Atari A8 design critique

[rdefabri]

On another thread, there was a quote from a Lucasfilm programmer (David Fox?) about why they moved almost exclusively to the C64 platform.  That quote included a comment that he thought the A8 platform was "better", which I find interesting.

 

That made me recall another discussion (maybe here on AtariAge, not sure) about the C64 engineering team, and their thoughts on the A8.

I'm curious what they did think - was that ever documented?  I'm in the tech field, and we always look at our competitors - not from a "we're better!!" POV, but rather from a "hey - what they did is pretty cool" / tech perspective.

 

This isn't a "who's better" question.  It's more about hearing their thoughts from a technical perspective - what was good, what could have been done differently, etc.  I know some of Bob Yannes' thoughts have been documented, but not much else.

[zzip]

Lucasfilm were some of the best at showing off what A8 could do, and their games usually lost something with the ports to other platforms,  so I suppose that's not surprising their programmers had that preference.

 

 

[xrbrevin]

time was also a factor back then. the A8 was released quite a bit prior to the C64.

i think the PET was released nearer to the A8 so the C64 development will have benefited from having more understanding of the market and also being designed to suit it's needs

[rdefabri]

1 minute ago, xrbrevin said:

time was also a factor back then. the A8 was released quite a bit prior to the C64.

i think the PET was released nearer to the A8 so the C64 development will have benefited from having more understanding of the market and also being designed to suit it's needs

Yep no doubt.  I'm sure there was some examination or look at the A8, especially given MOS Technology / 6502.  

 

I'm not an engineer, but I love watching the videos of Joe Decuir breaking down some of the design concepts / strategy around the 2600 / A8 / Amiga.  That's why I was curious about the C64 team - they surely looked at Atari / Apple / Tandy, etc...I'd love to know what they thought (good / bad / ugly).

[CommodoreDecker]

35 minutes ago, xrbrevin said:

time was also a factor back then. the A8 was released quite a bit prior to the C64.

i think the PET was released nearer to the A8 so the C64 development will have benefited from having more understanding of the market and also being designed to suit it's needs

 

The XL computers existed, but apart from 2x the color shades available it didn't have much more to offer.

 

Marketshare was definitely an issue, so how did Atari respond to the C64 with the new XE line? Well, the high-end 130XE model had 2x the RAM of the 800XL, and an exciting case design that made you think you were in Ten-Forward on the USS Enterprise NCC-1701-D... there were lots of re-released recycled games, but no other actual hardware improvements - such as graphics modes - to really one-up against the C64...  Seems a bit of a misfire...  Pokey was still, okay but by 1985 they could have increased octave range... or even be a trendsetter for stereo...  But even just adding in some higher resolution modes to complement the additional RAM and they could have had the C64's same higher resolution with far more colors than what it had and C64 would be smoked...  "Masters of Time" showed the strengths and weaknesses of both A8 and C64 - A8 had a far richer color palette, which helped, but the C64's resolution made the A8 look too blocky and ancient by comparison.   Other games like "Panther" made use of C64's graphics modes far better and looking more lively throughout, with A8 looking sufficiently inferior (though these newer games were probably built natively on the C64 then ported to A8 and not making as much use of the custom features...) But by 1985, Atari and Commodore were working to push the new 16-bit technologies to the customer market.

 

A shame, as Atari was the trendsetter and Lucasfilm's earliest games definitely show the A8 in a superior light.  

[zzip]

23 minutes ago, CommodoreDecker said:

"Masters of Time" showed the strengths and weaknesses of both A8 and C64 - A8 had a far richer color palette, which helped, but the C64's resolution made the A8 look too blocky and ancient by comparison.

When C64's full resolution was used, it was hard for A8 to compete, but because of technical limitations of that mode, many C64 games used 160x200 mode as a compromise and ended up looking more like the A8 versions.

[+MrFish]

Listen to the interviews with David Fox, Peter Langston, Aric Wilmunder, and Douglas Crockford on the Antic podcast; some good information there. Although, Fox has apparently forgotten that LucasFilm Games did The Eidolon and Koronos Rift for the Atari 8-bit computers too (speaking of the cuttoff point for Atari releases; probably because he didn't work on either of them...).

 

[Jetboot Jack]

3 minutes ago, MrFish said:

Listen to the interviews with David Fox, Peter Langston, Aric Wilmunder, and Douglas Crockford on the Antic podcast; some good information there. Although, Fox has apparently forgotten that LucasFilm Games did The Eidolon and Koronos Rift for the Atari 8-bit computers too (speaking of the cuttoff point for Atari releases; probably because he didn't work on either of them...).

 

I don't think that's true...

sTeVE

[+MrFish]

2 minutes ago, Jetboot Jack said:

I don't think that's true...

He forgot about The Eidolon and Koronos Rift being done on the Atari 8-bits in the interview he gave to Antic. Listen to it...

 

[Jetboot Jack]

not disputing that - just quoting the thread where in 2018 he clearly does remember the games...

 

David Fox gives a totally different version of events regarding Labyrinth:

 

Regarding an Atari 8 bit version of Labyrinth...

 

Sorry to disappoint you, but there never was one. By the time we decided to

do this game, the Atari was on its way downhill, having been beaten in the

marketplace by the C64. The last games we did for the Atari were Koronis

Rift and The Eidolon. Atari sales weren't especially good for either (I'm

pretty sure C64 sales for both were much better), and we saw the writing on

the wall. (Even though the Atari was by far my personal favorite, and I

still think it was a better machine.)

 

So when the Labyrinth game got the go ahead, it became the first Lucas game

to lead on the C64. If it had been commercially successful, we might have

done versions for other platforms. But the film didn't get great reviews,

and without a strong push from the film, Activision (the publisher on this

title) didn't do much to promote the game. I doubt it sold more than 25-30K

copies, but don't quote me on that.

 

David Fox.

[+MrFish]

5 minutes ago, Jetboot Jack said:

not disputing that - just quoting the thread where in 2018 he clearly does remember the games...

OK... yeah, I read that from the link you posted above already.

 

No big deal. He probably forgot it during the Antic interview, as I said, because those weren't titles he worked on. Great interview though, along with the rest.

 

[thorfdbg]

If I recall correctly, Lucasfilms Games was founded as a cooperation between Atari and Lucasfilms, thus no miracle the games first appeared on the Atari platform and not on the C64.

 

There is also an IEEE article on the design of the C64, and if I remember correctly, their authors mentioned that they studied the industry quite well before designing the VID. A particular point of critique was the rather limited sprite system of the 8-bit.

 

It's a shame Atari never invested into the 8-Bits anymore, the XL system did not add much, and neither the XE. However, creating an extended chip set is a major investment, and they didn't want to take that risk, apparently. Thus, they followed the cheap approach: RAM was getting cheaper, and putting more into the system did not require a lot of re-engineering of the design.

 

Big companies just work like this - they usually don't invent or invest, and leave the risk to smaller players (like Hi-Toro aka Amiga later).

 

[ivop]

They could have added an 80 columns mode with an integrated XEP-80 (well, the chip, not connected to PIA obviously ) in the 130XE, with added 80 column E: driver in the OS. That would have been great. IIRC the C128 has an 80 columns mode, too. Reusing XEP-80 tech would have dramatically reduced cost compared to creating a new chipset.

Edited September 28, 2022 by ivop
spelling

[rdefabri]

1 hour ago, thorfdbg said:

Big companies just work like this - they usually don't invent or invest, and leave the risk to smaller players (like Hi-Toro aka Amiga later).

Particularly in tech.  My company was recently acquired by a larger org that didn't want to recreate what we were already doing.

 

WRT Atari at the time, I seem to recall memory was a big point the C64 hammered home, and as you noted with memory prices coming down, that was an easy thing to implement.

 

I'm sort of a purist, but best I could see them doing would be some sort of graphics accelerator chip that piggy backed on the GTIA.  Something that would free up processing from the main CPU and allow more "sprites" via software manipulation (which I think is what VBXE does?).

 

Anyone have those IEEE articles?  I'd love to see that sort of thing.

[reifsnyderb]

1 hour ago, ivop said:

They could have added an 80 columns mode with an integrated XEP-80 (well, the chip, not connected to PIA obviously ) in the 130XE, with added 80 column E: driver in the OS. That would have been great. IIRC the C128 has an 80 columns mode, too. Reusing XEP-80 tech would have dramatically reduced cost compared to creating a new chipset.

I stated my opinion elsewhere and will state it here again.  I think Atari considered their home computer to be a game machine with a keyboard and treated it as such.  They never considered it to be the serious computer it was.  Not only is the 80 column issue evidence of this but the monitor port on the XL's as well.  A business would buy a monitor for a computer without thought.  Why not connect the chroma pin?  I think they figured it didn't matter because home users would use their TV sets.  

 

The 1090 would have allowed business upgrades as well.  It was cancelled.  The Apple II's had expansion ports built-in.  Why not combine an XL with a 1090 and run with it?

 

The technology was there.  The implementation was not.

[+MrFish]

7 minutes ago, reifsnyderb said:

I think Atari considered their home computer to be a game machine with a keyboard and treated it as such.  They never considered it to be the serious computer it was.

There was a division in Atari over the matter.

 

[+MrFish]

2 hours ago, ivop said:

They could have added an 80 columns mode with an integrated XEP-80 (well, the chip, not connected to PIA obviously ) in the 130XE, with added 80 column E: driver in the OS. That would have been great. IIRC the C128 has an 80 columns mode, too. Reusing XEP-80 tech would have dramatically reduced cost compared to creating a new chipset.

Actually, someone (I'll have to look up who again) in Atari R&D demonstrated a low-cost 80-column (or at least very close to 80-columns) off-the-shelf chip implementation using anti-aliasing through a television set. It was shown to various people who could have done something with it -- and who agreed that it looked great -- but nothing was ever done. I'm pretty sure this was in the XL days.

 

[zzip]

4 hours ago, Jetboot Jack said:

Sorry to disappoint you, but there never was one. By the time we decided to

do this game, the Atari was on its way downhill, having been beaten in the

marketplace by the C64. The last games we did for the Atari were Koronis

Rift and The Eidolon. Atari sales weren't especially good for either (I'm

pretty sure C64 sales for both were much better), and we saw the writing on

the wall. (Even though the Atari was by far my personal favorite, and I

still think it was a better machine.)

So dumb question time...

 

My copies of the Lucasfilm games were all on Flippy disks (release by Epyx)   Atari on one side, C64 on the other.    How did they know how bad Atari sales were vs C64 when it was the same sale?

[zzip]

3 hours ago, thorfdbg said:

If I recall correctly, Lucasfilms Games was founded as a cooperation between Atari and Lucasfilms, thus no miracle the games first appeared on the Atari platform and not on the C64.

Correct,  I believe Atari provided the funding and the games were supposed to be exclusive to Atari systems at least for a time.

 

The then Atari sale happened,  Atari never released the games, and they were later released by Epyx

1 hour ago, reifsnyderb said:

They never considered it to be the serious computer it was.  Not only is the 80 column issue evidence of this but the monitor port on the XL's as well.

This was typical for home computers at the time.   When they came out, monitors were not common or cheap and most consumers would want to use a TV.   80 Columns on a TV is hard to read, so you'd see 40 or sometimes 60.   But it allowed these companies to sell millions of computers cheap - $200 give or take

[FifthPlayer]

It's been a while since I read it, but the book "Commodore: A Company on the Edge" by Brian Bagnall interviews the C-64 designers, and they most certainly did look at the hardware architectures of all the machines in the market at the time they were designing the C64, including the Atari 8-bit.  There is a quote in there about them reviewing all the computers and consoles in the market with sprite hardware including the TI 99/4a, Intellivision, and A8.  That review informed the design of the C64 sprite architecture, for instance.

 

Given that the C64 designers went second, and had several models in the marketplace to learn from, it's not surprising the C64 does some things better than the A8.  What surprises me is that the C64 isn't clearly better than the A8 across the board in all aspects, given that on top of Commodore's competitive research the C64 came to market 3 years after the A8.

Edited September 29, 2022 by FifthPlayer

[+MrFish]

23 hours ago, MrFish said:

Actually, someone (I'll have to look up who again) in Atari R&D demonstrated a low-cost 80-column (or at least very close to 80-columns) off-the-shelf chip implementation using anti-aliasing through a television set. It was shown to various people who could have done something with it -- and who agreed that it looked great -- but nothing was ever done. I'm pretty sure this was in the XL days.

I looked this up yesterday, and I was incorrect about a few important details. It's from the Rich Pasco interview with Antic podcast; you can hear it starting at 19 minutes into the interview. Rich was an engineer who worked on Freddie; so, this would have been in the late XL, early XE era. He worked at PARC before coming to Atari.

 

The 80-column device was never actually developed. What Rich did was demonstrate the possibility by using a video camera, focused on an Atari book/manual, which was fed into a television set. It produced an average of over 80 characters per row, but the font was proportional. With a fixed-width font, you could probably expect somewhere in the mid 70's per row, at the same point size.

 

The intention was to produce a new chip, which would have taken several engineers a year of more (estimated by Rich). This never got any traction, for obvious reasons, especially given the time frame. Interesting approach, though, and not surprising given Rich's background.

 

As a result of my mixup, I've decided to check myself into the AtariAge Alzheimers Patient Care Center - Fantasy Hardware Devices Division -- for a couple of weeks...
 

[+slx]

13 hours ago, FifthPlayer said:

What surprises me is that the C64 isn't clearly better than the A8 across the board in all aspects, given that on top of Commodore's competitive research the C64 came to market 3 years after the A8.

Don't forget that the C64 was designed to allow a very low retail price. That certainly required some compromises. I recall an interview where it was explained that the C64 wasn't even priced to generate revenue itself but to generate revenue through sales of peripherals (which sold far better than expected). 

[+MrFish]

1 hour ago, MrFish said:

I looked this up yesterday, and I was incorrect about a few important details. It's from the Rich Pasco interview with Antic podcast; you can hear it starting at 19 minutes into the interview. Rich was an engineer who worked on Freddie; so, this would have been in the late XL, early XE era. He worked at PARC before coming to Atari.

 

The 80-column device was never actually developed. What Rich did was demonstrate the possibility by using a video camera, focused on an Atari book/manual, which was fed into a television set. It produced an average of over 80 characters per row, but the font was proportional. With a fixed-width font, you could probably expect somewhere in the mid 70's per row, at the same point size.

 

The intention was to produce a new chip, which would have taken several engineers a year of more (estimated by Rich). This never got any traction, for obvious reasons, especially given the time frame. Interesting approach, though, and not surprising given Rich's background.

 

I might have been partly confusing this with a device mentioned by Jim Tittsler (Atari 1600 Prototype, among other products), in another Antic interview. He talks about working on a grayscale adapter for 80-column text on a television that never shipped (starting at 15:28 in the interview). It's unclear if it was completed/fully-working or not, since no other details were discussed (unfortunately).

 

[thorfdbg]

The IEEE paper is unfortunately behind a paywall: "Design case history: the Commodore 64" by "Tekla S. Perry;Paul Wallich" appeared in IEEE Spectrum 1985.

 

https://ieeexplore.ieee.org/document/6370590

 

For more sprites, one would need additonal DMA channels at ANTIC, and of course additional logic at GTIA to mix the sprites to the playfield. At least at normal and narrow playfields, ANTIC still has a couple of free slots before the display starts, so that would have been feasible without changing the clock rates or the overall system design. Thus, it would have been possible to trade the display start with available P/M graphics. Actually, a similar tradeoff was possible with the follow-up chipset in the Amiga.

 

More colors would require a higher bandwidth to the RAM, and that isn't possible without a major system redesign. In the C64 design, this was possible since VID has two busses, the one to the system memory, and then another 4-bit bus to the "color RAM".

 

Anyhow, this would have required to upgrade the ANTIC design, and to upgrade the GTIA design, and apparently Atari didn't want to take the risk. Instead, they made a minimal investment to upgrade the available system RAM. This required only very minimal changes to the system without touching the custom chips.

 

[rdefabri]

1 hour ago, MrFish said:

I looked this up yesterday, and I was incorrect about a few important details. It's from the Rich Pasco interview with Antic podcast; you can hear it starting at 19 minutes into the interview. Rich was an engineer who worked on Freddie; so, this would have been in the late XL, early XE era. He worked at PARC before coming to Atari.

 

The 80-column device was never actually developed. What Rich did was demonstrate the possibility by using a video camera, focused on an Atari book/manual, which was fed into a television set. It produced an average of over 80 characters per row, but the font was proportional. With a fixed-width font, you could probably expect somewhere in the mid 70's per row, at the same point size.

 

The intention was to produce a new chip, which would have taken several engineers a year of more (estimated by Rich). This never got any traction, for obvious reasons, especially given the time frame. Interesting approach, though, and not surprising given Rich's background.

 

As a result of my mixup, I've decided to check myself into the AtariAge Alzheimers Patient Care Center - Fantasy Hardware Devices Division -- for a couple of weeks...
 

That you remembered this at all is amazing.  I'm reasonably hardcore Atari, but wasn't aware of this.