JJ,

If the file is not fragmented, the system doesn't need to determine
the next cluster number, when reading a file for e.g. copying. i.e.
when reading data, it only need to increase the next cluster number
by one, each time

And in the other case it uses the current sector # and uses it as an index
in a look-up table. I don't think you will notice the difference. Not
even when retrieving a million sectors.

strikethru
But yes, when you define the the time consumption of everything else as
being zero - meaning you do not allow something as common as a second
thread/background process - and jack up the ammount of sectors that you are
going to retrieve I guess you could get an actually measurable time
consumption somewhere along the line ...

.... but it would be be devoid of any meaning.
/strikethru



Hold the presses:
I thought that MOVing a register from a table would cost at least double the
cyles of in INC (on an X86) but some googeling seems to show they cost the
same ammount ...

So, the answer is: No difference.

(but I left my origional answer there as "strikethru").

Regards,
Rudy Wieser