Monday, July 26, 2004
Posted by Jason Dunn in "HARDWARE" @ 09:00 AM
When it comes down to which card to buy, you'll be thinking about more than just speed. Other factors to consider include price, warranty, availability, and brand – this article will not address those issues. Instead, I'm focused completely on raw speed, and how the speed of the card relates to its overall value.
The Flash Cards Tested
Wanting to eliminate as many variables as possible, I focused on testing only 512 MB Secure Digital Cards. I requested cards from over half a dozen vendors, but only Sandisk, Simpletech, and Delkin responded to my request. Each vendor was asked to send samples of their currently shipping 512 MB SD cards, both the consumer level cards and their high-end professional cards (if applicable). Sandisk sent me their Ultra II card. I had already had a 512 MB consumer-level Sandisk card in my possession for a few days, and it was fully tested, but I lost my first batch of test data when Windows XP abruptly died on my laptop and I didn't have a backup (I know, I know…). Simpletech sent me a 512 MB SD card in a ProX package, but the label indicated it was their standard consumer-grade card. Testing confirmed that it was their consumer level card. Delkin sent me both their consumer grade eFilm card and their professional grade eFilm Pro card. The cards covered in this shoot-out are:
Sandisk 512 MB SD Ultra II
Simpletech 512 MB SD
Delkin eFilm 512 MB SD
Delkin eFilm Pro 512 MB SD
First, a word about the methodology I chose: Flash cards will perform differently based on what type of device is reading and writing to the card. So while a card might be blazing fast in a high-end digital camera, it might perform slowly in a phone. Thus, it's impossible to predict exact performance values across all devices. This article is focused on several constants, the only variable being the brand and type of card. The tests were performed using a Sandisk 6 in 1 USB 2.0 memory card reader connected to the USB 2.0 port on a Fujitsu P5010D laptop. No other programs were running on the laptop, and it was disconnected from the network to ensure maximum performance. No other USB devices were connected to any other port, and the laptop was connected to AC power and the CPU was set to maximum speed (900 mhz in this case).
Figure 1: HD_Speed in action
The software used to measure the speed was HD_Speed. Special thanks to Louis Solomon from Steelbytes for adding a timer to HD_Speed specifically for my use. Each test was run once for five minutes in length, so any speed blips would have been averaged out over time. It was for this reason that I only ran each test once – in initial repeated tests I saw zero variance in the final average after five minutes. Six tests were performed in total: three tests to benchmark how fast a device could read data from the card, and three tests to measure how fast a device could write data to the card. Each test utilized three sizes of data blocks: 256 KB, 1 MB, and 16 MB. I felt that these represented common scenarios: reading/writing files on a PDA (256 KB), taking and viewing photos (1 MB), and reading/writing RAW files in a high-end camera (16 MB). Those values are approximations of course, because every device will have different file sizes – my Canon EOS 300D takes JPEG images around 2 MB in size and the RAW images are 10 MB or so. The test sizes I used are a "best effort" approximation.