to the server-oriented Fujitsu, but the X25-M leaves its opponents far behind.
By the way, the new SSD feels best at short queue depths, showing a performance jump.
The X25-M enjoys a sevenfold advantage over the closest opponent in terms of performance rating. HDDs just can’t deliver such performance.
As opposed to the previous load, the Web-Server pattern doesn’t have write operations, thus being ideal for SSDs. Samsung’s SSD is faster than the HDDs in this test while the X25-M delivers 4000 and more operations per second. Even if we imagine a RAID array with ideal scalability, it would take no fewer than eight SAS drives with a spindle rotation speed of 15,000rpm to overtake the X25-M.
The X25-M shows its preference for short queue depths again. There is only one unpleasant thing: the drive’s performance is somewhat lower at a queue depth of 32 requests than at zero queue depth.
The X25-M is eight times as fast as Samsung’s SSD, which is itself faster than any HDD in this test.
The new SSD from Intel is good in the Workstation pattern while Samsung’s SSD is slower than the hard disks. Note that while the HDDs are increasing performance along with the growth of the request queue, the SSDs are slowing down. This is especially conspicuous with the X25-M.
Our formula gives higher weights to the results achieved at small queue depths, so the X25-M enjoys a huge advantage over its opponents again.
When the test zone is limited to 32 gigabytes, the X25-M changes its behavior at small request queue depths. It begins to resemble HDDs then. At long queue depths its performance is higher than in the previous test, when it could use its full storage capacity.
The X25-M has a smaller advantage than in the previous test, yet it is still three times as fast as the closest of its opponents.
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.
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 asynchronous threads (one for reading and one for writing) when copying files.
This test produces too much data, so we will only discuss the NTFS results. You can use the links below to view the FAT32 results:
We’d like to make one remark about the very low results of the Fujitsu: all SAS drives deliver similar performance in this test. We seem to have some driver conflict here. As for our SSD, its ten-channel architecture has indeed made it comparable to the best of modern hard disk drives. Take note that the Samsung F1 disk with its very dense platters is only ahead in the ISO pattern, i.e. on very large files. As soon as the file size is reduced, the X25-M goes ahead. The smaller the files, the bigger advantage the SSD enjoys. And it is with small files that we work oftener in real life.
The X25-M is unrivalled at reading. As opposed to the hard disks, it doesn’t care much about the size of the processed files. Therefore it is far faster on small files. It is only on very large files that the Fujitsu gets close to it. It’s time to recall that our controller obviously limits the SSD’s capabilities. We guess the next time, under better test conditions, it will be absolutely superior, delivering speeds higher than 200MBps.
When copying within the same partition, the X25-M is also ahead in every file pattern, especially in Programs and Windows. The Samsung F1 is unable to catch up with the SSD even in the ISO pattern although gets very close to it.
The X25-M wins the Far Copy test, too.
PCMark tests are the closest to real-life applications.
PCMark04 benchmarks drives in four different modes: 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; File Copying measures the HDD performance when copying a set of files; the General Usage parameter reflects the disk activity in a number of popular applications. These four parameters are used to calculate the overall performance rating.
We ran each test ten times and averaged the results.
Of course, this test is not absolutely equivalent to a real boot-up procedure. Expressed in seconds, the results won’t differ so much because PCMark is a trace-based benchmark. So we can say that the X25-M shows thrice higher performance at system boot-up in comparison with its opponents. Even Samsung’s SSD with its low sequential speeds is ahead of the HDDs in this test thanks to excellent speed of random reading. And Intel’s SSD with its multi-channel architecture couldn’t but win this test!
The Application Loading test produces the same picture due to the same reasons. The X25-M enjoys an even bigger lead over the HDDs. It is five times as fast as them!
The copying test agrees with what we have seen in FC-Test. The Samsung SSD takes last place due to its low sequential speeds, but the X25-M has high sequential speeds thanks to its multi-channel architecture. Coupled with superb processing of small files, this makes Intel’s SSD an indisputable leader.
Data are read more than written at ordinary usage of a computer. Therefore the SSDs are in the lead here, too. The old model from Samsung is ahead of the HDDs whereas the Intel X25-M is absolutely unrivalled.
The X25-M has the best overall score, of course. Take note that it is three time as good as the older SSD, four times as good as the server-oriented 15,000rpm HDD, and five times as good as one of the fastest desktop HDDs available now!
PCMark05 is an updated version of the previous benchmark. Instead of File Copying, there is now a File Write trace. A new trace called Virus Scan is added. Its name is self-explanatory.
Again, we performed each test ten times and averaged the results.
These three tests produce the same picture as the same traces of the previous version of PCMark: the X25-M enjoys a huge lead over the others. The Samsung SSD is second best while the HDDs take last places. Frankly speaking, the difference in performance is awful. The new guy in town, the newly-released X25-M, leaves no chance to the oldies.
Well, HDDs do not give up without a fight. The Fujitsu wins first place in the Virus Scan trace, probably due to the larger cache and efficient algorithms of using it. We wonder if the SSD will be slower when benchmarked on a faster controller.
There is one load type at which HDDs are still faster than SSDs, though. We mean writing, of course. But we haven’t yet tested SSDs based on SLC memory that delivers higher write speeds. So far, this test is the last bastion of HDDs.
The Fujitsu has a higher overall score in this benchmark than Samsung’s SSD but the Intel X25-M is still almost two times as fast as the 15,000rpm HDD.
Ten-channel architecture of the controller’s interaction with flash memory chips, a good processor with effective algorithms, and buffer memory have been combined by Intel to produce the high-performance storage X25-M. Our bravos go to the developer who has managed to transform the ugly duckling of early SSDs into a swan. We must also keep it in mind that the X25-M employs multi-level-cell memory which is considerably cheaper than single-level-cell chips, being inferior to the latter in write speed and service life.
The speed characteristics of this 2.5-inch SSD are so superb that it will easily outperform RAID0 arrays built out of multiple disks, which is one of the very few available methods of boosting your disk subsystem performance. And the SSD needs less power and takes less space than a RAID. It seems that RAID arrays (except for those comprising lots of disks) will soon be relegated into the sector of redundant data storage or will be used when high speed must combine with high capacity.
A hard disk drive has but few advantages left over a solid-state drive: low cost of storage, high storage capacities, and somewhat higher speed of writing (as yet). The latter advantage may vanish quite soon because the upcoming SLC-based X25-E series is declared to deliver a write speed of 170MBps. HDDs with such a high write speed are not expected any time soon.
The SSD is yet behind the HDD in terms of capacities, even though the gap is getting smaller. The capacity of the SSD we have tested today is 80 gigabytes and Intel plans to introduce a 160GB model in Q4 of this year, but 2.5-inch hard disk drives of all the manufacturers have already reached a capacity of 320GB and are approaching 500GB. 3.5-inch HDDs are going to step from the 1TB milestone even further.
Well, the capacity is closely linked to the price factor. And it is from this aspect that SSDs are far inferior to HDDs. High speeds and low access time come at a high price. It is technically easy to increase the capacity of an SSD by using a bigger case and more memory chips, but the resulting device would be far too expensive. The 80GB X25-M comes at a price of $595. So, the price of each gigabyte of its storage space is about $7.5 whereas the price of one gigabyte of hard disk storage is far less than 1 dollar. On the other hand, SSDs have progressed in terms of pricing, too. The manufacturers obviously have some elbowroom determined by the manufacturing cost of the chips. We just have to wait for a real price war and reap the results.
So far, we can see a clear picture: rather small but very fast SSDs are going to be used in applications that require high performance of the disk subsystem whereas HDDs will be used for storing the bulk of data and will work as the main storage only when high speeds are not necessary.
There is only one thing we could not check out practically in our tests. And unknown as it is, this factor is the most alarming one. How long will the service life of an SSD be? How long will the X25-M work in an expensive notebook, a high-performance workstation or a server? How soon will those 10,000 rewrite operations guaranteed for modern MLC memory expire in practical applications, even considering the wear-leveling algorithms? How long will an SLC-based model with its guaranteed 100,000 rewrites work in a highly loaded server with lots of writes? These questions surely need more investigation.
Anyway, we’d like to end this review with a positive note. The Intel X25-M boasts such a quick access time that it is just incomparable to HDDs and many previous SSDs. It also delivers superb performance at random-address operations (at both reading and writing), a very high speed of sequential reading and a sufficiently high speed of sequential writing. It is a new star in the high-performance storage world. And we promise you to keep our eye on it.
