OK. As it has been explained to me, the controller chip on the CF card is
responsible for making it appear to have cylinders, heads, and sectors just
like an IDE HD. However, it doesn't really work like that. As a matter of
fact, there are extra memory areas built into the card. The controller chip
doesn't always write to the same block. It knows how to vary the areas it
writes to, to extend the life of the card.

As far as NetBSD is concerned, it is not optimized for compactflash. For
example, I am sure that some form of fragment management and defragmentation
is built into the code, whereas fragmentation wouldn't be a concern on a
compactflash card because of the way it is read. The card reads at the same
speed across the card and it doesn't know the difference between getting
continuous and non-continuous data. This is because a hard disk is read in a
circular motion and it only goes one way. If you read one sector and then
need to read the previous sector, you have to wait a revolution.
Compactflash is closely related to an array. You can read any part any time
you want.



Terry
----- Original Message -----
From: godwin <godwinchan@hotmail.com>
To: <port-hpcmips@netbsd.org>
Sent: Thursday, April 20, 2000 11:04 PM
Subject: NetBSD file system on z50


> Hi all
>
> I just had my computer science exam and it was about file systems. I had
> this question that I had never considered.
>
>     Is the file system used NetBSD currently is optimized for Flash? or it
> doesn't really matter?
>
> What I mean optimized is the factors like fragmentation etc. Well
> considering something like my SanDisk 128.. it's way more expensive per
> megabyte than any hard discs + there is a finite number of times we can
> write to the cards (has anyone busted one of those cards yet?). So we
would
> want to minmize swapping and fragmentation right? But right now as far as
I
> can figure out Windows and even NetBSD just treat the CF cards as normal
> hard discs right? On the upside flash is much faster access time than hard
> discs, so what we consider long/ big overhead operations in hard discs
would
> be minimal in CF right?
>
> I don't know enough about *NIX filesystem to comment, that's why I like to
> find out! :-)
>
> Thanks all
>
> Godwin
>