We test the speed of a sampling of memory cards to see what difference the higher speed cards make in reality with current digital cameras.
With the frequent releases of new memory cards for digital cameras a common question on the discussion groups is whether you need these faster cards or whether they will make a real difference to your shooting.
In some situations they will, but normally they will not. All digital cameras have an internal memory buffer. When you take a shot the image goes to the internal memory, with whatever processing is being done, and is then written out to the memory card. If you take another shot before the first is written out, it goes in the buffer and gets written out after the first one. All cameras have a limit to the buffer size. So if you are shooting as fast as your camera is capable of you will fill the buffer quickly, after a certain number of images. The camera will not then take a shot until at least one has been written to the buffer. At this point you can keep taking shots but only at the speed it takes to write one image to the memory card.
Now in normal shooting you will only rarely encounter this. In common shooting situations you take one or perhaps a few shots in rapid succession but then you pause, either to examine the image on the screen to check exposure or to find the next subject. The only times I hit my camera’s buffer limit is on the rare occasions that I shoot sports or when shooting kids and their fast action. But then I’m a pretty staid type of photographer. There are plenty of others who hit this limit all the time. So what do you do?
The buffer size of the camera provides an almost absolute limit to how many shots you can take at the full speed of the camera before it will pause. But if you shoot below this maximum speed there is a way to get effectively infinite continuous shooting. There is also a way to get the camera to empty its buffer more quickly so you can do the next burst sooner. The way is to buy faster memory cards.
Two questions need to be answered: can you camera make use of the extra speed and is it worth the extra cost? Let us answer these.
To do this test I took four dSLRs that took CompactFlash cards and one compact camera that SDMC cards. These are the Canon 350D/XT, Canon 400D/XTi (mine), Canon 40D (in for testing), Nikon D3 (in for testing) and the Canon G9 (in for testing). Each was set to manual focus, manual exposure, fastest continuous shooting, RAW and a shutter speed of 1/250 second. With the camera on a tripod so it was easier to handle I started a stopwatch at the same time as I held down the shutter button. I then held the shutter as long as the camera kept taking shots at the full speed. Then I release the shutter and stop the stopwatch when the memory write light on the camera goes out (indicating the buffer is now empty). I note the time and the number of shots recorded and move on to the next card.
For this test I had four CompactFlash cards: SanDisk Ultra and Ultra II(1GB cards, so early Ultra II, not the latest revision); and Lexar Professional 133x and 300x. I also had two SDMC cards: Sandisk normal and Extreme III.
The dSLR results are shown in the graph below. In all cases there was a massive improvement in going from the Sandisk Ultra to the Ultra II. From the Ultra II to the Lexar 133x gave an improvement but nowhere near as massive. In all cases except the Nikon D3 there was no benefit in going to the Lexar 300x. I believe this is because in these cases we were hitting the limit of the sensor readout and processor speeds. Of those tested only the D3 could be pushed faster. One thing should be noted in these times, and that is the number of images taken. The 350D only took six images with the Sandisk Ultra but seven with the other, faster, three cards. The 400D took 12 images with every card. The 40D took 21 images with the Sandisk cards but 23 with the Lexars and the Nikon D3 took 16 with every card except the Lexar 300x, when it took 17.
To investigate further I took these times and divided them by the number of frames taken to give the time per image. This is shown in the graph below.
What you see is that with the Sandisk Ultra the time to write is proportional (approximately) to the size of the RAW files, as you would expect. I did find it interesting that as you get to the fastest cards you find the cameras pretty close together. If you use the fast Lexar 133x or 300x cards the two Rebel models should be able to continuously take one image every 1.5 seconds. With the 40D and D3 they should be able to maintain one frame per second indefinitely.
With the SDMC cards I only had two models on hand, the Sandisk normal and the Extreme III. On the G9 I held the button down for five frames and timed the interval from the start of this to the time the write light went out. With the basic Sandisk card this took 15 seconds. This dropped to 9.7 seconds with the Extreme III card.
So what did I learn from this exercise? Well faster cards do make a difference if you shoot in a way where continuous shooting is necessary. With all except the Nikon D3 there is no point going to the Lexar 300x, but each step up with the other cards was worthwhile. I checked the prices on these cards. To make it easy I took the US Amazon price for a 2Gbyte CompactFlash card. They are $30.25 for the Ultra II, $42.95 for the Lexar Professional 133x and $53.95 for the Lexar 300x. Given these prices it is definitely worth going to the Lexar 133x or similar from other manufacturers if you shoot action.