by Aleksey Meyev
03/23/2008 | 11:11 AM
Memory cards is something every one of us uses because digital cameras, PDAs, smart-phones are all equipped with one or even several such cards. You can see them in cell phones, players and car stereos, too. Memory cards are also used with ordinary PCs even though the small size of this storage doesn’t seem to be an important factor for them: the low power consumption and shock tolerance make such cards a good replacement to hard disks in industrial PCs, compact servers, notebooks. You may also want to use a memory card with your desktop PC because Windows Vista can utilize flash memory to accelerate disk access (thanks to the relatively low access time of such memory at reading) by means of ReadyBoost technology.
Memory cards have been improving their key characteristics, capacity and data-transfer rate, year by year. But besides the theoretical speed, there is the real one that depends not only on the card’s capabilities but also on the card-reader’s. The reader is built into the device’s electronics in a camera or PDA, but if you want to use the card with your desktop PC, you can choose the reader yourself. It would be wise to do so because cheap readers, based on slow chips, cannot reveal the potential of a modern card fully.
Before proceeding to the devices I am going to test, I will tell you about the SDHC and MMCplus formats that have come to replace SD and MMC, respectively. I’ll mention these formats below in connection with some models of readers, but some users have a rather vague notion of the difference of these formats from their predecessors.
The story of these form-factors (they are thicker than the full-size versions but it is the only difference) began in 1997 with the introduction of the MultiMediaCard by Siemens and SanDisk. As opposed to the older Compact Flash format, which mostly employed NOR flash memory chips, the new standard was based on more compact NAND chips. The resulting card was very small at 24x32x1.4mm. Later on, the Secure Digital standard was developed on that basis. An SD card was 0.3mm thicker but offered data security features (not very demanded by the users, though) and a somewhat higher speed due to the higher interface frequency, 25 or 50MHz as opposed to the MMC’s 20MHz. The further progress was logical enough: the cards were getting larger in terms of capacity, yet retained their compact size. They eventually won a share of the market from the larger Compact Flash format. Their speed has always been somewhat lower in comparison with Compact Flash, but the difference is not too big in practical applications. Reduced-size versions of these cards appeared under the pressure from the manufacturers of compact devices (cell phones in the first place): RS-MMC, SecureMMC, mini-SD, and microSD. RS-MMC is just a twice shorter MMC while the others have differently positioned connectors and can work in SD/MMC-supporting readers by means of an appropriate adapter.
After a while, both formats approached a point when the steadily accumulating problems called for an introduction of serious changes into the design of the cards. For MMC the main problem was the data-transfer speed across the interface, which was too low. For SD, there was a fundamental capacity limit of 2GB resulting from the addressing mechanism originally implemented in the standard. That’s why the attempts to introduce 4GB SD cards were not a success as most devices just wouldn’t work with them. Anyway, both standards solved their problems one way or another. There appeared MMC 4.1, also known as MMCplus, in which the bus frequency was increased to 52MHz while the bus itself was enlarged with two additional contacts. This helped maintain backward compatibility, even though at low speeds (if you insert an MMCplus card into an old card-reader, the latter will work with it as with an ordinary MMC), and achieve excellent speeds, up to 290x, in new card-readers (to remind you, 1x equals 150KB/s as in CD drives). The SD standard developed to version 2.0, better known as Secure Digital High Capacity, in which byte-based addressing was replaced with sector-based addressing. The maximum capacity of a card thus increased to 2TB. Unfortunately, the introduction of the new addressing type doesn’t allow SDHC cards to be read by old, SD-supporting readers.
Summing it up, MMCplus cards can work in any MMC-compatible device but can only show the full speed in new card-readers specially designed for MMCplus. SDHC cards do not work at all in old SD-compatible readers. You should make sure your card-reader (PDA, camera, etc) supports SDHC before purchasing such a card.
Finally, the SD 2.0 specification introduced speed class ratings. Each SDHC card can be class 2, 4 or 6. The number means the minimum write speed the empty card provides. These speeds are 2, 4 and 6MB/s, but the manufacturers can install faster chips for higher write speeds. The speed just cannot be lower than the declared class.
You can go to the MMC and SD websites for more information about these flash card standards. Now that we’re done with the theory, it’s time to get to practice. I mean, to the card-readers to be tested.
Every model covered in this review, both internal and external, is connected via the USB interface, so the difference between them is in the design of the case, the connector at the end of the cord, and in the controller chip. I can tell you right away that each model comes with the same accessories that include a USB cord and a disc with drivers for old OSes (card-readers don’t require special drivers from Windows 2000).
As for the controller, you don’t even have to take the device apart to find what model it is: every controller has a specific code, two four-character strings. The Vid string denotes the manufacturing firm, and the Pid string denotes the model number. These strings can be read with special programs, but I prefer the way that doesn’t require additional tools.
Just connect the card-reader (with any card inserted in it) to a PC running Windows and open the Device Manager. Find the USB Mass Storage Device among the USB devices. If no other external storage is connected to the PC, there is only one such string in the list. Then you open the Properties window, switch to the Details tab, and select Hardware IDs in the drop-down menu.
I will describe external models first.
That’s a nice-looking square box with two pairs of card slots in its two sides. A third side offers a mini-USB connector. The manufacturer says the Mega Steno AM300 supports CF, SD/MMC (all versions up to MMCplus and SDHC), MS and SM (xD). This doesn’t seem much in comparison with 52-in-1 or something models, yet these four slots allow working with every existing card type.
As for the numbers in the model names of many card-readers, most of them are just a marketing trick. Manufacturers claiming that their devices support dozens of card formats just puff up the list deliberately, including cards of the same format but marketed under different brands. For example, Transcend, Kingston and SanDisk turn out Compact Flash cards under their respective brands and some card-reader maker just calls them different formats to write a bigger number in the device’s specs. So, the modesty of Apacer should be viewed as the company’s honesty with respect to the customer.
According to the Vid and Pid combination, this card-reader is based on an Alcor Micro AU6362 chip with firmware 01.29.
A typical representative of its product category, this card-reader has a flat translucent plastic case you can see its internals through. One longer side offers four standard card slots. A shorter side provides a mini-USB connector. There is an Alcor Micro AU6362 chip with firmware 01.26 inside. The long list of supported formats is built up artificially in the way I’ve described above: for example, besides Secure Digital format it includes Secure Digital Kingmax (ordinary SD cards manufactured by Kingmax) and Secure Digital Extreme (ordinary SD cards manufactured by SanDisk) that don’t require any special support.
This card-reader has a black plastic case and a rather flimsy cover above its connectors. Otherwise it is alike to the previous model having four universal card slots and an Alcor Micro AU6362 controller with firmware 01.26.
In this model the case remained transparent, has no additional cover. The only thing new is the color of the casing. As you may have already guessed, there is the familiar Alcor Micro AU6362 chip inside with 01.26 firmware.
This one looks completely different: the card slots are placed on the two longer sides of the case. One shorter side offers three output ports of the integrated USB hub (unfortunately, you cannot plug devices wider than the USB connector into the neighboring ports), and the other shorter side provides an input mini-USB port and a power adapter connector. The power adapter is not included into the box. The hub can work without it unless you attach USB devices consuming over 500mA in total to it. The card-reader is based on an Alcor Micro AU6362 controller with firmware 01.26.
This card-reader looks very much alike to the Apacer AM300 and even more imposing due to its cute metallic surface but the assembly quality is depressing. The internals are installed awry making it difficult to insert a card into the SD slot. I was also confused at unexplainable problems with cables: the device would not be identified on its native cord as well as many other cables. It only worked with certain USB cables I had. Inside it I found a NeoDio ND3260 chip with firmware 02.00. This rarity absolutely refused to work with SDHC cards.
As opposed to most other models in this review, this reader is very compact but only supports MMC and SD formats and does not support SDHC. It looks like a USB flash drive that opens to reveal the USB connector. It is pretty, original, small and handy. The reader is based on an Alcor Micro AU6332 chip with firmware 1.13. We tested this model some time ago but with different cards.
This model is rather new in SanDisk’s product range and one of the fastest among USB-interfaced models. It lacks an SM card slot but supports SDHC. The case with sharp angles differs from other models, yet it is nothing fundamentally new: it’s got card slots in one butt-end and a mini-USB connector in the other. Unfortunately, it is impossible to find out the chip model in SanDisk’s devices.
As opposed to the previous model, this one has four card slots and supports all formats including SDHC. It has a remarkable design: you can connect the card-reader with an ordinary cord using the mini-USB connector on its butt-end or you can install it onto a neat stand that has a USB plug inside. Another distinguishing feature of this card-reader is the button on its surface that can be pressed to copy files to the PC (you need to install special software for that feature to work).
Yet another example of the minimalistic trend, this device is but slightly larger than SDHC and MMC cards it can work with. One of its butt-ends is entirely occupied by a card slot, and the other offers a full-size USB connector. So, you can just plug this card-reader into a USB port of your PC. This model is supplied with SanDisk cards but can also be seen selling in retail.
This is yet another rather old model from SanDisk differing from the other card-readers in design. The single slot on the butt-end of the silvery plastic case supports MMC/SD (alas, without support for SDHC) and MS cards. Unfortunately, this reader can only work with full-size MS cards while their reduced-size versions go so deep into the slot that you may not pull them out afterwards. Unlike the other card-readers, this one connects to the PC by means of a full-size B-type USB connector. Of course, it offers the quick copy button we’ve seen on the SanDisk Image Mate 12 in 1.
This must be the biggest external card-reader in this review. It resembles the Eurodata 19-in-1 as it has a black case, too, with a lid covering four connectors. It is connected via a full-size USB connector on a short cord located at the back of the case. This reader is based on a Genesys Logic GL819 controller with firmware 93.17.
Yet another square device with four connectors on two side panel, this reader’s case is painted a bright orange. It connects to the PC by means of a mini-USB plug. The reader is based on a Genesys LogicGL819 controller with firmware 93.21. It doesn’t support SDHC format.
The second model from Transcend is smaller but it supports only SD/SDHC, MMC and MS formats. This model would be the smallest in this review if it were not for the SanDisk MicroMate. Based on an Alcor Micro AU6333 controller with firmware 01.20, it connects to the PC by means of a mini-USB plug.
This card-reader looks effective due to its mirror-like surface but doesn’t differ much from the above-described all-in-ones overall. It’s got four connectors on one side panel and a mini-USB on the opposite side. It is based on an ICSI IC1210 chip with firmware 01.9. Unfortunately, it doesn’t support SDHC.
55 in 1 plus
These no-name products differ from each other with the shape of the case. The CR 702A also has a full-size USB connector on a short cord rather than a mini-USB plug. Supporting all card formats except for SDHC, these devices are all based on a Genesys Logic GL819 controller with firmware 93.17.
This no-name device resembles the above-described Eurodata, which was painted black. Otherwise, it is exactly the same, being based on an Alcor Micro AU6362 controller with firmware 01.26.
I will also test a few internal models that are to be installed into a 3.5” bay of your system case. This is a handy solution if you often work with memory cards and don’t have much space on your desk.
The internal reader from Gembird looks like a typical representative of its class. It’s got a plastic case and four connectors on the front panel. Its cord is connected to the mainboard’s onboard USB headers. The device is based on a Micro AU6362 controller with firmware 01.26 and does not support SDHC. There is a place for an ordinary USB connector on the front panel, but the connector itself is missing.
This reader has a different color of the case than the previous model and a USB connector on the front panel, but differs greatly on the inside. The Microsonic CR 90 is based on a new Alcor Micro AU6377 controller with firmware 01.00. As opposed to most other internal readers, it supports SDHC.
This model has a lot of additional connectors on the front panel. Besides the reader’s slots, there are USB, FireWire and two audio connectors that can be attached to the mainboard’s internal connectors. This reader is based on a Genesys Logic GL819 controller with firmware 93.17. Trying to meet every variant of the USB port pin-out on the mainboard, the manufacturer put a bunch of individual pins instead of a monolithic connector on the end of the cord.
These three models differ in the position of an additional USB connector on the front panel and the design of the connector on the cord (monolithic or not). In fact, this is one and the same model based on an Alcor Micro AU6362 chip with firmware 01.26.
And finally, there are these queer devices from China. The 3.5” one lacks a top panel and connects by means of an ordinary external USB connector. The 5.25” device offers a bunch of cables, one cable for each interface on its front panel. The oddest thing is that the cables end in ordinary USB and FireWire connectors. The user is probably supposed to put the cables out through the back panel of the system case and plug them into the mainboard’s external connectors. These readers are based on an Alcor Micro AU6362 controller with firmware 01.26.
For the sake of convenience, the information about the controllers is listed in the following table:
We used the following software:
Using two versions of FC-Test allows to compare the results with those we got in our earlier tests while transitioning to the newer version of the testing program.
The following PC configuration was used for the test:
I used the generic OS drivers for the test. The memory cards were formatted in Windows XP as one FAT32 partition with the default cluster size. The card-readers were connected to the mainboard’s connectors with the standard cables, without USB hubs or extension cords.
The following cards, which are among the fastest available now, were used for the test:
We did not test SM and xD cards due to their low popularity and a total lack of progress in their characteristics.
The speed of a card-reader depends entirely on the controller installed in it, so I will compare not the reader models, but the controllers – you can look up what reader a specific controller belongs to in the table above. SanDisk seems to prefer to change the Vid and Pid strings for the chips of its card-readers, so I will call these chips by the name of the device they are installed in.
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 we will see the dependence of the reader’s linear read/write speed on the size of the data block.
The results are divided in two groups for better readability. The card-readers based on the AU6263 controller with firmware 01.26 and 01.29 are the worst at reading, reaching a speed of 3MB/s only (the card can yield data at a much higher speed than that). Interestingly, the same controller but with firmware 01.00 is much faster at nearly 20MB/s. The IC1210 and ND3260 controllers are very poor at sequential reading too, their maximum speeds being lower than 10MB/s. That’s too low a speed for a modern flash card. SanDisk’s card-readers are traditionally good. The readers based on the Genesys Logic controllers are good, too. Alcor Micro’s new AU6377 controller is first, achieving a speed of 30MB/s on large data blocks.
It’s the same as in the writing test: the AU6377 is in the lead, and there are two groups based on speed. The AU6362 with firmware 01.26 and 01.29 is on the losing side. Note the performance of the SanDisk Extreme USB: it is far faster than the other readers on small data blocks (from 2 to 8KB).
The average read/write response time is measured in a 10-minute test to read/write 512-byte data blocks with a request queue of 1. The reader has to process over 60,000 requests, so we get a sustained response time as the result. The numbers are sorted by read response.
The readers have similar results here. The difference between the worst and best result is a mere 3%. The read response time is very small especially in comparison with other types of storage (for example, hard disk drives). The AU6377 is again the best controller. It is followed by the AU6362 (with firmware 01.00 while the other firmware versions are slower), the GBL819 with both firmware versions, and SanDisk Extreme USB.
I’ll evaluate the speed of random read and write operations by measuring the time the readers take to perform them. The data block size is varied from 0.5 to 8192KB in this test.
I don’t build a diagram for this test – the table is enough. It’s clear that the controllers line up in the order of their results in the sequential reading test. The AU6377 is first, and most other controllers follow it in a dense group. The AU6263 with the “slow” firmware versions is last, and the ND3260 and IC1210 are rather slow, too.
There are no significant changes when we change reading for writing.
We continue checking out external storage media for their compliance with Windows Vista’s ReadyBoost technology using IOMeter. To remind you, the point of that technology is in using a flash drive or a flash card as an additional cache with a data access time lower than that of the hard disk. To qualify for this application, the external disk must meet certain performance requirements and have a capacity of 256MB and higher. The OS itself benchmarks the speed of the attached device and proposes that it be used for ReadyBoost. What are the requirements? The flash disk must ensure a data-transfer rate of 2.5MB/s and higher when reading random 4KB data blocks and a data-transfer rate of 1.75MB/s and higher when writing 512KB data blocks. So, we measure these two speeds in a 10-minute test. For better readability, the red vertical line marks the ReadyBoost-compliant speed in the diagrams:
Only the slowest readers – the AU6332 and ND3250 – cannot pass this test when working with the fast CF card. The IC1210 was very close to a failure as well, yet it passed the test. The other readers coped with the task easily.
The requirement of 1.75MB/s proved to be easy for each reader. This test is more demanding about the card’s speed on small data blocks rather than to the controller’s capabilities. To remind you, a SanDisk Ultra II card couldn’t pass this test in any reader.
FC-Test will help us examine the readers under real-life conditions. The test writes and reads a few file-sets and measures the time it takes to perform each operation. This allows to calculate the resulting speed and see how it depends on the number and size of the processed files. We use three file-sets that differ in the size (1, 10 and 100MB) and number (1, 10, and 100) of files included. Practice suggests that a 100MB file is large enough to reveal the maximum performance of a flash-based storage media and using a larger file doesn’t affect the results much
I will only publish the diagrams for the reading and writing of ten 10MB files to make the review shorter. You can view the other diagrams clicking on the links below.
Other read diagrams:
There are no surprises. The controllers perform up to the reputation they have earned in the previous tests. The AU6377 is in the lead, being 30% ahead of the closest pursuer, which is the SanDisk Extreme USB. The two AU6362 controllers are the slowest of all. The IC1210 and ND3260 are rather slow, too.
Other write diagrams:
The SanDisk Extreme USB goes ahead in this test, outperforming the AU6377. The four slowest readers are the same as in the previous test.
I will now test the cards in the newer version of FC-Test using the same methodology.
Other read diagrams:
The newer version of the test outlines the tendency better: the AU6377 enjoys a huge lead over the others. Next go five more controllers with similar results, and there are four very slow chips.
Other write diagrams:
The SanDisk Extreme USB goes ahead, pushing the AU6377 down to second place. The other chips do not show anything exceptional.
So, it is all quite clear about this part of my tests. The card-readers based on the new AU6377 controller deliver the highest read speed while the SanDisk Extreme offers the highest speed of writing. The readers based on the IC1210, ND3260 and, especially, AU6362 controllers (the latter with firmware version 01.26 and 01.29) are slow. Even 45x card are going to surpass the capabilities of these readers.
I should note that the fastest Compact Flash cards have reached such high speeds that non-specialized card-readers cannot match them. The best models in this test session have but half of the speed the card could deliver. Most controllers supports speeds of 100-150x, but not 266-300x.
Now let’s check out Secure Digital format.
Like with Compact Flash cards, the results and, accordingly, the controllers can be divided into two groups. One group includes those readers that deliver a read speed of less than 10MB/s. This group includes the AU6332 controller with firmware 01.26 and 01.29, the IC1210 and the ND3260 controller. The SanDisk USB provides a somewhat higher speed, but doesn’t reach 10MB/s, either. The other controllers fall into the second group which in its turn splits into two subgroups with absolutely identical results. The highest speed on 64MB and smaller data blocks are delivered by Alcor Micro’s AU6263 (with firmware 01.00), AU6332, AU6333 and AU6377 and by the SanDisk MicroMate.
The overall picture doesn’t change much at writing. The SanDisk USB is again the best among the losers, almost reaching a speed of 10MB/s. The leaders go in two groups again with a difference of 1MB/s on small data blocks.
The ND3260 acts up in the response time test, having much worse results than the other models. Take note that the average write access time of Secure Digital cards is one third lower than with Compact Flash cards, and the average read access time is, on the contrary, one third higher. The formal winner of this test is the GL819 with firmware 93.17.
The five slowest controllers from the sequential reading test are the slowest here, too. They are 2-4 times slower than the other chips that deliver so similar results that it’s impossible to pinpoint a winner.
It’s no different in the writing test: we’ve got the same five slowest readers and the same leaders with very similar results.
Now we’ll check out the readers with the SD card for their compliance with the ReadyBoost technology. The point of that technology is in using a flash drive or a flash card as an additional cache with a data access time lower than that of the hard disk. To qualify for this application, the external disk must meet certain performance requirements and have a capacity of 256MB and higher. The OS itself benchmarks the speed of the attached device and proposes that it be used for ReadyBoost. What are the requirements? The flash disk must ensure a data-transfer rate of 2.5MB/s and higher when reading random 4KB data blocks and a data-transfer rate of 1.75MB/s and higher when writing 512KB data blocks. So, we measure these two speeds in a 10-minute test. For better readability, the red vertical line marks the ReadyBoost-compliant speed in the diagrams:
The IC1210, ND3260, and the AU6362 chips with the two slow firmware versions are again the slowest. The SanDisk USB passes the test but barely. The other controllers deliver similar results.
Like with the Compact Flash format, the chips all pass the random-address writing test. The slowest controller delivers two times the required speed.
Other read diagrams:
Just as you’ve seen in the previous tests, there are five slow controllers while the others deliver similar performance.
Other write diagrams:
No changes in the writing test.
Other read diagrams:
Other write diagrams:
The diagrams speak for themselves: there are several controllers that cope superbly with SD cards, and there are five slow controllers. The SanDisk USB is reasonably slower than the leaders, but the AU6362 (with “slow” firmware), IC1210 and ND3260 are obviously slow.
Summing up this section, I can tell you that most controllers coped well with Secure Digital format, delivering similar speeds in all the tests. On the losing side are the AU6362 (with firmware 01.26 and 01.29), IC1210 and ND3260. The SanDisk USB is not fast enough for modern SD cards, but its results are not so depressing.
Perhaps such a high level of performance is due to the fact that there are no SD cards that would be as fast as Compact Flash. On the other hand, SD cards are not slow. Their speed is enough for most modern devices, excepting modern professional-level reflex-lens cameras (Compact Flash is dominating in this field anyway). In most devices it is the integrated card-reader rather than the card that limits the data-transfer speed. As for ReadyBoost technology, modern SD cards deliver sufficient speed unless the card-reader is based on a slow controller. The main reason why memory cards of the first version of SD format are leaving the market is the limit on the maximum capacity of the card, only 2GB. Not much by today’s standards. That’s why SD is being replaced not with some new format, but with its logical follower Secure Digital High Capacity.
So, let’s see what we have with SDHC cards that are free from the capacity limitation problem. I should note that many card-readers are still unable to work with this format and do not recognize the inserted card. That’s why there are fewer devices in the next section of the review.
Four out of the six controllers passed the test at the same speed. The AU6377 has a twice lower speed. The AU6362 couldn’t cope with small data blocks altogether and had a very low speed on large data blocks.
The AU6362 has zero writing speed but didn’t issue an error. Running a little ahead, I should tell you that this chip passed the tests one way or another, so it is not completely incompatible with this card format. Among the other card-readers, the AU6377 chip, which has been very fast in the Compact Flash tests, is a disappointment. The other four controllers are similar, but the AU6333 and SanDisk MicroMate have somewhat higher speeds on 8KB and 16KB data blocks.
The first thing to catch your eye in the diagram is the very low write access time of the AU6377 controller. This must be some flaw in the test program because the result is beyond the capabilities of flash memory. The AU6362 with firmware 01.29 is thus the leader of this test. Comparing this diagram with the diagram for the same test with the SD card, you can see that the write access times are somewhat higher.
The AU6362 seems to have problems in this test judging by the time it took to perform the random-address read operations. The AU6377 differs from the others, spending two times more time to process blocks of every size than its opponents do.
The AU6377 is, on the contrary, ahead of everyone, having a much better result. It is worse than the others, except for the AU6362 (which had problems again), on large data blocks, but I’m more interested in small blocks as it is in such blocks that data are written to the card in real applications.
Alas, the AU6362 still has problems – it passes the test but its result is awful. The other controllers overtake the 2.5MB/s barrier although the AU6377 is but slightly above that mark.
Every controller copes with the writing test easily, excepting for the AU6362. It resembles the test of the SD card.
Other read diagrams:
Five out of the six readers cope with reading well enough. The AU6362 even tries to show some results, yet it is awfully slow in comparison with the other chips.
Other write diagrams:
The controllers split up at writing: the AU6333, SanDisk MicroMate and SanDisk Extreme are ahead while the AU6377 and SanDisk ImageMate are behind the leaders by 1.5-2GB/s.
Other read diagrams:
Other write diagrams:
The AU6377 has considerably lower results both at reading and writing in the newer version of the test. The 6362 chip works with SDHC cards but does it too slow. The other chips are equally good with this card format.
So, SDHC cards are no faster than SD. No wonder as the addressing mechanism is the only difference between them – we were not promised any performance benefits. You may have noticed that the speed was generally higher than 6MB/s as guaranteed by Class 6. As for compatibility, the controllers that are declared to support SDHC do work with this format except for the AU6362 with firmware 01.29 that obviously had problems with it.
Now it’s time to discuss MMCplus format that theoretically provides a higher speed than the above-discussed SD and SDHC (290x as opposed to 150x) on the condition that the new format is supported by the card-reader. Otherwise the card will work in MMC compatibility mode, i.e. at a reduced speed.
I guess the diagrams make it clear which controllers support the new standard and which of them work in the compatibility mode only. The AU6362 (with firmware 01.26 and 01.29), IC1210, ND3260 and SanDisk USB are in the slow group. As opposed to the previous tests, the controllers from the fast group don’t have similar graphs. The GL819 with firmware 93.21 is the best on large data blocks (by the way, it is the worst in the fast group with firmware 93.17). The AU6332 outperforms it on small data blocks, though. The AU6333 has a highs speed as well, but slows down heavily on 16KB blocks.
We’ve got the same five slow controllers again. It’s hard to find the winner because the fast controllers have similar results. The AU6332 differs with its fastest growth of speed along with the data block size. The AU6333 is suddenly slow on 1024KB data blocks.
The write response time varies by little among the controllers. The read response varies more. The same five controllers are on the losing side again, the ND3260 being three times worse than the leader GL819 with firmware 93.21. Note that the write access time is two times that of SD and SDHC cards if used in the same card-readers.
The five losers do not change, but this time it is the AU6362 with firmware 01.29 that is the best among them. The leaders are the GL819 (with firmware 93.21), AU6332 and AU6333. Note that the gap isn’t large and mostly occurs on large data blocks.
There are no definite leaders here, and the slow group isn’t far slower than the leaders.
As you could expect from the sequential read diagrams, all the controllers, except for the slow five, overtake the 2.5MB/s limit. Take note that the speeds are somewhat higher than with SD and SDHC cards.
There are no surprises at writing. There are only three controllers that don’t pass the test: IC1210, AU6362 (with firmware 01.26) and SanDisk USB.
Other read diagrams:
The slow controllers are slow in FC-Test, too. First place goes to the GL819 with firmware 93.21. It is followed by the AU6332 and AU6333 that have had good results in IOMeter.
Other write diagrams:
As opposed to reading, every fast drive copes with writing well enough. The “slow” chips are again very slow.
Other read diagrams:
Other write diagrams:
The second version of the test doesn’t have anything new to tell us except that the AU6333 has a surprisingly low result (in comparison with the leaders).
So, MMC provides almost the same speed as SD. But as opposed to SDHC, this format is supported by all card-readers. MMC is somewhat faster at reading but slower at writing. The leader of this section is the GL819 controller with firmware 93.21. The winners of the previous sections performed fast enough, too.
Finally, I’ll discuss the fastest variety of Memory Stick format – Memory Stick Pro Duo.
The chips form two distinct groups again. One group includes those chips that have problems with support of this version of the standard, which shows up as the reduction of the sequential read speed on 64KB data blocks. The other chips, having similar speeds overall, accelerate on 64KB data blocks. The GL819 with firmware 93.21 is the leader on this test. The AU6333 has a very good result, too.
It’s overall the same as at reading but the AU6362 with firmware 01.29 differs from the others in the slow group as it doesn’t slow down on 64KB data blocks. Anyway, its speed of 5MB/s is lower in comparison with the fast group that deliver a write speed of 13MB/s on 64KB data blocks. Note that the fast chips behave differently with small data blocks, the AU6362 with firmware 01.00 and the AU6333 being best among them.
The write response time is disappointing in comparison with the other memory card formats. We’ve got the same losers, excepting the AU6362 with firmware 01.26 that performs better here.
The AU6362 with firmware 01.26 joins the group of losers again in the random-read test while the SanDisk USB leaves it. There is no definite leader as the other readers have similar results.
All the chips are very slow at processing small data blocks. It’s impossible to identify a definite leader or loser among them.
According to this test, flash cards of this format do not comply with ReadyBoost technology. None of the chips can deliver the required speed. I guess it’s the problem of Memory Stick format at large rather than of the specific chips.
Again, the card cannot reach the desired speed in any of the card-readers. The speeds are too low in comparison with those of the other card types.
Other read diagrams:
The SanDisk USB leaves the group of outsiders to overtake the fast chips in this test.
Other write diagrams:
The writing test produces a more interesting picture that shows both leaders and losers. Interestingly, the lowest speed comes from the AU6377 and the AU6362 with firmware 01.00 that have not been spotted among the outsiders before. The best result is provided by the GL819 with firmware 93.21 and by SanDisk’s ImageMate and Extreme USB.
Other read diagrams:
Other write diagrams:
The read diagrams are similar to what we’ve seen in the first version of the test but there are differences at writing. The group of losers that has got new members above is now also joined by the AU6333 whereas the SanDisk ImageMate has left its place in the top three to the GL819 with firmware 93.17.
Summing up this section of the tests, I can’t but regret the low speed of writing shown by every chip. It is not the problem of the specific chips, though. It is the problem of this memory card format. The low speed makes such cards incompatible with Vista ReadyBoost technology, by the way. The Extreme USB and the GL819 with firmware 93.21 can be named the leaders of the Memory Stick tests.
Now I would like to return to the specific card-reader models and how the test results apply to them. First of all, most no-name products are based on slow AU6362 chips with firmware 01.26 – and most of them are internal models. There’s no talking about high speeds here because the functionality of such card-readers is just enough to read and write the card. The Apacer MegaSteno (AU6362 with firmware 01.29) is the main disappointment as it never left the group of outsiders throughout the tests. The products with the Gembird brand are poor, too.
The Microsonic CR 90 (AU6377) is quite a good internal card-reader. It is one of the few such devices to support SDHC and its speed characteristics are good, especially for its low price. Moreover, it is the only card-reader to deliver a higher speed with Compact Flash than the much more expensive SanDisk Extreme USB. Hopefully, this controller will be installed in both internal and external card-readers.
SanDisk’s products performed well as always except for the old SanDisk USB model. These products are expensive, but enjoy deserved popularity. I am especially fond of the cute MicroMate. It is going to be appreciated by people who need support of SD/SDHC cards only.
The products from Transcend are good, too. These are the M1 (on the AU6332 controller) and the TS-RD13R (on the GL829 with firmware 93.21). The latter is out of production now, but can still be seen in shops. It has been replaced with a similar model that supports SDHC, probably on an Alcor Micro chip.
The compact and appealing Pretec e-Disk II is not a new model, but its speed characteristics are good. Unfortunately, it doesn’t support SDHC cards.
Comparing the card-readers’ capabilities with those of the cards proper, I should confess that most readers are no match for the fastest of cards. So if you purchase a 200x or faster card, think about buying a reader like the SanDisk Extreme, preferably with a FireWire interface.
Now that the Secure Digital standard has reached its end and is being replaced with SDHC, which is free from the 2GB limitation of maximum capacity, the card-readers don’t have problems with it. The speed of both cards and card-readers is high enough, and your only problem may be to find a device capable of reading new SDHC cards, which are incompatible with older readers. Both formats are identical in terms of speed while the rating system of classes is only necessary to show the user the device’s requirements to the minimum write speed and, accordingly, to simplify the purchase process.
MMC proves to be a viable format that is successfully supported by today’s card-readers. Unfortunately, it is not very popular due to unsuccessful competition with SD and SDHC.
The Memory Stick standard (in all its versions) seems to be destined to a slow decline. First of all, Memory Stick cards are more sensitive to the type of transferred data and do not comply with ReadyBoost technology as the result. Moreover, these cards show low speed irrespective of the card-reader.
The last thing I want to tell you is that if you want a fast card-reader, you should buy a tested model or check out the version of the controller installed in the device you are interested in. It’s easy to learn the Vid and Pid parameters of a device if you can connect it to a PC. After that, you can predict the model’s characteristics which depend directly on the controller installed in it.