[Coco] NEC PC6001, a shadow of the CoCo

[Steve Ostrom]

Steve B, I really love your stories.  Thanks for your updates and sidebars.

-- Steve O --



----- Original Message ----- 
From: "Steve Bjork" <6809er at bjork-huffman.net>
To: "CoCoList for Color Computer Enthusiasts" <coco at maltedmedia.com>
Sent: Monday, March 03, 2008 10:32 PM
Subject: [Coco] NEC PC6001, a shadow of the CoCo


> As someone that programmed a great deal on both the Z-80 and 6809, I say 
> the 3.8 MHz Z-80 is a match (in general) for .89 MHz 6809.
>
> Most op-codes on the Z-80 take about same time to do the same work as the 
> 6809 with memory running at those speeds.  Yes, there are times the 6809 
> can pull ahead of the Z-80. (But the same can be said for the Z-80 too.) 
> It's all how you code them.
>
> Back in my days at DataSoft, I was giving the task to write a version of 
> Mega-Bug for NEC's PC 6001.  Like any new system, I took a peek under the 
> hood to see just what the computer could do.  Let's just say I did NOT 
> like that I saw.
>
> The big flaw in the (over) design of the computer was the marriage of the 
> Z-80 to the 6847 and their incompatible memory interface. The 6847 must 
> get its data on a regularly time cycle without missing a byte. (During H 
> and V blanks the 6847 could miss out on data without messing up the 
> screen.)  This works will with the 6809 because it can time the data 
> request so not to get in the way of the 6847.  (This is key why both the 
> 6809 and the 6847 can access the same memory at full speed without messing 
> anything up.)
>
> But the Z-80 can NOT sync its self to the timing cycle of the 6847 because 
> a op-code's memory fetch could come in 3 to 7 (or more) clocks apart.  So, 
> the Z-80 would add wait states to all memory fetches till the 6847 was 
> done with the memory.  These wait states slowed the Z-80 from 3.8 MHZ to 
> something more like 1.1 MHz.
>
> Because the system runs so slowly (with unreliable timing), they had to 
> put a microprocessor in every peripheral including the Cassette Recorder 
> Unit.  Everything was overpriced for that computer! (Boy, they could of 
> used someone like Steve Woz designed the computer to get the most out of 
> the hardware!)
>
> You may ask how did the slower speed of the system effect Mega-Bug? 
> Because of all the wait states on the Z-80,  I could not put the magnifier 
> screen overlay in the game. (Well, I did try to put the overlay in but it 
> ran at an unplayable 4 frames per second.) Without the magnifier overlay, 
> all you got was the screen full of a maze with a few small dots moving 
> around on it.  Not very impressive!
>
> NEC was not happy with the way Mega-Bug turned out on their system. (I 
> shipped it with an option to turn on magnifier.) The producer of the game 
> over at NEC and I had a little talk about just how slow this new computer. 
> (He could not belive it!)  He ask for a report on my findings so he could 
> pass it on to his boss.  As it turned out, Mega-Bug was the first port 
> competed for their computer.  Everyone else writing games was also having 
> problems getting their games working too.
>
> After sending my report, I got a call from the head of NEC personal 
> computer division. He wanted me to fly me out to Japan to talk with his 
> team.  While it would be nice to visit Japan (on someone else's dime) but 
> I knew that would have to go "head to head" with the whole NEC computer 
> department.  Thanks, but no thanks!  I just told him that I had too many 
> projects on my plate to fly out there.
>
> Well, the next day, he called and asked to bring a "few" of team members 
> to meet with me.  I said sure and setup for a meeting the next week.  That 
> "few" turned out to be 8 engineers and 4 of NEC's top executives.  I gave 
> a little song and dance on why the Z-80 and 6847 was a bad match and how 
> well 6809/6847 synchronize bus worked.  It did not take long before the 
> engineers where in hot water for no knowing how poor their design was 
> compared to computers like Apple and Tandy's systems.  At the time, the 
> floppy drive for the CoCo cost about 1/3 the price of the drive for PC6001 
> and it loaded data at  cassette tape speeds.
>
> DataSoft (or any other American company) never got another port project 
> for the PC6001 and it was very quickly replace with a better designed 
> model.
>
> That the story and the reason for the short life of the NEC PC6001.  I 
> could not wait to ship that piece of crape back to Japan.
>
> Steve (Mega-Bug) Bjork
>
>
> At 04:14 PM 3/3/2008, you wrote:
>
>>I think that a  3.8 Mhz Z80 is no match for a 1.8 Mhz 6809, may not even 
>>be for a 0.9 Mhz
>>All my friends had Sinclair Spectrums (or clones), and when we managed to 
>>find some similar programs, the CoCo was usually faster.
>>Of course, it might have been because many other reasons beside CPU power, 
>>but....
>>
>>Diego
>>
>>>>Can we create a similar game on arguably worse hardware (I still say
>>>>YES)...?
>>>>
>>>>
>>>Arguably?  I'll argue that point!  :)  According to old-computers.com
>>>the PC-6001 had 16-32K of RAM.  The CoCo easily meets or exceeds that
>>>capability.  I think one could make a case that the 6809 is a generally
>>>superior processor to the Z-80.  Comparing a 3.8 MHz Z-80 to a .89 MHz
>>>6809E is apples and oranges of course, but I think the 6809 generally
>>>gets more done per clock cycle than the Z-80.  Still, that's a pretty
>>>big gap in clock rates, so the 6001 might have a slight edge in raw
>>>CPU.  I'll bet Sockmaster could give us a pretty good comparison, given
>>>his experience with the Donkey Kong translation.  But the 6809 may still
>>>make up some ground in sophistication that it loses in RPM.  We can MUL,
>>>doggone-it.  And a 64K CoCo might find opportunities to trade storage
>>>space for speed in ways that a 32K 6001 couldn't.
>>>
>>>For graphics, it looks like they are at a dead tie, unless the chip in
>>>the 6001 has a few enhancements that the stock 6847 doesn't.  So they
>>>can't boast better graphics.  It looks like the CoCo's only real
>>>deficiency in comparison to the PC-6001 is in its sound hardware.  But
>>>the sound chip in the 6001 is (I believe) a subset of the chip in the
>>>Radio Shack Speech and Sound Cartridge.  So if you are lucky enough to
>>>have one of those, the 6001 has basically nothing on the CoCo.  If the
>>>6001 has better hardware than the CoCo, it's marginal -- maybe a little
>>>more raw CPU for some purposes, and better built-in sound, though a CoCo
>>>expansion betters it with an improved version of the sound chip, plus
>>>hardware speech synthesis to boot.
>>>
>>>So I'm with you, Andrew.  I don't think there's any reason the CoCo 1/2
>>>couldn't have that game.  The CoCo 3 of course could blow the 6001 out
>>>of the water, but that's not really a fair comparison.
>>>
>>>JCE
>
>
> --
> Coco mailing list
> Coco at maltedmedia.com
> http://five.pairlist.net/mailman/listinfo/coco
>