Does the Speed of Your Memory Card Really Matter?
Mark Pemberton
Mark Pemberton
May 4, 2010
Have you ever been out shooting and had the experience of pressing the shutter button and nothing happened. You realize immediately what had happened. You had been shooting that eagle swooping in to pluck a fish from the river and, shooting at 5 frames per second, filled your camera’s buffer. You curse yourself as you miss the climactic moment when the eagle grabs the fish and flies off.
Both Cathy and I have experienced this problem on multiple occasions. Even when we are trying to be good and not lay on the shutter we have had occasions when the camera locked up and would not fire. Needless to say this usually results in a string of expletives that I will not share here.
The question is how can this situation be avoided. The camera manufacturers aren’t helping. They keep increasing the continuous high-speed burst rate. Our Nikon D200 camera bodies can fire at the rate of 5 frames per second; the D300 at 8 frames per second.
After our most recent bad experience I bought a couple Lexar Professional 300X 8 GB UDMA compact flash cards. These are not the fastest cards on the market but they are a lot faster than the cards we had been using. Until now we had been using pq1 100X 4 GB CF cards, or A Data 120X 4 GB cards that we bought many years ago at a bargain price at Frys. At a cost of $60 per card I wanted to know if the expense was worth it. I decided to do some testing and find out for myself.
TEST #1
I designed two sets of tests. In the first series of tests I would compare the read / write speed of three different cards while uploading and downloading images to and from the computer. The three cards that I chose for this test were:
Pq1, 100X (15 MB/s) 4 GB
SanDisk Extreme III, 30 MB/s (200X) 4 GB
Lexar Professional 300X, (45 MB/s) 8 GB
To measure the read / write speed of the cards I copied 100 raw images into a file folder. The size of the 100 images was 1110 MB. To copy the images between the computer and the cards I used a Lexar RW019 Firewire 400card reader. For one additional test, I also used a 9 in 1 USB card reader (manufacturer unknown) with the Lexar 300X card just to compare card readers.
The results of this experiment are shown below.
Lexar RW019 | 9 in 1 | |||
pq1 100X | SanDisk Extreme III 200X | Lexar 300X | Lexar 300X | |
Rated Speed | 15 | 30 | 45 | 45 |
Write Time (s) | 266 | 129 | 94 | 160 |
Read Time (s) | 100 | 98 | 100 | 133 |
Write Speed (MB/s) | 4.2 | 8.6 | 11.8 | 6.9 |
Relative Write Speed | 100% | 206% | 283% | |
Read Speed (MB/s) | 11.1 | 11.3 | 11.1 | 8.3 |
It was also interesting to note that the 9 in 1 USB card reader performed much worse than the Lexar Firewire card reader. The USB card reader took 70% longer in the write test and 33% longer in the read test. I won’t be using that card reader any more.
This experiment tells me that using my current equipment, the new Lexar cards that I bought will not speed up the transfer of my files from the card to the computer. However, that isn’t why I bought the cards. Thus we proceed to test #2.
TEST #2
For test #2 I compared the performance of the same three CF cards under real world shooting conditions. However, before I describe the test conditions it is important to recognize that there are many variables associated with capturing an image. In terms of speed, one of the most important is the file format that you choose.
Cathy and I are Nikon shooters. We have the option of shooting in RAW, compressed RAW, TIFF or JPEG formats. JPEG offers the option of Fine, Normal or Basic (Good, Okay or Crap). For Test #2 I chose to test 12 bit RAW, 12 bit compressed RAW and JPEG Fine. The approximate file sizes for these are 19.5 MB, 12.1 MB and 6 MB respectively. The compressed RAW and JPEG files are obviously much smaller but require more processor time to compress the files.
For the test I chose to use a Nikon D300S body with a 35 mm f2 lens. All tests were performed in manual mode with a shutter speed of 1/400 sec and f8. Autofocus was set to manual as well.
The testing process was as follows. Each CF card was inserted into the camera and formatted. The camera was set to continuous high (8 frames per second) and the image mode to RAW, compressed RAW or JPEG Fine. I aimed the camera at the target and then pressed and held the shutter button down until the buffer filled up, as evidenced by the camera ceasing to fire (RAW) or a rapid reduction in the rate of shooting (JPEG). I then recorded the number of images collected and the time from when I initially pressed the shutter button until transfer to the card was compete.1 I also recorded the approximate average file size. This was repeated for each file type and each card. The results are presented below.
When shooting RAW the cards performed precisely as expected. The San Disk 200X and the Lexar 300X cards were 233% and 353% faster than the 100X card. However, when shooting in either of the compressed file formats the situation changed. With compressed RAW the numbers were 176% and 235%. For JPEG the relative speeds dropped even more to 132% and 142% respectively. The reason for this drop in transfer speed is that the camera’s processor has to spend a lot of time compressing the file before it can be transferred to the CF card.
| CF Card | File Type | Photos to Fill Buffer | Seconds to Clear Buffer | Write Speed (MB/s) | Relative Speed |
| 100X | Raw | 17 | 70 | 4.6 | 100% |
| | Comp. Raw | 17 | 44 | 4.3 | 100% |
| | JPEG Fine | 32 | 51 | 2.1 | 100% |
| 200X | Raw | 17 | 30 | 10.8 | 233% |
| | Comp. Raw | 17 | 25 | 7.5 | 176% |
| | JPEG Fine | 43 | 52 | 2.7 | 132% |
| 300X | Raw | 18 | 21 | 16.3 | 353% |
| | Comp. Raw | 20 | 22 | 10.0 | 235% |
| | JPEG Fine | 80 | 90 | 2.9 | 142% |
When shooting RAW the cards performed precisely as expected. The San Disk 200X and the Lexar 300X cards were 233% and 353% faster than the 100X card. However, when shooting in either of the compressed file formats the situation changed. With compressed RAW the numbers were 176% and 235%. For JPEG the relative speeds dropped even more to 132% and 142% respectively. The reason for this drop in transfer speed is that the camera’s processor has to spend a lot of time compressing the file before it can be transferred to the CF card.
One of the most important numbers is how long it takes to clear the buffer so that shooting can resume. When shooting RAW the 100X card required 70 seconds, the 200X card required 30 seconds and the 300X card cleared the buffer after only 21 seconds.
The other interesting observation was the number of images required to fill the buffer. With RAW and compressed RAW formats the buffer filled up after17 or 18 images. One might expect that the compressed RAW should have accumulated more images than the RAW format due to the smaller file size but, in test after test there was almost no difference between the two. And because the buffer filled up so fast, approximately 2.3 seconds, there was no difference between the three cards.
The situation was very different with JPEGs, however. With the 100X card 32 images were accumulated, the200X card allowed 43 images to accumulate and the 300X card allowed a whopping 80 images to accumulate before the buffer filled up. Unfortunately this then required 90 seconds for the buffer to clear.
CONCLUSIONS
So is the expense of a fast card worth the cost? That depends upon the type of photography that you do. If you are a landscape photographer you would be wasting your money on a fast card. However, if you photograph sports or wildlife and shoot RAW I would highly recommend it. The first time you miss a crucial shot because the buffer is full you’ll realize this.
If you normally shoot JPEGs the answer is not as clear-cut. Even with the 100X card I was able to capture up to 32 images before the buffer filled up. Under normal shooting conditions of course you would probably be shooting in short bursts of 5 or 6 shots. Under those conditions the buffer should not fill up. However, I should point out one problem with my data. The average JPEG file size in my test was 3.3 MB. This is very small and reflects the target that I chose. The degree of compression is highly dependent upon the complexity and contrast of the target. I was using a very simple target. Normally JPEGs are more on the order of 5 or 6 MB in size. If that were the case my numbers would have been much different. However, I would guess that you could get away with a slower card if you are shooting JPEGs.
Despite the claim of the manufacturers I saw no difference among the three cards with respect to the time required to upload images to my computer. It is clear that card reader is the rate-determining step here. While the measured time of 100 seconds to read 100 RAW images doesn’t sound too bad you have to put that into context. Between Cathy and I, we have been known to shoot over 1500 images in one day. Admittedly, that is a rare event but with a read speed of 1 image/second 1500 images requires 25 minutes to upload. That’s a long time for impatient people such as myself. I guess it’s time to start looking into faster card readers next…
Footnotes
1 To measure the time required to clear the buffer I simply watched the memory card access light on the back of the camera. When the light went off the transfer was complete. This technique worked well for the RAW and compressed RAW formats but not for JPEGs. Because so much time is required to compress the files prior to transfer the memory card access light is flashing making it hard to know precisely when transfer is complete. So for JPEGs I watched both the indicator lamp and the LCD preview screen. I knew transfer was complete when the LCD shut off. I then subtracted 4 seconds from the total to account for the fact that the LCD preview time is 4 seconds long.
