Re: Have any of you put this chip in your //gs?

[mjmahon@aol.com (Michael J. Mahon)]

Bill Garber wrote:

>Are you saying they have broken the 64K barrier on
>a machine with only 16 address lines? I suppose that
>if you were to switch at each 64KB and at each 4 MB
>which would be 64K of 64K, then you could go as high
>as you want, but you would need to go pretty fast to
>make it worthwhile. Ok, I'll agree that there is reason
>to have so much ram.

Actually, 4MB is just 64 x 64KB, or a 6-bit bank number.

Since 8-bits is a convenient size, a typical extended banking
scheme will permit 256 x 64KB, or 16MB of memory.

RAMWorks and similar cards use such an 8-bit bank number,
and the design naturally extends to 16MB.

Bank switching is inferior to direct addressing, since making
an access to a random location requires a store of the bank
number prior to making the access, and addressing controls
to allow code fetches to continue to occur from the original
code bank.  Apple II's and their memory cards typically do
not have these controls, so the actual bank accessing code
must usually be stored either in a non-bank-switched area
of the address space, or the code must be replicated in
all banks which it can access so that bank switches do
not disrupt code execution.

Notwithstanding the programming issues, bank-switched
memory can be quite useful for memory expansion,
particularly if the accesses are to "segments" of memory
that can be swapped into the bank containing the code.
A bank-to-bank move is much faster than a disk read!

-michael

Check out amazing quality sound for 8-bit Apples on my
Home page:  http://members.aol.com/MJMahon/