Was the Coco3 competing against the Amiga and Atari ST?

[random_rodder]

My bad - it wasn't Eric who did Gold Runner 2000 on the CoCo 3 - it was Chet Simpson. Eric said it couldn't be done...

[JamesD]

 

...

This entire thread has been coulda, shoulda, woulda.

The Aquarius could have blown away an A4000T had Mattel not taken so many shortcuts, shut down video when benchmarking and spread some chicken bones around it.

Um yeah, you are reading something into my comments that isn't there.

At no point did I ever say the CoCo 3 was a direct competitor.

I said even with the advanced chipset it would have been a competitor only as a budget machine.

 

The point you missed on shutting down the video to run that benchmark was that it was on a CoCo 1.

The CoCo 3 didn't require shutting down the video, which means a CoCo 3 at 1.7MHz can out benchmark a 4.77 MHz PC normally.

With a 6309 it's around the speed of a 6-7MHz 8088 PC while still running 6809 code at 1.7MHz.

With 6309 code it's probably the equal of an 8MHz 8088 PC... while still running at 1.7MHz.

So you might not want to laugh too hard about the comparison.

BTW, there is a 4MHz hack for the CoCo 3 that's been around a long time so don't be too surprised if a CoCo 3 can be noticeably faster than the PC for under $25 in mods.

 

[JamesD]

...

JamesD already addressed the CPU differences and each one's capabilities. I'll add to his comment about the benchmark, though. I remember that as well as an article in Computer Shopper (I think) in the late '80's discussing machine language programming. It stated that moving from a PC to a CoCo was like getting out of a Pinto and climbing into a Ferrari. It was easier to code for and, and the code would execute faster. I'll let historians and programmer argue that one.

...

The 8088 has a programming model more like the Z80 than the 68000.

If IBM hadn't used it in their PC it never would have been as widely accepted as it was.

[mos6507]

I never even knew the CoCo3 existed back in the day. I did spend a fair amount of time with the older CoCos window-shopping at Rat Shack and found them underwhelming, especially in the graphics and sound department. If the CoCo3 had been called something else and sold in a whole new modern looking case it probably would have done better, but to most people, the CoCo was a decidedly acquired taste by then and people just didn't give it the time of day. Really a failure of marketing more than anything else.

[Bill Loguidice]

I never even knew the CoCo3 existed back in the day. I did spend a fair amount of time with the older CoCos window-shopping at Rat Shack and found them underwhelming, especially in the graphics and sound department. If the CoCo3 had been called something else and sold in a whole new modern looking case it probably would have done better, but to most people, the CoCo was a decidedly acquired taste by then and people just didn't give it the time of day. Really a failure of marketing more than anything else.

 

There was clearly no place in the market for a completely new 8-bit computer by late 1986 (in fact, a large portion of the competitive market from just a few years earlier was already orphaned by this time). It would have to have been an extension of an existing line, like it was. The CoCo 3's sole purpose was to wring a few more years out of the CoCo platform in as economical and profitable way as possible. In that role, I think it did its job.

[potatohead]

The CPU does not wait for RAM refresh on a coco 3. At 4mhz, it will be very fast.

 

Dang, I should mod mine and get a 6309...

[JamesD]

 

There was clearly no place in the market for a completely new 8-bit computer by late 1986 (in fact, a large portion of the competitive market from just a few years earlier was already orphaned by this time). It would have to have been an extension of an existing line, like it was. The CoCo 3's sole purpose was to wring a few more years out of the CoCo platform in as economical and profitable way as possible. In that role, I think it did its job.

I agree to some extent but if you look at sales numbers of 8 bits they were still strong for a few more years.

 

The CPU does not wait for RAM refresh on a coco 3. At 4mhz, it will be very fast.

 

Dang, I should mod mine and get a 6309...

The CPU isn't socketed on the CoCo 3 or everybody'd do it.

[random_rodder]

No,its not socketed,but, that's easy enough to do. I might look up Sock Master's mod and make one.

 

And I've got about 40 6309s...

Edited May 5, 2015 by random_rodder

[potatohead]

Hmm... yeah that's a slow down for sure. Hard to deal with those old PCB's. I can probably do it... Good to know somebody here has a stash.

 

If anything, bumping the CoCo to 512K is probably the bigger bang for the buck upgrade for me.

 

Right now, I've an Apple card project to chip away at. But maybe one day this coming winter, when there is a lot of inside time for the CoCo.

Edited May 6, 2015 by potatohead

[random_rodder]

I never even knew the CoCo3 existed back in the day. I did spend a fair amount of time with the older CoCos window-shopping at Rat Shack and found them underwhelming, especially in the graphics and sound department. If the CoCo3 had been called something else and sold in a whole new modern looking case it probably would have done better, but to most people, the CoCo was a decidedly acquired taste by then and people just didn't give it the time of day. Really a failure of marketing more than anything else.

 

There's some truth to this statement about marketing. However, when Nick Marentes interviewed Mark Seigel (the CoCo's developer) about the CoCo 3 in his book CoCoNuts, Mark brings up a very valid point:

 

Nick: Many Color Computer enthusiasts have these bad views of Tandy. What is your feeling on this?

Mark: In a lot of cases they are right, but the Color Computer community never sent articles to the major magazines about the machine. You know, enthusiasm is catching, but you have to pitch it before someone catches. The Apple II started by word of mouth. So I think both Tandy and the Color Computer community were at fault here.

Nick: How do you feel being the developer of the Color Computer 3?

Mark: I was pretty happy with the way it turned out. I'm not even sure in retrospect that I would have done anything much different. The Color Computer 3 cost under $100 to manufacture at the start. I defy anyone to show me a better price performance than this.

Nick let me use the info in his book on an article I wrote on the CoCo line for Retrogaming Times Montly, also available here. If anyone wants even more in depth history of the CoCo line, I highly suggest getting a copy of CoCo: The Colorful History of Tandy's Underdog Computer. It's a great read that any retro computing fan would enjoy.

Mark was absolutely correct about the enthusiasm part: while Color Computer users were an enthusiastic bunch, they didn't spread it around much. They had three magazines completely dedicated to them: the Rainbow, Hot CoCo and Color Computer Magazine - the longest lived was Rainbow, with publication lasting over 10 years. I was fairly frequent reader of Computer Shopper and Compute!, and while an occasional article about the CoCo would appear, reader response and articles were overwhelmingly in favor of the more mainstream machines of the day. Tandy made a vain attempt at going a bit more mainstream with the CoCo by trying to sell the TDP-100 in other venues, i.e.: department stores. However, I don't think I ever saw one sitting next to an Atari or Commodore at Sears or K-Mart.

[JamesD]

Don't forget Color Computer News.

I don't agree with Mark's comment. Magazines only accepted articles for machines they supported.
You'd get a thank you but no thank you letter.
A couple TRS-80 magazines accepted articles and had a few good programs but support was pretty inconsistent.
Compute supported the CoCo for a while but they only wanted to support non-Extended Color Basic.

Everything was text... but then a lot of their games were just programmable characters.
They even printed a graphics based character generator for the Apple so they could use custom fonts to build games for the Apple II just like the other machines.

[potatohead]

Those graphics made sense for BASIC.

 

Some games did use low resolution.

 

The Coco had lots of clone games that seemed to give it a non legit feel. Many in my community also called Radio Shack, shit shack. This rubbed off on the computers for sure.

 

IMHO, the 3 didn't get fully exploited. It's quite capable for an 8 bitter.

 

Maybe somebody will go through and patch or write some 256 color software. That, to me, would have made quite a difference back then, but it went unused.

 

Rad Warrior is one I think about from time to time... That one probably could be spiffed up, particularly in 512k. It came on cart too.

 

Compiled sprites were not used back then either. Not that I know of. 6809 chips can move a lot of RAM by abusing the Stack. One byte per pixel graphics seems a great fit for the technique, demonstrated a while back.

 

I never did get into the Nitro much, though a few of us jammed on FLEX. From what I see today, that level of systems programming being available must have made the machine attractive to embedded types.

 

Most Coco users I knew either got stuck with it, or were tech, HAM radio types.

Edited May 6, 2015 by potatohead

[Joe Stella]

i'm in the camp, never heard of it. that's neither good nor bad. i love learning what else was out there, and this is fascinating.

[potatohead]

Yeah, check these out: http://atariage.com/forums/blog/105/entry-6693-color-computer-3-artifact-art/

 

http://atariage.com/forums/blog/105/entry-6701-a-well-known-image/

http://atariage.com/forums/blog/105/entry-6683-more-color-computer/

 

NTSC composite display required. Resolution is 160x200 or so lines, depending on vertical GIME setup.

 

Jason Law figured out quick image conversions.

 

When I had this machine in the day,I was rendering nice fractals on the TV, and had started on some graphics... but I only had cassette and just didn't progress too far. Thought I would have seen the technique used, but never did.

 

It is now. I shared it a while back, and some coco users ran with it, one guy playing movies. No link to that one.

 

FWIW, more colors are there to be had, but somebody is going to have to set up a screen with interrupts and palette entry changes.

[Bill Loguidice]

The mythical 256 color mode in the CoCo 3 is almost certainly just that, a myth. It has yet to be found despite some incredibly aggressive attempts at figuring out a way for many, many years.

 

As for the movies, John Linville was running them at various shows last year, so perhaps that who were thinking of.

[potatohead]

That is what I saw. Bill Buckles has recently been working on improved dither for the Apple and it's artifact colors. IMHO, the movie output can be improved considerably combining both techniques.

 

http://www.appleoldies.ca/bmp2dhr/

 

IMHO, the "real" mode seems like a test, or some other feature, and it never made sense to me. How do the more graduated color level actually get output?

 

Never went looking for it. Suppose somebody could de cap a GIME and go looking for real.

 

On the other hand, NTSC artifact color works well, and it can be used with very little fuss. A little creative palette management in the 320 mode would yield more colors too, in a way very similar to how the PC guys exploited CGA.

 

And, if scan line doubling is used, one gets a byte per pixel, graphics 7 type mode that would be fast.

 

There is also the idea of PAL type artifacting too. I don't have a PAL CoCo, or I would have already done it. The technique is used on many other machines from that era, just as the NTSC style was.

 

Horizontal resolution could be better in PAL. There is a little opportunity for somebody to do new things there, IMHO.

Edited May 6, 2015 by potatohead

[JamesD]

If someone wants a 256 color mode we'd be better off creating a new GIME in an FPGA.
Lets face it, we don't know if it made it past the prototype, if it did we don't know if all GIMEs have it and we don't know even how to activate it anyway.
Sadly, the odd number of pins the GIME uses means there are no existing socket adapters for it and a custom one would have to be made.
(that I found anyway)

1 byte per pixel would be awesome for wireframe games like Elite. The line drawing would be much faster.
Tile based games like Ultima or early Zelda games could look amazing.

[potatohead]

That is my thought on both points.

 

Well, that can be done on composite on a Coco 3 now.

 

About 233 unique colors result from the 640 pixel mode, writing bytes at a time, with the 4 colors set to black and the three Luma values.

 

That is how those images linked above were done.

Edited May 6, 2015 by potatohead

[JamesD]

Well, I don't have a CoCo 3 to try it anymore so I won't be doing it.

So John Kowalski, the guy that came up with the 4MHz mod, says the 6309 with the 4MHz mod offered a 38% speedup on 6809 code.
Some instructions like the multiply and 6309 memory transfer offered greater speedup (~50% for TFM) due to the number of internal clock cycles.

If you do the math, that may or may not be as fast as an 8MHz 8088 depending on the difference on that benchmark I mentioned.

That's about like a 6.58MHz 8088 if the 6809 is the same speed as 4.77 MHz. So it's probably about equivalent to a 7MHz 8088.+- a little.
For 6309 code that should be pretty close to an 8MHz 8088 +- a little so my guess isn't too far off.

[JamesD]

The 4MHz mod
http://users.axess.com/twilight/sock/4mhzproj.html

[random_rodder]

I'm going to try to work out the schematic based on logic diagram. Might take a bit....

[Mr SQL]

Don't forget Color Computer News.

 

I don't agree with Mark's comment. Magazines only accepted articles for machines they supported.

You'd get a thank you but no thank you letter.

A couple TRS-80 magazines accepted articles and had a few good programs but support was pretty inconsistent.

Compute supported the CoCo for a while but they only wanted to support non-Extended Color Basic.

Everything was text... but then a lot of their games were just programmable characters.

They even printed a graphics based character generator for the Apple so they could use custom fonts to build games for the Apple II just like the other machines.

 

And Dynamic Electronics, CoCo-ads (a CoCo newspaper mag) and T&D subscription software (a monthly periodical on tape or disk full of programs)

 

I wrote for all of them as well as The Rainbow; I had a specific colation order depending on the quality of the program. My best went directly under my software label Saint John Gallery Software, next up was T&D, then the Rainbow... CoCo-ads was almost at the bottom but it was still pretty good - I used a pen name for that periodical The mag at the very bottom I forget the name, someone probably remembers. They folded after less than a year and actually rejected the one program I sent them (a program that turned the keyboard into a musical instrument and let you save your riff's). that mag started with an S I think, came as a B&W cheap staple job, like Dynamic Electronics (that one was pretty cool content-wise though).

 

It was cool to see Compute! supporting the CoCo but I agree the graphics in the games were disappointing compared to the c64, atari and even the VIC-20 versions. I think BYTE also had some great articles early on.

[JamesD]

Tape magazine Chromassette also supported the coco for a couple years but none of these were general computer mags.

I only saw a few CoCo articles in BYTE that I remember but I may have forgotten some.
I thought there were a few articles on the 6809, a couple on the intro of the CoCo and then almost nothing.

 

*edit*
What's sad about COMPUTE! is they could have done a hi-res font routine just like the Apple II.
I've ported a 64 column one to a couple different machines/processors including the MC-10 and it's pretty fast.
It's faster than the ones that use an odd character width and I believe faster than the 64 column setting in VIP Writer.
Porting it to the 6809 would be a piece of cake.

Edited May 8, 2015 by JamesD

[Jinks]

So with the 256 color graphics mode and an accelerator it would spank c64 a8 st amag. But they decided it would be too close to the 1000 so they detuned it. That what I heard. That would be the ultimate retro find would be for someone to figure out the secret graphics mode.

[Bill Loguidice]

So with the 256 color graphics mode and an accelerator it would spank c64 a8 st amag. But they decided it would be too close to the 1000 so they detuned it. That what I heard. That would be the ultimate retro find would be for someone to figure out the secret graphics mode.

 

Unfortunately, that's nothing more than a convenient myth that gets repeated way too often. The truth is the CoCo 3 was constrained by cost and management indifference. As we detail in the CoCo book, there were plans for a more ambitious version of the CoCo 3, but that was quickly squashed before real work began on the hardware. It had nothing to do with concern over eating into sales of the Tandy 1000 series computers.

 

And again, the "secret graphics mode" is also a myth.

Edited May 11, 2015 by Bill Loguidice