Memory Card Speed Matters

All CompactFlash cards are not created equal. Just as some hard drives and CD-ROMs are much faster than others, some memory cards are faster than others.

This was news to me at last summer's Macworld Expo, where Lexar was pushing their CompactFlash cards as 3x faster than the competition. Fast forward to mid-2001, and we see Lexar selling cards rated as 10x and even 12x.

Have they really left the competition that far behind?

That's the question Rob Galbraith (among others) asked. Galbraith has twice performed comprehensive studies of CompactFlash performance in high-end digital SLR cameras. He discovered that some cards are faster than others, but that no card is 12x faster than the competition. For a brand-by-brand comparison, I urge you to follow the links at the end of this article.

The purpose of this article is not to identify the fastest or best value CompactFlash card - Rob Galbraith has already done that. Instead, we want to take a broad look at memory speed, which should apply to other media as well.

Write Speed, First Report

In his February report, Galbraith found write speeds for a "fine" quality JPEG ranged from 366-960K/sec, a range of 2.6:1, on a Nikon D1. Writing a RAW file produced comparable results, with speeds ranging from 319-937K/sec, a 2.9:1 range. In other words, the fastest CompactFlash card tested was roughly 2.5-3 times faster than the slowest.

Galbraith next tested the memory cards in a Canon EOS D30, which had JPEG write speeds of 438-1008K/sec, a 2.3:1 range. For RAW files, speeds varied from 474-1022K/sec, a 2.15:1 range. The Canon EOS D30 writes faster than the Nikon D1, especially on the slower cards, but faster cards still have a significant advantage in write speed.

Galbraith's third test uses the Kodak DCS/Canon D2000 camera. RAW speeds range from 628-1485K/sec, a 2.35:1 range. In each test, the fastest memory was at least twice as fast as the slowest.

Read Speed, First Report

Galbraith next tested all the memory cards using various readers and computer platforms. We'll look specifically at his PC USB, Mac USB, and Mac FireWire results.

Using a Delkin eFilm Pocket Reader on a PC, read speeds ranged from 844-888K/sec, showing only a 5% difference between the fastest and slowest card. That's essentially unnoticeable.

Galbraith also used the eFilm Reader on a Mac, where read speeds ranged from 678-714K/sec, again a difference of just 5% between the fastest and slowest cards.

To see if the USB adapter itself makes a difference, Galbraith also examined the Lexar Jumpshot USB adapter, which gave speeds of 792-803K/sec on the Mac (not all CF cards were tested). Although still 10.5% slower than USB throughput on the PC, the Lexar Jumpshot was 12.5% faster than the Delkin eFilm on this test.

Of course, USB is not a particularly fast bus for moving data. For speed, you want FireWire. Galbraith tested the Microtech FireWire CameraMate and Lexar FireWire Reader on a Mac. The Microtech, which was not tested with all CF cards, had a range of 2186-2665K/sec reads - the fastest memory provided 22% faster reads. With the Lexar adapter, speeds ranged from 2268-2934K/sec, a difference of 29%.

Using fast memory, FireWire is over 3x faster than USB, which is especially helpful when working with high capacity memory cards.

Write Speed, Second Report

In his second report, Galbraith looks specifically at the Nikon D1X. He finds write speeds for "fine" JPEGs range from 617-1525K/sec, much faster than in his February report. Write speeds for RAW files ranged from 617-1732K/sec, a 2.8:1 range. Some of the same cards were tested in both reports, so it's evident the Nikon D1X writes much faster than any of the cameras in the first report.

Read Speed, Second Report

Using the Lexar FireWire Reader on a Mac, Galbraith obtained read speeds of 1163-2765K/sec, a 2.4:1 range. Even throwing out the lowest result, speeds were 2325-2765K/sec, showing the fastest card 19% faster than the slowest.

Conclusion

Remember that these tests were performed in high-end SLR digicams, so the results with your camera may vary. Still, the bigger the image your digicam creates, the more benefit you may see from faster memory.

The other variable is time - every manufacturer will continue to push the speed envelope, so the next generation of cards and cameras will probably be even faster.

Write speeds show a lot of variance; there's a lot less range when it comes to reading. All things being equal, look at the cards with better write speeds, then compare price to see which provides the best value for your needs. If possible, find out how fast your camera can write data - there's little point spending money on speed your camera can't use.

For instance, the Lexar 12x cards test at 5% faster then their 10x cards, but you're likely to see a much higher premium for the faster card. Unless you need the best possible speed, you'll probably find the best value a step or so back from the cutting edge.

When it comes to read speeds, use FireWire if that's an option. USB is fine for lower capacity cards, but if you're working with 64 MB or more, you'll appreciate the improved throughput of FireWire.

Now pop on over to Ron Galbraith's articles to get the speed information you need to pick the best CompactFlash card for your own use.

Resources

• Selecting a CompactFlash card for a photojournalist's digital SLR camera, Rob Galbraith, 2001.02.17
• Selecting a CompactFlash card for the Nikon D1X, Rob Galbraith, 2001.07.04
• CompactFlash speed tests on the Nikon 990, d-store, undated
• CompactFlash memory cards in a speed race, digitalkamera.de, 2000.08.19. Recommends faster cards for 3.3 megapixels and beyond.
• Digital film comparison, Digital Photography Review, updated regularly. A look at the speed of Compact Flash memory cards, including real world read and write performance. IBM Microdrive 1 GB still the write speed champion.