by Vasily Melnik
08/10/2004 | 02:17 PM
Comparing the so recent, but so forever gone era of film-based photography with the current digital photography, one may come to a varying opinion: from “nothing changed really” to “everything is now different”. Beyond doubt, digital technologies brought the long-sought-for speed of processing the image and getting the result, but still they made it no easier to become a good photographer or a professional in this field! The photographer used to know a bit of chemistry, but now he/she has to have certain designing skills and know something about computers and digital image processing.
Digital media replacing film freed us from many problems, but, as usual, raised new ones, peculiar to digital technologies. This review is not about those problems, though. I am going to talk about the media: the choice of film/paper used to be widely discussed in the photographer community. Nowadays, however, the choice of the memory card has come to the fore. Although this choice doesn’t affect the quality of snapshots directly, many aspects of photographing do depend on it.
But what and how and to what extent depends on the choice of the memory card? What to choose – a fast and more expensive card or a slow and cheaper one? Can you be sure that you’ll get exactly the same performance as measured by various benchmarks? Do memory cards of the so-called “professional series” have a higher performance? These are the questions I’ll be trying to answer in this review.
Communicating with photographers, I found that many of them purchased a “fast” memory card without knowing exactly why it was better at everyday work. Their reasoning was usually trivial, “It is just faster!” Running a little ahead, I should say that in most cases – and I will try to prove my point – the performance of the flash card is not the main factor, contributing to the speed of a particular digital camera. Moreover, even a professional camera may not feel any difference between a slow and fast card – and that’s not necessarily bad!
So why, then, does it make sense to purchase the highest-performance flash media? It may be reasonable in those rare cases when the fast card does give you certain advantages or – something else. For example, a high-capacity memory card is usually a considerable money investment, which you may want to secure somehow. Being limited-life devices (and due to some other reasons), flash cards fail sometimes. It is in this case that you realize that “professional” media series, although cost more, come usually with a long-term – sometimes lifetime – warranty. Of course, such a flash card is going to be the most secure investment of your money.
Anyway, many reviews and tests of flash cards limit themselves with the performance issues only. That’s understandable – one and the same memory card model may vary greatly in price and warranty terms in different world regions, so this information cannot be taken as a comparison basis. But we find ourselves on an insecure ground here: benchmarking the performance of a flash card is no trivial task, with numerous pitfalls.
Making your choice of the fastest memory card, basing on the results of a test session, is only reasonable if you are absolutely sure that you are purchasing a card from the same series as the one used in the tests. Considering that memory cards of different series, made at different factories and from different parts, differ very slightly (a casual user may not find any difference at all), their manufacturers are free to change everything in the device, including the supplier of the controller or the memory chips. Flash cards are produced in huge batches, so the manufacturers do change their supplier as well as the chip type from time to time, or simply use those parts that they now have at their storehouse. For example, fast and expensive flash cards usually use the so-called SLC (Single Level Cell) memory chips, while their slower and cheaper alternative are MLC chips (Multi Level Cell). As you understand, flash cards on chips of these two types don’t practically differ externally.
However, it is still possible to identify a majority of memory cards. A CompactFlash card usually carries a serial number, which can be displayed at several places, on the card’s butt-end as well as sides (the so-called edge stamp), or on the backside of its cover.
For you to be sure when shopping, I post the identification information about the cards I use in my tests. Each manufacturer has its own identification nomenclature and they are not very willing to share this information with outsiders. So, I give no clues as to deciphering the cards’ IDs – you may find certain regularities by yourself.
So, I’m about to check out the performance of sixteen memory cards of CompactFlash format from different manufacturers. I’ve got samples from “professional” as well as “ordinary” series. Thus, I will see if expensive cards provide a higher performance and to what extent. We will also check out how this performance depends on the camera or the card-reader model.
There were three digital cameras in my tests, two professional and one for “advanced amateurs”, and I used fast serial shooting as the most illustrative mode for determining the card’s performance. I also selected the maximum quality settings (RAW format if possible); the cameras were controlled manually; I used an exposure of less than 1/250s and a white balance preset, appropriate to the lighting. Installing the cameras on a tripod, I was shooting one and the same scene, focusing manually. The lighting didn’t change during the tests, so I got practically same-size files. During the tests, I measured the time interval between my pressing the “shutter release” button and the camera’s reader activity LED stopping to blink.
I also measured the performance of the flash cards with two fastest card-reader models (FireWire and USB 2.0 interfaces), using the same set of benchmarks: the FC Test with “1x400” and “400x1” patterns (i.e. writing and reading a single 400MB file and 400 files, 1MB each) and the Disk Benchmark plug-in from AIDA32. Plugging the card into the reader, I formatted it with the OS’s tools (FAT32, default cluster size).
Unfortunately, I couldn’t get any cards from Lexar Media – this manufacturer is now changing its model series and professional cards are being transferred from the “40X” series to “80X”. I don’t have SanDisk Ultra II and IBM/Hitachi MicroDrive, either. Otherwise, I tried to embrace as many available products as I could.
I decided to present my comments to the results in two parts: the first deals with the devices I tested the memory cards with, and the second part deals with the products of each flash card manufacturer and their results.
This device formally belongs to the class of reflex cameras with interchangeable lenses. Informally, this is a very popular model of a digital mirror-based camera now available in the market, and professionals and amateurs all like to use it. With a resolution of 6 megapixel and a 22.7x15.1mm sensor, the camera provides an excellent image quality under any conditions and for any photo tasks.
The Canon EOS 10D has an integrated frame buffer that allows shooting at a speed of three frames per second, up to nine frames in a series. After the shooting is complete or the maximum series size is exceeded, the buffer’s contents are written to the memory card and the shooting is stopped until the buffer has enough storage space for at least one frame.
You can see that the difference between the tested memory cards is rather small when you use them in the Canon EOS 10D camera – here, the gap between the fastest card (the Delkin Devices 1GB, with a result of 35 seconds) and the slowest one (the Pretec Platinum 512MB, 48 seconds) is only 13 seconds. That’s the maximum speed advantage you may obtain by replacing a slower card with a faster one – it may be even less! Moreover, you will only feel this advantage at reportage or sports shooting where making several shots in a series is the norm. Considering the rather low shooting speed of the Canon EOS 10D in this mode, it is unlikely to be used for such photo-work – especially because Canon, and other companies too, have cameras specifically targeted at press reporters and sports photographers. So, here’s my verdict:
This is the only camera today that complies with the open standard “Four Thirds” proposed by Olympus, Kodak and others. This is the first camera of that class in Olympus’ series – until now this manufacturer didn’t produce digital reflex models with interchangeable lenses. Although developed from scratch, the camera is reported to be a success. Its resolution of 5 megapixel and a 18x13.5mm sensor are enough for getting high-quality images, and the Zuiko Digital series has lenses for any even in your life and is being constantly expanded.
This camera is equipped with a 128MB buffer that stores up to 12 frames and allows for a serial shooting up to 12 frames at a speed of three frames per second. Although this is rather an average speed, the camera can be considered as a universal product that can be used for reportage (sport) shooting, too.
My tests proved that this camera writes the image to the memory card very quickly. The Olympus E-1 reveals all the advantages of fast cards compared to slow ones. Just take a look at the following diagram:
The RAW+JPEG writing mode – when two files are written to the memory card (an original image and a compressed version) – is the most convenient and universal mode for a number of applications and it loads the “camera – memory card” interface very hard. The total size of a series of 12 frames was about 160MB. It is here that the controller and chips of a fast flash card can show their best. There’s a triple difference between the best (48s) and worst (143s) runners in this test! That’s a nice bonus, really.
However, this speed gap is somewhat diminished at ordinary photographing, which doesn’t require a high fps rate. That’s because of the frame buffer – the camera is nearly always ready to take a shot, quickly throwing the contents of the buffer out to the flash card when preparing for the next shot. Thus, you don’t feel any difference in the camera’s responsiveness using a fast or a slow memory card.
So, again, fast media only make sense if you often use serial shooting. If you need to make many frames in a short time interval, you should definitely invest into a fast flash card. As for everyday shooting, there’s practically no difference between memory cards – the camera’s large frame buffer negates any.
Note also that the tested cards occupied slightly different positions in the results table for this test than in the test with the Canon EOS 10D. It means that every digital camera may have its own “favorites” among flash cards, which are not the same for all cameras.
Finally, I decided to check out a camera of the amateur class. The F828 has a big-resolution sensor (8 megapixel) and can record high-quality video (30 frames per second in the 640x480 resolution) with sound and zoom (thanks to the metal zoom circle on the objective lens).
A CompactFlash slot in a Sony camera is an unheard-of accident. Until now, the company has been sticking to its own standards, inventing proprietary formats whenever possible and developing products compatible with those formats only. Considering the influence of the company in the audio-video equipment field, the users of digital cameras from Sony had to put up with it, buying Memory Stick/Pro media.
Marketing considerations must have prevailed over the company’s engineering principles, and the F828 came equipped with slots for CompactFlash as well as for MemoryStick/Pro. This fact certainly added popularity to this camera, although with one reservation: you need MemoryStick/Pro or MicroDrive flash cards to record high-quality video – CompactFlash cannot be used for that. That’s strange since the required bandwidth can be provided by a fast CompactFlash card. The performance of MemoryStick Pro and MicroDrive devices cannot be considered overall higher than that of CompactFlash cards. Well, that’s a subject for another review…
You can feel the borderline between professional and amateur products in the serial shooting mode. The Sony F828 allows photographing up to seven frames at a speed of 2.5 fps, but only in JPEG format. The camera can also shoot in TIFF and RAW formats, but only with single frames. These limitations are probably due to the insufficient size of the camera’s frame buffer or may have been imposed deliberately. Anyway, I’m going to check out the camera’s writing speed in the serial shooting mode (in JPEG format):
You see that the camera has no favorites. Moreover, the write speeds of the first and last cards are 16s and 25s, respectively. This is no great difference. A more interesting fact is that you get about the same numbers if measure the write speed with single RAW or TIFF frames. Thus, if you shoot in JPEG format, fast media won’t help you much. And otherwise, if you’re getting the highest quality from the camera and always shooting in RAW or TIFF formats, you need to use fast memory cards with it.
And, somewhat off the topic of this review, I want to remind you that you need Memory Stick Pro or MicroDrive to record video in 640x480x30fps format.
If you’ve been examining the results carefully enough, you have certainly noticed that the leaders vary between different “camera – memory card” combinations. It is also evident that the gap in performance of the fastest and slowest card also varies depending on the camera. I suppose there are cameras (not included into this review) that wouldn’t have any speed difference working with any flash card! My point is that you should always consider the performance of a memory card with the device you are intending to use the card with. Each “reader – memory card” system will have its own unique performance level.
So, the first fact that can be derived from my tests reads as follows: make sure it makes sense for you to buy a faster memory card. Compare the performance of your current card to the one you intend to purchase, with a simple stop-watch: serial photographing, exposure of no less than 1/250s, white balance preset and aperture size – in accordance to the lighting, manual focus, maximum quality of files (RAW or TIFF would be optimal). If the difference is obvious – the purchase makes sense. And vice versa…
I decided to use several readers in this testing session to check out my supposition that the performance of a memory card may vary between card-readers as well. As you will soon learn, this supposition was confirmed. But let’s first examine the devices I plugged those cards into:
This single-format card-reader (supports CompactFlash media) is among the fastest devices of its class today – no wonder many flash cards had the highest speed in it. The enclosed interface cord is rather short and thin, but I didn’t try the reader with another cord.
This device is a multi-format USB 2.0 card-reader of an original design. You can use it in two ways: mobile and stationary. In the first case, the reader unit can be taken off the docking station and used independently with an interface cord. Or it can be plugged into the docking station with its USB port.
There’s a common opinion that universal card-readers are slower than single-format ones and that FireWire devices are faster than their USB 2.0 counterparts. Of course, I can’t make any generalizations comparing just two devices, but anyway I can say that that opinion is not always true.
So, let’s run some benchmarks.
As I already said, I used two patterns to ensure a higher precision of the results. One operates with a single and large file and another with numerous small files. First, the readers will be doing some writing. Although a reader is supposed to read information, writing tests will help us estimate better the speed potential of the flash cards.
The first thing that catches the eye is the SanDisk reader being slower than the Lexar with the fastest flash cards. However, some cards (Kingston Elite pro 1024MB and Transcend 512MB, for example) are noticeably faster in the SanDisk reader. The most curious results are those of the 1GB card from Apacer: it performs well enough in the Lexar, but awfully bad in the SanDisk! I checked this result out several times, but to the same effect. What’s strange, benchmarking in the AIDA32 suite doesn’t show this phenomenon…
Let’s proceed to data read tests:
In fact, a memory card’s reading speed is not very important for digital photography applications – you can wait a bit here. Anyway, I can’t help marking the excellent performance of the SanDisk cards in this test. This is the more curious since these cards had average results in the write tests. It means that their controllers are specifically optimized for writing. We should recall here that CompactFlash flash cards have numerous applications, besides digital photography, and some of them may require high read speeds.
Otherwise, the situation with the two readers remains the same: some cards are faster in one reader and slower in another.
This benchmarking suite works with storage media at a lower level than the FC Test and this accounts for the difference in the results. So, the Linear Write Speed test comes first:
Although the leaders remained the same, there are surprises starting from the third place. For example, the Transcend 45X 512MB card, which didn’t have good write results in the FC Test, approached the top of the diagram here, with very good showings. You may note that some other cards also behave this way. I suspect that different benchmarks produce different results (both absolute and relative) since they use varying methods and techniques to measure the performance. This is the nature of synthetic tests and you can’t do anything about that. We could argue for a long time about the advantages and shortcoming of a particular benchmark, and with no result.
Next goes the Linear Read Speed test:
This time there are no surprises at all. The SanDisk cards are again far ahead of the group, especially in the Lexar Media reader: the gap to the closest rival, the 1GB Transcend 45X card, is as big as 4MB/s. The last test is Average Access Time, whose results are of little importance in real life. But anyway, I’m curious…
The results of the SanDisk cards look like a requital for the high read speed! The access time of these devices is really too big – this must be due to their controller. The rest of the cards have normal results. Note also that the access time is in average 0.5-0.6msec better in the Lexar reader than in the SanDisk.
I will compare the memory cards according to three scores: overall score, camera score and card-reader score. I calculated the scores by finding the geometric mean of the results the card had got in the appropriate tests (the AIDA32 Average Access Time didn’t affect the overall score), and normalizing the results by a 5-grade scale.
I’m now going to add a few words about the products of each particular manufacturer.
The products from this manufacturer left me confused. Although the 512MB card from the PhotoSteno Pro series is supposed to be the photographer’s choice, the ordinary Apacer 512MB is its superior in every respect, and even leaves all the other test participants behind in the camera tests. Meanwhile, the 1GB card from Apacer is no record-breaker, sometimes finding itself at the bottom of the results table.
The mystery is solved easily, though. The backside of the PhotoSteno Pro card reads “This Product incorporates SuperFlash technology licensed from SST”. You can’t find this text on “ordinary” flash cards from Apacer. So, I suppose that when the PhotoSteno Pro series was introduced, memory chips from Silicon Storage Technology Inc. were among the fastest then, and Apacer took them for that professional series. But times are changing and cheaper products may have overtaken chips from SST in performance. Anyway, the 512MB card from Apacer is very fast – it is the winner of my today’s tests.
At the same times, you cannot rely on the fact that other Apacer cards have the same speed – the results of the 1GB card are very poor.
I had only “ordinary” flash cards from this manufacturer, although they also ship the eFilm.Pro series, for which a higher performance is specified. Anyway, the Delkin Devices 1GB card did well in the camera tests, although worse in the card-readers. The 512MB device from the same series passed the tests with opposite results.
Analyzing the edge stamps of several cards from Delkin Devices we came to a conclusion that a majority of products of the ordinary eFilm series are supplied to Delkin Devices by Toshiba, but we couldn’t identify the origin of the 512MB card – and this must have been the reason for the two cards to be so different in the tests. You may want to pay attention to the edge stamp; it looks like “THNCF…” on cards made by Toshiba.
I tested only one card from Digitex and to a great surprise: the device notched a record of 6.51MB/s write speed in the Lexar Media FireWire RW019 card-reader. This card doesn’t belong to any “special” professional or speedy series, but this didn’t prevent it from behaving very confidently in the camera tests. Unfortunately, we couldn’t identify the origin of the card, so I can’t tell you what supplier ships such fast devices. Anyway, Digitex gets my respects as its card made it to the third overall position in my tests. I hope other cards from this company are as fast.
The two cards from Kingston that I used in this test session belong to two different series: the 512MB one is “ordinary” and the 1024MB one belongs to the “speedy” Elite Pro series. Kingston plays it correctly: the Elite Pro does have a higher performance, both in digital cameras and card-readers. However, the “ordinary” card is not much slower.
The edge stamp on the Elite Pro series card says it is manufactured by Toshiba; there is no stamp on the 512MB card.
This renowned manufacturer is represented with four flash cards in my review. Three of them belong to the legendary Platinum series. A Platinum card astonishes with its very design – it is in a metallic case that increases the robustness of the product. The labels on the face of the card are holographic and shimmer with all the colors of the rainbow under light.
The exterior alone is not enough to gain popularity among professional photographers, though, and the two Platinum cards of 1GB and 2GB capacities are on a high performance level – the former is a bit faster than the latter. The 512MB device, however, did rather badly in my tests, even worse than the card from the “ordinary” series. That said, I cannot but wish that Pretec were stricter about the selection of the components for its memory cards. The elite “Platinum” series should come equipped with the fastest chips available.
Unfortunately, we couldn’t get samples of the popular “speedy” Ultra II series for our tests – I had to benchmark two “ordinary” cards instead. These devices were record-setters in the reading tests, but their writing speed was rather low. As for the untypical results in the Average Access Time test, I can’t name an application where this parameter of the CompactFlash card might be important. So, these cards are excellent at reading, but rather average at writing.
The 45X series cards from Transcend behaved similarly in the tests, notwithstanding the difference in storage capacity. Moreover, these devices turn to be well-balanced as they work fast both at reading and writing and both in cameras and card-readers. The 1GB card was better in readers and the 512MB one – in cameras. Overall, the Transcend 512MB card scored second best after the Apacer 512MB. You should note that a majority of Transcend’s 45X series cards of other capacities are high-performing, too.
After all the winners and losers are identified, I’d like to summarize the results of the today’s test session as a few facts: