🚀 go-pugleaf

In article <4686@shlump.nac.dec.com> balzer@frambo.enet.dec.com (Christian Balzer) writes:
                        [...]
>2. A stoopid(tm) thing to do would be an A3000 without a 32 bit version of
>the custom chips. Enhancements that go beyond that of the ECS would be
                        [...]
>designers have learned this lesson. A really nice thing would be a 68000
>as fallback CPU on the motherboard. Whilst giving utmost compatibility with
>all the "standard" Amigas, this feature could be implemented at very low
>costs...

The motive being downward compatibility with the 68000 equipped Amigas.
But is that (financially) possible with a 32 bit custom chip set?  I
mean, it would be possible to make the 32 bit custom chips appear to
be standard 16 bit chips to the 68000, but how much extra would it cost
to design them as dual mode chips?  And if they can not emulate the
standard 16 bit chip set, then why on earth would they put a 68000 in
there if the result would still not be binary compatible?
                        [...]
>-The solution is simple, supply the A3000 with a powerful chipset (like large 
> screens [1024 by 1024 min.] more colors...) but don't try to do the 
> (financially) impossible and have the whole system try to refresh these
> displays at flicker free rates (25Hz are sufficient for animations). 
                        [...]
Faster DRAMs are beginning to appear in the market.  In a year or
two they will probably be much more available.  So what is (financially)
impossible today probably will not be (financially) impossible in a
year or two.  The 3000 is aimed at the higher end PC market anyway.
It could be designed with a significantly faster display and it would
be more expensive, but many people who are in the market for a machine
like the 3000 would buy it anyway.  And a year later the price would
come down when faster chips drop in price.

I think that the most affordable way to get the bandwidth is
through dual interleaved banks with two ports, implemented using
custom chips and standard 100 ns DRAMs.  The full bandwidth of
the chips is then available to the screen refresh through a
dedicated port.  Of course faster chips will increase the
bandwidth proportionally, but having 100% of that bandwidth
available to the screen is significant.

But my guess (not knowing anything factual) is that they probably
won't make it that radical.  It might be an overly risky approach
for Commodore right now.  Then again, they could decide to just
go for it :-).

--Steve
-------------------------------------------------------------------------
	  {uunet,sun}!convex!swarren; swarren@convex.COM