Data From Tests of multiple format methods on individual cards
set of test data and further reports.
Spread sheet of above results below.
Conclusion thus far:
By label the card that should be the fastest/best is not. When it was fresh out of the box it was the fastest. The Sandisk 6G Class 4 is currently the best in all categories. The answer to this may be in what they were formatted with (program) and/or what parameters were used. All I can conclude thus far is "what you see is not necessarily what you get". I will now test to see if the format programs make a difference.
Update:
New Test Data concerning SD Formatter 3.0 (Beta). I formatted the 4G Class 2 Sandisk with the above program using every parameter it allows. You can only "overwrite erase an SD Card". The Quick and Full formats with erase on and off were tested. Who won? Fooled me. The Quick had faster write/read speeds using post format H2Testw benchmark test as compared to the full (which takes forever to format). It was only a little faster on read and the same on write. This would seem to indicate that you might as well take the easy, and quick, way out! Did not test the larger cards.
Update: (8/19/10):
Tests of Format programs and cluster size on Sandisk 4G, Class 2
This is the smallest card in size and rated class that I have.
It was interesting to see the Quadrant scores come out so close and in the same range as all of the larger/faster cards. This would seem to indicate that the Quadrant test is not dependent on the SD cad for it's scores.
Note: During these tests I noticed that the first run of the SD Card Tester under Android would run extremely slow. Usually giving a speed for write of 0 or 1. I would also get errors and SOD's when I used Windows or WinMo to format the card. I do not know why the Android program runs so slow on the first pass. Possibly some difference in windows card format and Android. I do not know. (This also happened on the upcoming 16G, Class 2 tests.) After the 1st pass the tests were normal. Maybe this initial slow down is the cause of some of the problems ppl have with class 2 cards. It is total speculation, but maybe running some initial write/read tests would help smooth out some of the problems on the smaller cards.
Update: Tested a card with direct copy of backup on fresh format and there is no fragmentation shown by windows. Pocket Mechanic shows 0.01 fragmentation so I think this one can be tossed out the window. As files are added like music etc. this may become a problem. But fragmentation does not have any effect on the tests done here.
Final Update:
And the winner of this adult tricycle race is the original 16 G Class 2 card. I am now trying this card in everyday use to see if the Windows 64 format or the WinMo in the phone format have any different characteristics in use.
If I was going to choose "A" Format Program for "Any" card, I think I would choose WinMo in the phone.
Using programs that only write one FAT, versus the normal 2, might help once the disk get's fragmented, but this is easy to fix by defraging or backup to hard drive, format and recopy. The second FAT is great insurance for file system recovery and this test shows that on a defraged card there is little if any downside in speed.
What we learned: (By this very limited test.)
1. Card Class on the label does not have much to do with true read/write speeds. So you can't buy a card by it's cover and expect price/performance.
2. Card size and Class have little, if any, effect on Quadrant scores.
3. The format program and parameters you use, at least by these results alone, are not going to make a great deal of difference. Some programs/parameters might have a slight advantage. Fragmentation and usage may make some format/parameter methods behave differently in the long run. This was not tested here.
Conclusion: So go, my children and format your cards haphazardly. It probably is not going to make a lot of difference.
Update 8/22: Tested suggested Ubuntu format method and found no advantage as far as speeds or scores. (This suggestion seems to have been based on incorrect knowledge.)