kaihp said:
The 7D has a deeper buffer (~25 RAWs), vs 18 RAWs using a UDMA-7 card on the 5D3. Here the 5D3 and 50D are similar (16 RAWs).
The 7D vs 5D3 buffer depth in real life was not as in the Canon manuals per my experience (although the 7D metering slowing fps makes it hard to compare the exact same way), I got:
Lexar 1000x, it honestly didn't really do any better than a Sandisk Extreme Pro 90MB/s card with a 5D2 or 7D (or surely xxD) cam but with the 5D3.... WOW. And it lets the 5D3 deliver best in class buffer and longer contrinuous shooting times than either the 5D2 or 7D. (UPDATE: 2.0 for 7D boosts 7D performance with 1000x card)
A few numbers (all tests at 1/800th shutter, no IS, no AF, no NR or any other in cam jpg special processing options, max continuous drive mode selected (7D pointed at a bright light so as to maintain max frame rate)) with the raw number being number of shots before you hear it slow down shooting and timed number is the continuous time you could shoot non-stop at the camera's max frame rate (derived by a simple frames/max fps of camera):
Test 1 (lens cap on or super underexposed near black frames (i.e. with requiring minimum for storage) at ISO100 on 5D3):
14 shots with a slow Lexar 16GB 200x card
17 shots with a 30MB/s SanDisk Extreme III 8GB card
20 shots (3.3s) with a 32GB SanDisk Extreme Pro 90GB/s card
35 shots (5.8s) with a 32GB Lexar 1000x card (whoa! 35! vs 20! 5.8s of continuous shooting vs only 3.3s!)
(so the Lexar 1000x makes a surprisingly large difference on the 5D3 and is well worth the $ if you seriously shoot action with the 5D3, do note that the 16GB 1000x card is said to be slower than the 32GB and larger 1000x cards so the advantage may be less if you go for the 16GB size; with this speed card the 5D3 continuous shooting time easily beats the 5D2 and 7D)
Test 2 (as above but on a 5D2):
16 shots (4.1s) with a SanDisk Extreme Pro 90MB/s card
17 shots (4.3s) with a Lexar 32GB 1000x card
(so the extra $ for the Lexar 1000x makes no difference on a 5D2; with this speed card the 5D3 actually lets you shoot continuously for less long than the 5D2 or 7D)
Test 3 (as above but on a 7D):
23-29 shots (up to 3.7s) with a SanDisk Extreme Pro 90MB/s card
23-28 shots (up to 3.5s) with a Lexar 32GB 1000x card
(not sure one time I got the large 28-29 shots and now I get more like only 23, must be related to how little detail I got on the image, 7D shoots slow fps in the dark so I couldn't do it with lens cap on, on a similar test condition when 7D got 23 frames the 5D3 got 27 frames)
(so the extra $ for the Lexar 1000x makes no difference on a 7D, same result)
Test 4 (as 7D above with now with 2.0 firmware):
getting about 27-30 instead of closer to 23 for the current bright scene being shot (5D3 gets about 27) so the 7D now does about the same to a bit more frames than the 5D3 although at 8fps the total continuous shooting time is LESS than for the 5D3
and then trying it with a Sandisk 90MB/s extremepro instead 7D 2.0 gets:
about 23
(so with 2.0 firmware the Lexar 1000x now DOES make a HUGE difference for the 7D even if not quite as insanely huge as with the 5D3; also appears to show that you'd need one of the best of the best, the Lexar 1000x cards to take any advantage of the 2.0 buffer)
Test 4 (these have an actual image being shot and the ISOs are higher so each RAW file is larger, the performance drops considerably, especially above ISO800 and NOTE that the numbers will vary depending upon the scene shot as different scenes and exposures will produce different file sizes and any given scenario may bump all the numbers noticeably up or down; all are on the Lexar 32GB 1000x card):
ISO3200 5D3 - 20 shots (3.3s) - a big drop from the 35 for ISO100 black frame
ISO1600 5D3 - 24 shots (4s)
ISO800 5D3 - 30 shots (5s)
Test 5 (as above only this time a SanDisk Extreme Pro 90MB/s card was used in all cases):
ISO3200 5D3 - 14 shots (2.3s)
ISO1600 5D3 - 20 shots (3.3s)
ISO800 5D3 - 22 shots (3.7s)
Test 6 (as above only this time a SanDisk Extreme III 30MB/s card was used):
ISO800 5D3 - 15 shots (2.5s)
So yeah for most Canon cameras the Lexar 1000x will not be worth the money at all (unless you are want to pay that much for faster transfer speed to your computer and your computer supports USB3 and you have a USB3 card read that also takes full UDMA 7 advantage) but for the 5D3 wow if you are serious about shooting action and dont want to be saddled with a shorter shooting time buffer than the old 5D2 or 7D then get the Lexar 1000x, suddenly you actually can shoot for longer times non-stop than with the 5D2 or 7D and often by a large margin. So for the 5D3 and action it is well worth it. (UPDATE: and with the 7D the Lexar 1000x is well worth it too now)