[Coco] What would a CoCo successor have to have as a minimum?

[Roger Merchberger]

On 11/24/2010 08:49 AM, Steve Batson wrote:
>
>
> Man! Lot's of different ideas here!
>
> Just wanted to add my 2 cents. Some seem to be willing to leave
> compatibility behind or drop it at the CoCo 3 level. Does that really make
> sense?  If you drop compatibility, there goes a huge sofware base which
 > means you have this new computer with limited stuff to run on it and it
 > will not be that appealing to nearly as many.

To some, it does. If you're building something brand new, it takes more 
resources the more "compatible" it is. It also adds "kluges" to the 
newer system as well (look at the discussions WRT the 65816 & all the 
wacked out (IMHO) 80x86 segmentation issues).

Also, voicing opinions helps inform the people actually creating the 
hardware/software what may be marketable... rather like a poll. If 
someone brings up CoCo1/2 compatibility and 300 people say "Keep it" and 
I'm the only guy saying "I wouldn't miss it" -- then it's pretty 
important. If the opposite happened, then not designing in the 
compatibility[1] might make things easier and the hardware faster as it 
could be optimized differently.

[1] For example: one of the lowest semigraphics modes on the CoCo1/2 was 
only used on a couple ROMpacks, one of which was the Appliance & Light 
controller system. Well, you can't run the controller from an emulator 
(as far as I know) -- the controller itself used the cassette port for 
communication; not the serial port. (if it had, that might've been a 
different kettle of fish.)

If dropping this mode that was almost never used (and can't be used for 
certain things in the new system anyway due to hardware limitations) 
should one use up valuable resources *insuring* full compatibility is 
there? Tandy didn't think so -- that mode isn't even in the CoCo3...

[[ Editor's Note: I'm not advocating one way or the other, or even 
playing "devil's advocate" at this point - I'm just speculating what's 
going through the architects' minds about how to build this new, faster 
yet mostly compatible system. ;-) ]]

> The question I have is, if a new system is going to be bult, why limit it
> like many have been taking about? Many have been suggesting designs that
> are more powerful than a CoCo, but still limited by today's hardware
> standards. Why do you want to build a computer that probably should have
> been built 10 or 15 years ago.

Compatibility. Any system that's compatible with the CoCo (whether in 
hardware or software emulator) is going to be slower than today's 
standards unless you just happen to have a functional 32nm chip fab 
system out in your garage.

...And gwarsh, if you did, think of what one could do with a 3GHz 6309 
with 8Meg RAM (all cache, all the time, eh?) and GIME all on a single 
chip... I'd bet that for general processing, it would run rings around 
any PC because it would never have to talk to anything external (slower 
RAM, etc.) except in the rare case of slower I/O..

[[ Dingalingalingalingaling... That's the alarm clock waking me up from 
dreamyland... ;-) ]]

 > Build a more up to date system with today's
> hardware capabilities (more memory, better graphics, fast processors, etc.)
> and make it compatible with the CoCo's so that the software base is not
> left been behind.

That's the problem -- eventually at least some compatibility has to go 
away... How much CGA/EGA/PGA software just won't run on today's machines 
because the hardware doesn't support it anymore? Should that 
compatibility have stayed forever? Now that 8Mhz "un-turbo" buttons 
aren't on PCs any longer, how much 8088/286/386 software's been 
abandoned because it *just runs too fast*? Who misses PC games *that early*?

[[ As an aside -- when I first got my dual-processor AthlonMP system (10 
years old now) -- 1.4GHz each -- I installed DOS on a small partition 
(for my EPROM burner) and I put an 8088 DOS pinball game on the 
system... and it ran... kinda. It ran so fast that it would play & drain 
_all 5 balls_ all within the time of the first keypress to release the 
first ball. The first go, I was like "whaddya mean 'game over?'" ;-) ]]

That's the balancing act of compatibility with new hardware.

> How about building a CoCo 4 board or device that plugs into a PC and
> leaverage all the PC Hardware somehow. Might require some type of basic OS
> or Loader to make it work.

[[ Now I am playing "devil's advocate"... ]]

What interface would you use? PCI? (I'm only guessing, but probably the 
fastest "general" interface a homebrewer could design for) -- That 
already limits you to 32-bit 33Mhz data transmission -- (very) roughly 
the same capacity & capabilities of the FPGA implementation.

How about PCI-X? 66Mhz at 64-bit -- you've now quadrupled your 
throughput, but seriously reduced who can use it; as those were only 
used on server boards. I can run it, but most people don't have the 
hardware.

AGP? You're already limiting yourself to a "dead" interface which is 
going away quickly, and most PC's only knew how to deal with a video 
card on the other end of that -- not much (any?) non-video I/O uses ever 
came out for that port. (You'd also have the catch-22 of: Once you 
install the new "AGP SuperCoCo" in your PC, what do you do for a video 
card?)

PCIe would be the "logical" new choice -- but can a hobbyist nowadays 
even *do* anything with that port?

> It just seems that many of the ideas people are throwing out are causing
> concern because of the cost. Why design something that one is going to
> build or buy because of cost?

Ah... most hobbyists have limited funds?

[[ And usually, the ones that don't have arguably better hobbies like this:

http://www.sinclairzx.com/spec-x-1.html

or, IMHO, even much more doubleplus fun:

http://www.dailymail.co.uk/tvshowbiz/article-1328742/Sir-Clive-Sinclairs-wife-Lady-Sinclair-sizzles-sun-sexy-beach-calendar-shoot.html

I'm just sayin'... ;-) ]]

> Just wondering. I'd buy something that is reasonable, works well and has
> support. But not If it's over priced or not compatible with a good software
> base. As a PC and Mac owner, I'm not going to want to go on to yet another
> system just because it's called a CoCo 4.

Not to mention... didn't you just answer your own question above?

I like threads (and posts) like this as shows the inventors that there's 
interest (sometimes quite passionate interest ;-) ) in projects like 
this and hopefully that spurs the people much smarter than me to keep 
building kewl stuff... and I let them know with my checkbook (when I 
can) that I always appreciate their efforts! :-)

Laterz!
Roger "Merch" Merchberger