http://www.robgalbraith.com/bins/camera_multi_page.asp?cid=6007-9424
Not sure where you ^ get the idea that larger cards are faster? Data speed is a function of... well... data speed. If anything, evidence points to smaller cards being faster.
There are some vague differences in the above referenced table between larger and smaller cards, but not enough to be significant.
This review states this, but gives no reference, so, kinda weak:
http://twwilliams.com/5d-mkii-faq/
I've found that buffer size of the camera is a bigger issue unless you're trying to take >17 pictures in a burst, and that's only been an issue when shooting in RAW+Jpg on the 5D MkII. Card speed only really affects how long it takes to clear the buffer and be ready to shoot again - unless you really cheap-out and get <50x cards.
http://www.cameratown.com/reviews/needforspeed/
"Because of the camera's internal buffer each of the memory cards were able to fire off an initial 12 images in the same amount of time. However, the faster memory cards recovered quicker thus freeing the buffer more quickly. Because of this I decided to hold down the shutter for 30 seconds with each memory card to see how many photos each would capture. This is where the speed differences become apparent - at least when shooting."
Card speed was only a really limiting factor when using 8x cards (!), and even then they were getting a continuous burst of 18 pictures with the XTi.
I use the Kingston 133x 16GB cards, I think I paid $45 for them at Fry's. I have a 4GB 266x card, and there's no difference in how many pictures I can take in a burst, or the FPS - the camera buffer is the limiting factor. RECOVERY time is faster with the faster card, but not by much - the camera's data pipe is the bottleneck, not the card speed.
Personally, I'd go for the 133x - which is about 'average' these days anyway, and get a bigger card, or get 2, smaller cards of equivalent total capacity than spending more on a 266x. (Fry's sells an 8GB 266x Kingston for $45, their 16GB 133x is up to $49 now).
Now, card speed makes an ENORMOUS difference in how long it takes to get the sodding files off the card and onto the PC! THAT might be worth it all by itself.
You should be able to look up the buffer size of your camera, then decide if you want to take more pictures than the buffer will hold in a burst, you'll need a fast card. But much above 133 seems to make no difference on the camera side of things, unless you start talking about 1D's and the like - and that's just conjecture on my part.
I know I can do 17 full-sized RAW+JPG images in a burst, no matter what speed card I have, and the FPS isn't affected at all either. This is on a 22 megapixel camera.
Of course, then I can go make a cup of tea before it's ready again if I'm using an 8x card, but shit, even finding one that slow these days would be hard
I'm not trying to start a pissing-contest, just trying to educate over the marketing hyperbole that floats around. There ARE advantages to getting faster cards for downloading, and recovery time after taking a burst- BUT - FPS is totally a function of camera buffer speed, and total burst is
mostly camera buffer, and once you get above 133x, it appears to make no difference (yet). I'm sure camera manufacturers will soon speed up the pipe enough that card speed will be the limiting factor again, but it isn't right now unless you get slow-as-dirt cards.
