OT: A call to arms (sort of)

From: Allison J Parent <allisonp_at_world.std.com>
Date: Sat Jul 3 14:01:38 1999

S100 was one of the first to see people using different cpus as they could
still use their old boards (sometime with mods) or they could "bend"
the cpu to fit the bus adaquately.

<You may indeed be onto something here, Allison, but the ISA is no less
<general in its inherent qualities than the S-100, and I'd submit that a
<major case for the S-100's popularity for non-8080 applications was the
<ready commmmercial availability of numerous desirable functions at
<reasonable prices. That's how the ISA occurred to me.

Indeed, the key was commonly available generic functions. ISA8/16 fits
the bill fairly well. Infact myself and another are working on a ISA16
motherboard with a z280 insted of intel. Why, keyboard, serial, video,
floppy and hard disk interfaces are all done and cheap to free. We do not
need a raft of interrupts, so thats not a big problem. It's not a perfect
match but general enough if we bend a rule or two it will be fine. After
all we want to run z280s at full bore speed for software development and
the specifics of the platform are relitively unimportant other than we'd
like the two of them to be the same.

In 1989 S100 would have been a choice for the same reasons.

<Aside from that, a general purpose not processor-biased architecture would
<provide a few control signals, e.g. IORD, IOWR, MEMRD, MEMWR, maybe a coupl
<of clocks, probably one fairly fast one suitable as a dot clock for a vide
<circuit, and one slower, suitable for bus transaction timing, a few
<interrupt and DMA support signals, and a couple of dozen address lines.
<Parity and maybe a "tilt" line would be handy, as well as a wait signal.
<Most busses have these signals in one form or another.

Sounds like ISA. ;) If memory is kept off the bus ISA is just fine for
most 8/16 bit cpus I can think of. The only provision would be the cpus
that may DMA impossible or very painful but even then DMA is not a required
capability for some smaller designs.

<The key element for generalized development, though, is whether or not you
<can afford to buy the functions you don't want to build right away. Don't
<you agree?

Bingo! or for many reasons don't care to build. A 16450/16550 serial is
uninteresting but, serial is one of those must haves in a system like
mabe a modem. Others make little sense to fabricate when even new
they are under $50!

Received on Sat Jul 03 1999 - 14:01:38 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:32:10 BST