[plug] usb storage howto

Arie Hol arie99 at ozemail.com.au
Sat Dec 2 13:23:01 WST 2006



On 2 Dec 2006 at 13:07, Kev wrote:

> Arie Hol wrote:
> 
> > FAT16 will go to a maximum size of 2Gb - maximum of 65,525 clusters 
with
> > a  size = 32,768 bytes
> >
> > Multiplying the maximum number of clusters (65,525) by the maximum
> > cluster size (32,768) equals 2 Gb. 
> >
> > FAT32 will go to a maximum size of 4Gb - maximum of 65,525 clusters 
with
> > a  size = 65,525 bytes
> >
> > Multiplying the maximum number of clusters (65,525) by the maximum
> > cluster size (65,525) equals 4 Gb. 
> >
> > Although - FAT16 can become inefficient with files sizes greater than
> > 64Kb.
> 
> I always understood FAT becoming inefficient with small file sizes, not
> large.  Bear in mind that the smallest amount of space a file can 
consume
> is 1 cluster.  Therefore a 1 byte file on a 2g FAT16 partition will 
still
> consume 32,768 bytes of disk space.  With large files, each cluster 
which
> the file occupies, except the last cluster, is filled to capacity (so 
to
> speak) and the only wastage is in the final cluster.  Hence /much/ more
> efficient use of space.  All file systems have their equivalent to
> clusters.  ie they have a minimum storage unit size.  Most good file
> systems allow you to manually set the 'cluster' size, so as to optimize
> disk usage for particular situations.  IBM's HPFS was one of the early
> /very/ good file systems, with a default cluster size of 512 bytes,
> manually changed if the user wished to do so.   ReiserFS and ext3 (I
> think) both use 4096 bytes as their default cluster size, but this is 
also
> manually changed if the user wishes to do so.
> 
> I think that stuff is correct.  At least it was once ;-)
> 

Yes you are correct.

It looks like I made a typo - I should have said ;

"files sizes less than 64Kb"


Regards Arie
------------------------------------------------------------------
 For the concert of life, nobody has a program.
------------------------------------------------------------------



More information about the plug mailing list