@ TheCoda - You're wrong I'm afraid. There is nothing "un-scientific" about this test. It may not be the most exhaustive one in the world (hence my saying so), but that doesn't make it invalid.
Look, let's say the OS caches all 13Mb into RAM first, and THEN transfers it all to HD.
It doesn't matter in the context of this particular test as it will do the same with ALL the cards. It won't "pick and choose". Therefore, caching into RAM will NOT be a variable....it will be a constant across the cards.
HD read/write speed will NOT be a variable....it will be a constant across the cards.
Card reader read/write speed will NOT be a variable....it will be a constant across the cards.
I think you've failed to grasp the basics of how to conduct a test properly
.
The whole point is to reduce variables you don't want (in this case only leaving the ones I do - card speeds) and reduce other things like confounding factors......which by ensuring that the OS/HD/RAM/Reader etc is all the same and
utilised the same way across the board means that I have
.
The ONLY thing that using 100Mb would do is to reduce the error in specifically calculating the speed (ie give a more accurate "to the decimal point" value). However, it WON'T alter enough to alter the basic rank of slowest -> fastest that I have demonstrated above (ie, they're similar).
As I
only intended to show a rank to debunk the myth of "one country always being faster than another", and didn't require speed times accurate to the nearest 500 decimal places (2 will do, which is what I have), then 13Mb is enough for this.
Plus, commercial programs that calculate speeds for you utilise MUCH less than 13Mb and cope fine. A homebrew app on my DS performs the test (on the inserted flash card) with moving only 512b, 4Kb or 16Kb and gives consistent results across the board....
So no...you don't NEED 100Mb, it won't alter the basic results I have given above.
Linear Mode
