If you own a camera or camcorder that uses SD cards, chances are you've confounded yourself at some point trying to figure out how fast a card you need (or want) for your device. This is complicated by a lot of factors, primarily that the Class specification defines a minimum sustained transfer rate--Class 6 is 6 megabytes per second, Class 10 is 10MBps, and so on.
That's like saying ais rated for at least 10 miles per gallon; it helps you filter a few other out of your buying decision, but it's not very helpful for sorting through the myriad others with a similar lowest-common-denominator rating.
Nor is there any official body that verifies the performance. The SD Card Association provides some guidelines and testing tools, you pay your licensing fee for the Class logo, and you're in business.
For example, according to Rob Galbraith's tests for SD cards in the Nikon D90, the sustained JPEG write speed for Class 6 cards runs from about 3.6MBps (Fail!) to as fast as 18.5MBps. But by the Class metric, you can theoretically stick both in your HD camcorder and have no performance worries. And what if the manufacturer simply decides not to pony up for the Class logo? Lexar's unlogofied 133X Professional SD claims a sustained write speed of 20MBps. Virtual Class 20! Update 5/22/09: Lexar wishes you to know that the card does indeed have a Class 6 rating.
Interestingly, the only card on which the company seems to put the official rating logo is the Full HD Video Memory Card, which is Class 4, and as far as I can tell, the only card for which it doesn't provide any additional performance specs.
So Panasonic U.K. announced the first Class 10 SD cards, the Gold series, with (obviously) a minimum transfer rate of 10 MBps and a stated maximum rate of 22MBps. Right next to me is a Class 6 8GB SanDisk Extreme III SDHC with a maximum rate of 30MBps that Galbraith's tests indicated had a JPEG sustained write speed of 18.4MBps in the
My point is that if you're confused, it's completely justified. And a rep from the SD Card Association told me earlier this year that this lovely system will continue with SDXC.
All this makes it difficult to figure out which card to buy, especially when presented with large price disparities between seemingly similar cards. Unfortunately, there's no easy solution, because few companies offer complete specs on their cards (for simplicity's sake!).
It might help to understand where you will and won't see performance gains from a faster card, however. For instance, your camera won't take pictures any faster, since that's primarily an autofocus speed issue, and the time to save a single shot has more to do with controller overhead than media write speed.
Nor will you get a better burst frame rate in a digital camera (unless the manufacturer specifically states that it requires an xMBps card to achieve a given rate) or better video quality simply because a card is faster. What we have typically seen with digital cameras is the ability to shoot longer burst runs with faster-than-minimum cards. But even there, once you reach the equilibrium speed where the card writing never bottlenecks at the camera's fixed maximum frame rate, the extra bandwidth gets wasted.
Right now, the highest bit rate at which most consumer HD camcorders and cameras are encoding video is 24 megabits per second, and the highest resolution is 30 frames per second 1920x1080, for which they're requiring Class 6 cards (though I've seen some contradictory requirements for Class 6 in one place and 10MB/sec in another for a given product). So I'm guessing that until we see encoding bit rates increase, there shouldn't be a need for faster cards.
Speedy SD mostly comes in handy when it's time to transfer images and video to your computer (a task for which the Class spec is completely meaningless, by the way). And here, as Rob G.'s testing shows, a reader that supports UltraDMA makes a big difference.