Jump to content
IGNORED

Atari Corp and Trameil with the 8-bits


kool kitty89

Recommended Posts

I haven't been around here in ages, but this is one topic that's been kicking around in my head for the last few years. I kind of came up a few times in some of the older discussions I've seen, but not really in detail and usually more sentimental or nostalgia related than historical or business. (especially after Curt and Marty's delving into the finer details on Atari History surrounding the whole transition to Atari Corp)

 

So I get the whole focus on the Atari ST line, but that's not the point I'm riding on here:

Atari corp was going to (and did) contrinue to sell, manufacture, and support the 8-bit computer line, but also needed to strealine operations in their new company due to limited funds and revenue and whatnot. It's the Atari 7800 that leaves me a bit confused compared to the rest of Tramiel's fairly pragmatic and focused business strategy. Atari Inc had already announced the 5200's discontinuation, the 2600 was well placed as the de-facto budget game machine (and dropping most of the advanced peripherals for the 2600 made sense for cost and the logistical mess during the Warner transition) and the 7800 had been announced and test marketed but not yet released with no real bond for customer support: so why not drop it and focus on the 8-bit computers for that same market slot along with the entry level computer market? (maybe keep the 600XL in production or replace it with a 32k machine more 'console like' ... maybe even go back to a cheap membrane keyboard to that end, something cheaper than the 800XL/65XE)

 

As neat as the 7800 is, from a sheer business and logistics standpoint I don't quite understand why Tramiel/Atari Corp went for it. The botched Warner management of the liquidation had complicated relations with GCC and forced litigation for payment for work on the 7800 and its initial games, so all the more incentive to just abandon it. One less unique piece of hardware to support and market and develop for (or encourage 3rd parties to develop for). The 800XL (during liquidation) and 65XE were very close to the 7800's price bracket through most of its market life when it was finally released and a more cut-down 'console' variant of the XE should have been even cheaper (or offered better margins). There's the added bonus that the 8-bit computers would be immune to Nintendo's anti-competitive market manipulation and licensing agreements, but that's not something Atari Corp management would have been privy to until after the 1986 release of the 7800 (or at least during the final stages of the build-up to its release).

 

The only practical counter-argument I can see is the existign stockpile of completed Atari 7800 hardware, components, and software was too compelling to just scrap, and the PR for the 7800's release was seen as worthwhile. (including the backwards compatibility feature ... though I'd argue that's a pretty mixed bag given incompatibility with the 8-bit computers themselves) This still seems like a hard sell for me given the logistical and marketing investment that comes into play on top of the delay and cost of clearing up the contention with GCC. And the added ability to allocate more resources towards marketing and supporting both the 8-bit computers and ST line seems like a pretty big advantage.

 

Plus, keeping the 600XL going in any form could've undercut the C64 during the price wars and give more merit to not dumping the 800XL all the way down to $100. ($150 would've plenty undercut what any sane retailer was placing the C64 at, while sticking the 600XL at $99/100 still covers cutting into the bottom end of the market)

 

 

 

 

 

And on a purely technical note: I've been pointing to the 32 or 16k configruations as those would also be the most useful for purely cart based games given the memory mapping in the 64k 8-bits; the added 32k in the 800XL/65XE being mostly blocked by I/O, BIOS, and cart ROM mapping unless I'm mistaken, plus most cart games released up to 1985 only required 16k with a handful using 32k. (in Europe with carts really unpopular, the 800XL/65XE would make more sense ... though the ST catching on was obviously the big deal there) I'm pretty sure 4-bit wide 8kB (4x16kbit) DRAMs were available in 1984/85 too so transitioning to a cheaper 2 or 4 DRAM chip configuration. (dealing with a bunch of 2kB DRAMs would be a mess and not cost effective -both board wise and with the much lower cost of 8kB density chips)

Useful reference for the DRAM market: http://phe.rockefeller.edu/LogletLab/DRAM/dram.htm(see Annual DRAM price per Mbit)

I suppose 128kbit (32kB) DRAMs got cheap fast enough to make the 8kB DRAM configuration a bit unattractive (and the late XE models did switch to 2 or 4 32kB DRAMs) so perhps a bit moot, at least in hindsight. (ie opting for 32k might have been a losing move all around; 16k was cheap enough to more than make up the difference until the early 90s, so that one makes sense in hindsight and back in 1984/85 -ie just don't drop the 600XL from the bottom end and release a 16XE to replace it)

 

And again, not knocking the 7800's merits or really arguing any of this because of personal bias one way or the other towards these systems (from a retro tech hobbiest standpoint, having the 7800 out there is certainly neat too). And I knot there's plenty of controversy over the 7800 vs 8-bit hardware on which is better for games, but honestly the two are close enough with their sets of trade-offs that it's pretty much a wash on the whole. (more confusing if you start dealing with on-cart RAM or sound expansion or that 16 vs 32k configuration on the 8-bit) The only definitive argument I can give to the 8-bit's credit (programming wise in 1985) was that it was established and old enough that some useful programming tricks were already being built on and less likely to go under-utilized than the 7800's unique MARIA graphics. (POKEY was just starting to get pushed a bit in music capability at that time too)

 

 

There's also the confusing marketing image of the XL to XE transition somewhat needlessly complicating marketing compared to the existing, known XL design and having it look more immediately distinct from the new ST line make sense too rather than the 'family look' approach. (plus the top cart slot was more convenient for a console) This is probably the least major of the arguments anyway, and if the XE case was cheaper than the XL's, all the better (plus it managed to be pretty nice and sturdy ... unlike the C64), but it still seems a bit unnecessary. (grafting a membrane keyboard onto the 600XL 'console' to cut costs seems like it'd have been pretty simple though ... if the full keyboard ended up being 'cheap enough' then all the better to compete as a /proper/ entry level computer)

 

 

 

 

 

 

 

All that said: am I making sense, or is there something else in Atari Corp's history that I'm totally missing?

Edited by kool kitty89
Link to comment
Share on other sites

Hindsight is the view being expressed - there's always the could have, should have etc possibilities that did not happen.

 

I don't think the 7800 was the dream machine it was suppose to. The lack of a good sound chip is one of many shortcomings - the other seems to be the lack of memory dedicated to the playfield?

I don't think it was the leap ahead of the 5200, compared to the 5200 over the 2600.

You need that technical leap ahead, that is apparent to anyone - not bolstered by PR.

 

Atari seemed to be comprised from the very start within it's home division.

 

I suppose it all reminds me of Sega with it's 32X add ons to the Genesis games console. It's a tough business to be in, to try satisfying these hard to please videogamers.

 

Harvey

  • Like 1
Link to comment
Share on other sites

Atari is a technology company first and foremost and there was no way Atari was going to abandon the Atari 7800. Tramiel was against all the high fees associated with the Atari 7800 but in the end, paid the price he wanted like everything else he ever bought. He was a hard negotiator and got deals done at a price that he wanted.

 

Tramiel (like Warner Bros.' Atari) was about having the latest state of the art technology as well.

 

I disagree that the Atari 7800 and the Atari 8-bits were similar enough that Atari could have abandoned the 7800. No way.

 

The Atari 7800 was the next level in 8-bit graphics. Atari certainly was known for offering great graphics in their systems and the 7800 continued that line. It was Atari's last best effort to offer a state of the art 8-bits gaming system at a price that would have given consumers a system that rivaled the graphics of many of the supposedly "technically" more powerful computers systems at the time.

 

Had Atari's promises panned out, the Atari 7800 would have been the cherry on top for their glorious 8-bit gaming systems. Atari laserdisc player, why did you become vaporware?!

 

Following your same logic on why Atari should've abandoned the Atari 7800. Atari could have very well abandoned the Atari 8-bits computer line instead. Just by having it around, it held back the Atari ST computer which was their computer future. Offering 2 different incompatible sets of computers (XE and ST lines) did not do a service to a public that was largely ignorant about how computers work back then. Owning an Atari XE or an Atari ST really did affect the experience you were going to have.

 

If Atari had dropped the Atari 2600 as well, it probably would not have been a big loss either. It added to the bottom line sure but never at the impact that it used to in the 80s or with an impact that could have helped Atari grow.

 

Perhaps if Tramiel really wanted a fresh start, he should have abandoned the old Atari completely instead as the consumers were very confused about its fate. The Warner Bros. stock crash sent consumers into a confused state about Atari's fate and whether it was alive or out of business.

 

Even today, ask someone about Atari and they will probably answer, "Atari? Didn't they go out of business since the 80s?"

 

No! Long live Atari!

Edited by TheGreatPW
Link to comment
Share on other sites

  • 2 weeks later...

Jack had been trying to release the 7800 since '84, and payment issues regarding who owed what for the rest of the development costs forced the delay. It was not brought out because of Nintendo, according to Mike Katz Nintendo wasn't even a blip on their radar. Remember, Nintendo was just starting a test marketing at that time in New York.

 

As far as why he still wanted to release it all that time remember we're talking about two completely different companies here. Atari Corp. was a completely different company from Atari Inc. Likewise, Jack's Atari Corp. staff was not Atari Inc.'s, and most of the people involved in hardware were already tapped for ST development. GCC had already developed the console for Atari Inc., and he was in contact with GCC and had actually upped the order for parts from GCC before the delay. Development costs are the most costly part of a new product, and with that done he had a ready made product. It's just good business sense.

Edited by Retro Rogue
  • Like 1
Link to comment
Share on other sites

It would've been interesting to see Atari Corp add the MARIA to the A8 computer line. It probably would've overwhelmed the bus competing for resources with the CPU and the ANTIC/GTIA but someone could've figured it out.

 

Inc and Corp both missed the opportunity to use Dual and Quad POKEYs in the A8 line along with a bunch of other things that could've improved the line but a lot of things became impossible to do from a commercial standpoint following the price war/console crash which took a lot of room out for profits.

 

Hopefully, Curt and Marty's upcoming book will answer why Corp just didn't go out and contract John Palevich and his former cohorts to get the AMY working. It's surprising to me that wouldn't have been explored while they were working with him over Dark Chambers but the lawsuit with the company they had sold the AMY tech to might've prevented that from going forward at the time even if it might've been on their radar. Whether you were on A8, ST, or the 7800, we all could've benefited from the AMY being completed and offered for sale. At that point, it probably would've been withheld from anything but the ST line and maybe some high-end games for the 7800 [cart-based instead of using a POKEY]. An add-on board like the Tweety Board for stock STs but featuring the AMY would've been a hot seller.

 

Looking back, I'm surprised nobody brought out an add-on board for the ST that would've supported POKEYs. Even the most pro-ST owners back then hated the stock Yamaha chip.

Edited by Lynxpro
Link to comment
Share on other sites

  • 3 weeks later...

I was at CommVEx this past summer and saw Leonard Tramiel while he was there, answering questions. The 7800 was one of the questions that came up.

 

He said that in 1984-85, there was a three-way pissing match between Commodore, Atari and Amiga; because of all of the changing of hands between the three companies, they were all suing each other to get things settled. While this was going on, it caused delays in the release of some products, like the 7800.

 

As far as the NES is concerned, he said that Nintendo had approached them about marketing the NES as a new Atari system is the U.S., but Atari was focusing a lot of their time and resources into developing the ST as quickly as possible, so they declined for that reason.

Link to comment
Share on other sites

It would've been interesting to see Atari Corp add the MARIA to the A8 computer line. It probably would've overwhelmed the bus competing for resources with the CPU and the ANTIC/GTIA but someone could've figured it out.

 

Inc and Corp both missed the opportunity to use Dual and Quad POKEYs in the A8 line along with a bunch of other things that could've improved the line but a lot of things became impossible to do from a commercial standpoint following the price war/console crash which took a lot of room out for profits.

 

Maria can't really be used in the A8 environment. It requires fast SRAM to operate which means it's needs its own bus and memory. It was practical for the 7800 because the system was completely cartridge based and didn't need much RAM.

 

The A8 was pretty much considered to be a legacy product once the ST was announced, so the only improvements we got were a few cost-reduced products that Warner had designed (like the 130XE and XF551). Jack certainly wasn't going to spend engineering dollars on keeping the 8-bits viable.

Link to comment
Share on other sites

Jack had been trying to release the 7800 since '84, and payment issues regarding who owed what for the rest of the development costs forced the delay. It was not brought out because of Nintendo, according to Mike Katz Nintendo wasn't even a blip on their radar. Remember, Nintendo was just starting a test marketing at that time in New York.

 

As far as why he still wanted to release it all that time remember we're talking about two completely different companies here. Atari Corp. was a completely different company from Atari Inc. Likewise, Jack's Atari Corp. staff was not Atari Inc.'s, and most of the people involved in hardware were already tapped for ST development. GCC had already developed the console for Atari Inc., and he was in contact with GCC and had actually upped the order for parts from GCC before the delay. Development costs are the most costly part of a new product, and with that done he had a ready made product. It's just good business sense.

 

I understand all of this and we've been over it several times before a few years back. (and a handful of added details in your book) It does make sense for the most part. Honestly, from Tramiel's standpoint in 1984 some of what I suggested likely was being considered and they just leaned towards releasing the 7800 being the better bet.

 

That said, I'm not sure the cancellation of the 600XL was totally merited or the 800XL price-slashing for that matter, but those two points are really one in the same. (having the 600XL around as the bargain basement system with the 800XL as the ... slashed but not QUITE so slashed upper teir product seems like it would have made more sense; there's a big range between the $99 the 800XL was slashed to -or whatever its European equivalent was- and the $200 the C64 commonly retailed for in late '84) Given how RAM pricing was going, I'm not sure a 16XE would have made sense or not. (outside of my 'instead of 7800' suggestion of a really basic 16k+membrane keyboard consolized 600XL -not 5200 style consolized, just repackaged)

 

The 600XL should have made a very good (superior) competitor to the Commodore 16 as well. (with a far more established software and software development infrastructure) I suppose my cheap consolized 600XL/16XE idea is a bit more like the C116 for that matter, chicklet instead of membrane but close enough. The 800XL certainly made a fine direct competitor to the Plus/4 on technical grounds and market positioning pretty much superior in all areas unlike the more mixed situation of the C64. (faster CPU, liner bitmap graphic modes, much faster disk drive with more standard formatting, better build quality, no overheating problems, better keyboard in most cases -for non-gaming purposes, the 800XL was technically superior across the board to the C64)

 

I suppose none of that is remotely mutually exclusive with 7800 production/distribution either and given the 7800 did end up profitiable in fairly short order after its introduction, it shouldn't have been a drag on resources in general. (there's probably a better argument for pulling more 2600 game development and marketing funds for 7800 and A8 games)

 

The XEGS is more what confuses me. THAT is the situation I was working backwards from to set up the premise of this thread. The XEGS in 1984/85 with no 7800 released might have made sense, but in 1987 it really, really didn't. I've seen Leonard Tramel's explanation regarding 'doing the 5200 right' which makes some sense in a vaccuum or as a hypothetical (the Atari 600 in place of the 5200 in 1982 would have made perfect sense as the 400 was being price-cut and phased out) but with their existing product lineup in 1986 let alone 1987 the XEGS really didn't make sense and made far less sense at the $200 price point (when the 7800 was pushing around $80 and 65XE itself $100).

 

A cut-down ST derived game system would have made sense at the time, especially if they got the blitter into volume production (hardware scrolling SHIFTER would be nicer, but the BLiTTER would at least offset a lot of the RAM needs for pre-shifted byte/word-alligned graphics and make it realistic to dramatically cut back on RAM in the system, though games would need heavy reprogramming ... maybe not a bad thing to get BLiTTER support popularized for ST computer ports too, especially if the BLiTTER had been added to the STf/fm as standard rather than waiting all the way until the STe -again, a scrolling SHIFTER would be a big deal too)

 

 

But back to the main point: I agree here, I think on the whole Tramiel made the right decision in releasing the 7800, those other areas still leave more questions, but the 7800 makes good sense all around. (getting it to Europe earlier would have probably helped a lot software-wise given the cheap 3rd party software model would have been excellent in that region, Atari was just very late to market there) There's other things like looking for lower cost sound chip add-ons too (unless they had a surplus of POKEYs, namely the SN76489s or possibly various AY-1910 derivatives should have been cheaper and very useful -mostly the low-pincount versions of the latter and Yamaha's offerings -aside from Ballblazer that predated Atari Corp)

 

 

 

 

 

 

 

 

Inc and Corp both missed the opportunity to use Dual and Quad POKEYs in the A8 line along with a bunch of other things that could've improved the line but a lot of things became impossible to do from a commercial standpoint following the price war/console crash which took a lot of room out for profits.

 

 

Looking back, I'm surprised nobody brought out an add-on board for the ST that would've supported POKEYs. Even the most pro-ST owners back then hated the stock Yamaha chip.

 

Throwing in a second POKEY might have been useful for an alternate 65XE-M (after Amy was abandoned). 8 sound channels, up to 4 pairs of 16-bit channels, and an added serial port that could potentially be used for a MIDI interface. (like a lower-end companion to the ST family for musical purposes)

 

The AY-8910 is actually a pretty decent sound chip with a lot of useful feaures for the time. Aside from the 16 I/O ports it has, it's got 12-bit frequency control for its three channels (much better note quality than the SN76489 or un-paired POKEY channels), an ASDR envelope, and the 5-bit logarithmic volume was nice. (and useful for 8+ bit approximated PCM playback, though POKEY was good for true linear 4-bit PCM playback -good for a lot of low bitrate stuff, and honestly probably a lot of what ST games managed for in-game samples) The Yamaha/General Instruments chip may also have been cheaper on the mass market than Atari's POKEY production sources. (and they lacked vertical integration to push the same level of bias towards custom chips as CBM had) Also remember you waste a lot more CPU time doing interrupt-heavy stuff on the 68k, so some timer based POKEY tricks useful on the A8 would be proportionally more taxing on the ST.

 

I had a thread discussing useful configruations of POKEYs in the ST design and it generally came down to it not being worth it overall. (the serial ports on 2 POKEYs would be useful, but unless you hacked things a bit, the keyboard and POT lines would be pretty useless -cutting down to a 24-pin package was also suggested -drop the 8 key lines and 8 POT lines; best case I recall was dual 24-pin POKEYs plus a PIA to add the pair of 8-bit I/O ports in place of the ST's ACIAs and YM2149F)

 

Besides that, including a memory mapped 8-bit (or better) DAC would have been a better cheap/simple added feature for the ST and was a far more common add-on too (including parallel port COVOX style DACs). A DMA sound circuit is really what it needed, and it took too long to get that with the STe too. (even something really basic like the Mac had would have been useful, or even if it shared the existing DMA chip's memory cycles and PCM audio was mutually exclusive with disk access)

 

Throwing in a cheap, low-end Yamaha FM chip might have been nice too, but they didn't really get to 'cheap' FM chips until the late 80s. (the really cut-down YM2413 seems up the ST's alley but I don't think that was available in volume until 1987, so a MEGA ST possibility or STe but not ST) The YM2203 was probably more costly than Atari was willing to go for in 1985 and probably more costly than adding a DAC and making any necessary changes to the DMA controller to make that work. (it might already be useful ... all it needs to do is be able to read and write 8-bit words in the kHz range without audible interruption) Note the YM2203 is compatible with the YM2149 including its I/O ports, but I'm fairly sure the YM2149+YM2413 would have been cheaper in 1987 or later. (more channels and 2 chips but the YM2149 was a very common part with competition producing compatible products while the YM2201 featured 3 4-op FM channels AND required a companion floating point -serial- DAC compared to the YM2413's 2-op FM synthesis and other limitations -albeit 9 channels- with built-in time-multiplexed DAC arrangement -it's also a lower cost version of the OPL chip used in PC Adlib and soundblaster cards though given how some of the pre-programmed instruments sound /better/ than a lot of PC games instrument sets, I'm not sure the cost cutting was that big of a loss -Adlib and Creative Labs may not have provided the best documentation for music either)

 

But I didn't intend to get that far into the ST discussion and the last thing I'd say there is MARIA really wouldn't have made sense as part of the ST. Packed/chunky pixel SHIFTER modes (including a 160 pixel wide 8bpp mode) and hardware scroll registers would have been exceptionally useful, but throwing MARIA in there (or even the PANTHER VDP -which is very much a 32-bit MARIA) into the ST would have made no sense at all.

 

 

 

 

 

 

Maria can't really be used in the A8 environment. It requires fast SRAM to operate which means it's needs its own bus and memory. It was practical for the 7800 because the system was completely cartridge based and didn't need much RAM.

 

The A8 was pretty much considered to be a legacy product once the ST was announced, so the only improvements we got were a few cost-reduced products that Warner had designed (like the 130XE and XF551). Jack certainly wasn't going to spend engineering dollars on keeping the 8-bits viable.

 

Indeed, though I don't think speed is the issue so much as lack of DRAM control and interface logic and lack of buffers or latches on the bus to accomodate MARIA. DRAMs used in the 5200 and A8 (at least by 1984) were fast enough to manage access and cycle times akin to what MARIA needed, but it would have delayed the design and to be fair it was GCC's biggest design undertaking at that time (and their first LSI chip). Atari's own R&D probably could have managed something signigiantly more impressive and/or something more easily compatible but Atari management didn't seem to be putting an emphasis on that (GCC+Warner didn't consult Atari Inc until the 7800 was well into development from what I understand). The 3200/Sylvia design using ANTIC mated with an upgraded Super-TIA (STIA) from 1980/81 was very much in the design ethic of the 7800 but was abandoned in favor of the 5200 for reasons only vaguely explained in Curt and Marty's book. (seems to be a complex mix of management bureaucracy and technical issues but it seems much more the former)

 

In any case, technically it should be possible to implement a middleman chip of sorts (like the ST's MMU) to handle DRAM control and interface to MARIA but it doesn't seem worthwhile given everything else going on at the time.

 

Atari Inc also had CGIA prototyped (ANTIC+GTIA combined into a single chip using a 68-pin LLC) that didn't end up carrying over into production with Atari Corp for whatever reason. Not an upgrade, but a useful cost savings for the A8 chipset. (Atari Corp eventually used the 6507+RIOT+TIA on a chip, so I'm not sure why CGIA didn't happen unless it was buggier and/or the documentation 'walked off' during Warner's horribly executed liquidation proceedings)

 

 

 

 

 

I was at CommVEx this past summer and saw Leonard Tramiel while he was there, answering questions. The 7800 was one of the questions that came up.

 

He said that in 1984-85, there was a three-way pissing match between Commodore, Atari and Amiga; because of all of the changing of hands between the three companies, they were all suing each other to get things settled. While this was going on, it caused delays in the release of some products, like the 7800.

 

As far as the NES is concerned, he said that Nintendo had approached them about marketing the NES as a new Atari system is the U.S., but Atari was focusing a lot of their time and resources into developing the ST as quickly as possible, so they declined for that reason.

From the research Curt and Marty have done on the subject, including intervies and cross-checking with other information and interviewers, I'm not sure this is entirely right. (not saying Leonard didn't say those things, but he might be mixing up a few memories)

 

From what I understand, Atari Inc (not Corp) was approached in 1983 to license-distrubute the Famicom BEFORE it had entered production in Japan. This was under Ray Kassar I believe (before Morgan took over) and the offer with both highly unfavorable in terms of cost and restructions and extremely risky due to Nintendo being totally unproven. Atari had wanted to string Nitendo along longer than they ended up doing (delaying Nintendo's approach to other potential distributors) but that all ended up being a bit moot due to Atari's management changes, liquidation, the market crash and overall mess going on pretty much up until things stabilized in late 1985. (Nintendo going it alone -or partnering with Worlds of Wonder depending on how you look at it- and pushing hard in 1986 changed things ... that and their near-complete monopoly on Japanese arcade game licenses and most computer to console ports ... and hoards of other anti-competitive business tactics that make Warner-Atari's golden-age strong-arming look like kiddie stuff -American Courts favored independent developers/publishers a hell of a lot more than Japanese courts did, there was no Activision revolution there)

 

Sega DID offer Atari Corp distribution rights to the Megadrive in late 1988 (just before Jack Tramiel retired). Mike Katz was in favor of the partnership but Dave Roden (Sega of America Board Chairman) and Jack Tramiel couldn't come to agree on the terms so it fell through. (Curt and Marty's research didn't explicly point to it, but I'd expect the deal being a North American exclusive contract made for some conflicts of interest given Sega and Atari Corp both had stronger market shares in Europe and it'd mean a rather complicated mess of Atari-Sega relations and competition in both markets -in practical terms the ST and Master System DID compete with eachother, especially in the UK and to some extent the ST and Mega Drive also competed with eachother)

 

In hindsight (or Mike Katz's own view in 1988), the complicated partnership might have been the better idea especially if Katz stayed around (Sam Tramiel probably would have ruined everything if Katz was gone). Mike Katz and Jack Tramel leaving had a huge negative impact on post 1988 Atari Corp -also remember Katz joined Sega of America in 1989 and was instrumental in establishing the Genesis before Tom Kalaniske took over in 1991. (Mike and Jack also fit wonderfully together in terms of business sense, ethics and just general non-nonsense practicality and at times being too honest and not charismatic enough -the main fact Katz cites for his leaving Sega, being unable to placate Japanese management or stroke their ego satisfactorily -not 'warm and fuzzy enough' per his own words) Sega would have almost certainly been dragged down with Atari had the deal gone through AND the bad shift in management occurred as it did historically. (between the DRAM price inflation and stagnation of 1988 and Jack Tramiel leaving followed by Mike Katz a few months later, the winter of 1988-1989 marked the beginning of Atari Corp's downward spiral)

Edited by kool kitty89
  • Like 2
Link to comment
Share on other sites

I was at CommVEx this past summer and saw Leonard Tramiel while he was there, answering questions. The 7800 was one of the questions that came up.

 

He said that in 1984-85, there was a three-way pissing match between Commodore, Atari and Amiga; because of all of the changing of hands between the three companies, they were all suing each other to get things settled. While this was going on, it caused delays in the release of some products, like the 7800.

 

As far as the NES is concerned, he said that Nintendo had approached them about marketing the NES as a new Atari system is the U.S., but Atari was focusing a lot of their time and resources into developing the ST as quickly as possible, so they declined for that reason.

 

That all sounds a bit off. I did confer with Leonard on what was talked about at CommVEx and there appears to be some confusion/misremembering above about the conversation. From every person in the know that would have been involved in such conversations that I've discussed this with (including Leonard), the only thing Nintendo approached them about was wanting money from them for the transfer of Nintendo licensed game stock to Atari Corp. They considered an assets purchase a sale, so they wanted the promised revenue payment based on the original agreement with Atari Inc. Likewise, Nintendo had approached Atari Inc. about doing the non-Japan version of the Famicom, which was a completely different company. Perhaps that's where the confusion lies.

 

Leonard himself was VP of Sofware Development, which would have put him knee deep in the development of RBP throughout that second half of '84. Not in any negotiations with Nintendo on the NES.

 

As for the Commodore/Amiga/Atari lawsuits, those had zero bearing on the 7800 and I can honestly say you're confusing/running together two different stories. The 7800 delay was entirely caused over on and off again negotiations between Warner/GCC/Atari Corp. over who owed GCC for the payments on the development of MARIA and the ten launch titles. GCC's contract was with Warner so they felt Warner owed them, and Warner felt Jack owed them if he wanted to release the product. An agreement was finally reached in spring '85 on MARIA and by early summer '85 on the launch titles. The lawsuit between Commodore and Atari Corp. and the countersuit by Atari Corp. at Amiga had no bearing on the Warner/GCC/Atari Corp. negotiations. The only thing those lawsuits had immediate bearing on was computer development, because Commodore had filed and then renewed injunctions barring Shiraz and several other ex-Commodore engineers (who all together formed the basis for Jack's hardware team) from doing any computer work for Jack. That shut down all computer development (RBP/ST and 8-bits) across the entire month of July until it was lifted in August. That's when wire wrapping on the RBP prototype immediately started, and plans for 8-bit development started up again as well. None of that would have any bearing on an already developed and produced game console.

 

Jack Tramiel leaving followed by Mike Katz a few months later, the winter of 1988-1989 marked the beginning of Atari Corp's downward spiral)

To clarify, Jack didn't "leave" Atari Corp. He was still there as chairman of the board. Rather, he had turned over day to day operations as CEO to Sam, who also retained his President position.

Edited by Retro Rogue
  • Like 1
Link to comment
Share on other sites

The Atari Home Computers division made very serious errors soon after the availability of the Atari 400 800 computers were on general sale - this is with not keeping important staff on, after the launch.

But this is the nature of the business of the home videogame market/etc. Nothing is ever stable or predetermined at the business end. The home consumer is lulled into a false sense of security and permanentability - when they take that piece of modern engineered hardware home, like a new born baby -it's life span was expected to be of the order of 5 to 8 years maximum - and any further years past that, is somewhat of a bonus.

Your new pet is not expected to live past 10 years - unless it's something like a tortoise - but then you know how incredibly slow they are.

 

Harvey

Link to comment
Share on other sites

The Atari Home Computers division made very serious errors soon after the availability of the Atari 400 800 computers were on general sale - this is with not keeping important staff on, after the launch.

But this is the nature of the business of the home videogame market/etc. Nothing is ever stable or predetermined at the business end. The home consumer is lulled into a false sense of security and permanentability - when they take that piece of modern engineered hardware home, like a new born baby -it's life span was expected to be of the order of 5 to 8 years maximum - and any further years past that, is somewhat of a bonus.

Your new pet is not expected to live past 10 years - unless it's something like a tortoise - but then you know how incredibly slow they are.

 

Harvey

 

If you're referring to the Amiga team, that's not quite right either. Many of the engineered intimately involved with the A400/800 chipset and system design remained at Atari Inc much later than that and some up until the liquidation of the company in 1984. (THAT is when a ton of valuable engineering staff was lost along with a fair amount of engineering materials -or at least any sort of organized access to the documentation/materials- due to the shoddy organization of the company's liquidation and assets transfer by Warner -properly organized and delegated proceedings through Atari Inc management, especially then-president James Morgan, should have allowed Tramiel/Atari Corp to make much more efficient use of Atari's intellectual assets and likely avoid some of the legal issues with disgruntled former staff as well -and even if Jack didn't keep many or any more of Atari inc staff on long-term, there'd have been more incentive for short term retention to key management and engineers in the interim)

 

But on the A8, Jay Miner didn't actually design any of the chips in that system he did TIA's design and I believe some concept work for the A8 chipset, but wasn't directly involved with the final design. I'm not sure if any of the other Amiga team members worked on the A8 chips either, but it IS very notable that Atari Inc had supported funded and licensed the Amiga chipset (along with several other companies) prior to Commodore's buyout (which was pretty much illegal given Amiga's existing licensees) and Amiga Inc had in fact lied to Atari Inc in early 1984 claiming their design didn't work and couldn't supply prototype chips for Atari testing.

 

I believe Atari's Amiga project was known as MICKEY and had plans as both a game and computer system. There were plans for a BSD UNIX derived OS to be employed on that project as well, I believe. (though it might not have been so useful given the Amiga's lack of protected memory -the ST had the same problem which led to Atari Corp dropping initial considerations for a UNIX derivative they'd attained a licensed for)

 

Also note that Atari Inc had plans for several different advanced 16-bit (68000 or possibly NS16032 based) machines aside from the A8/x86 hybrid systems being considered. RAINBOW, SILVER, and GOLD were code-names involved in those chipset(s) but I forget the details Curt and Marty posted on the project. They might have been more potent than the Amiga in some respects but again, the details (and my memory) are lacking here, they also might not have been as cost effective as the Amiga. I may also be mistaken above on UNIX and it may have been these systems and not MICKEY being targeted. (if they had a useful MMU of some sort, that would indeed make it more practical)

 

 

Atari had a ton of unused advanced (and not so advanced) hardware projects in their R&D division, but most of it didn't actually go anywhere. (tons of investment for relatively little actual product development -one of the many areas James Morgan was trying to reform just before Warner liquidated Atari Inc in mid 1984)

 

 

 

 

Honestly, losing the A8 chip designers (if that even occurred prior to the liquidation) would have had a bigger impact on complicating development of backwards compatible A8 derivatives than all-new incompatible designs. (that goes both for cost-reduction of the existing designs and chips -die shrinks and further integration- and full successors with added features and hardware -like a video chip with double the bandwidth, color depth, and a full 16 color registers) Given that Atari had prototyped the integrated CGIA (GTIA+ANTIC) chip prior to the liquidation, it seems they were making progress in that area anyway. (also note that the XL compatibility issues is down to memory map and OS changes, not anything related to the chips themselves -there were/are also more compatible memory mapping schemes possible, but they'd compromise some of the added OS features and possibly be slightly more expensive if requiring new I/O lines and not repurposing PIA's ports )

 

Though if I may jump into hypothetical hardware for a moment: a system based on a 2x speed (or bandwidth) integrated CGIA derivative with full A8 compatibility and the 6502 mainly relegated to I/O handling with the new main CPU a 68000 (somewhat like the Tandy Model 16 did with Z80+68k) could have ended up being a pretty useful system. (IBM CGA quality resolutions with much better color options and hardware sprites -at double the number, width, or color depth of A8 sprites- hardware scrolling and overscan capabilities and a slave 6502 to use as a sound/video coprocessor, possibly a second POKEY as well) The 68000 already has a 6800/6502 compatible I/O bus protocol, so interfacing should have been simpler than with Z80 arrangements. ( a lowres 96/80 pixel wide 8bpp 256 color mode would be neat too and chunky pixels too, nice singly byte chunks ... if only the ST had used packed pixels instead of bitplanes)

Edited by kool kitty89
Link to comment
Share on other sites

...

But on the A8, Jay Miner didn't actually design any of the chips in that system he did TIA's design and I believe some concept work for the A8 chipset, but wasn't directly involved with the final design. I'm not sure if any of the other Amiga team members worked on the A8 chips either....

My memory says that Jay Miner did work on the chip design for the Atari 400/800 computers - and looking at this Wikipedia info, it seems to confirm this?

 

"He moved to Atari in the late 1970s. One of his first successes was to combine an entire breadboard of components into a single chip, known as the TIA. The TIA was the display hardware for the Atari 2600, which would go on to sell millions. After working on the TIA he headed up the design of the follow-on chip set known as ANTIC and CTIA for which he holds a patent.[2] These chips would be used for the Atari 8-bit family of home computers and the Atari 5200 video game system.

 

I, myself did not think too much of the XL and XE upgrade lineage because the graphics hardware, etc remained the same - which dates back to it's 1978? architecture. It really needed something better to outclass the C64 with its 16? multi-coloured sprites.

Unlike coin-op videogames - they simply stuck in? more and more chips to handle more and more sprites.

However the XL/XE computers were of course cheaper to produce and could therefore be sold at a low price - but you couldn't really add the selling point - that it's graphic architecture dates back to 1978/1979.

 

Harvey

Link to comment
Share on other sites

Oh and I should comment on James Morgan's (at least apparent) decision to shelve most of Atari Inc's new computer projects but end up following up with the Amiga developments makes a lot of sense: cut down heavily on internal R&D spending, refocus efforts towards games development (per his primary business model) and continue development plans for console, computer, and arcade applications of the Amiga chipset. The fact the contract limited it to a game machine only in 1984 (computer with keyboard and 128kB max RAM in 1985 and unrestricted from 1986 onward, I believe) fit the plans rather well too while retaining the ability to promise a follow-on computer expansion unit to consumers. It would have been a very expensive game machine for the home market in 1984, certainly niche if it did launch, and likely with limited RAM (minimum possibly using 16kx4-bit DRAMs -like the 600XL- would be 32k, using 32kx1bit DRAMs -like the 800XL or C64- would mean 128k minimum) and possibly more useful as an arcade machine for that first year. (128k in 1985 is more realistic, though still high-end for a console, and the custom chips -and CPU- are going to be big considerations but potentially a fraction of the 520ST's price, let alone Amiga 1000; and remember Atari Corp had announced the -cancelled- 130ST at $299 that year)

 

And yes, a 32k Amiga would heavily limit available Amiga video modes, but with the ROM sizes practical at the time, you'd probably be using mostly 2bpp (3 or 4-color) graphics anyway, maybe some 3-bit stuff, probably not 4-bit. And you could use screen sizes smaller than 320x200. (though 64k would probably be the more reasonable base model)

 

In any case, more than enough potential to make halting any competitive internal projects a practical trade-off at the time.

 

 

 

 

The fact Amiga had licensed to multiple investors making the potential beginnings of a licensed hardware standard would have been a benefit or drawback depending on Morgan's business model. (not sure if his ideas tended more towards forcing proprietary design and such or profits through hardware sales -if it was stronger on the software end, the potential ability to more easily port Atari software to other Amiga-standard machines would certainly be a boon)

Link to comment
Share on other sites

My memory says that Jay Miner did work on the chip design for the Atari 400/800 computers - and looking at this Wikipedia info, it seems to confirm this?

 

"He moved to Atari in the late 1970s. One of his first successes was to combine an entire breadboard of components into a single chip, known as the TIA. The TIA was the display hardware for the Atari 2600, which would go on to sell millions. After working on the TIA he headed up the design of the follow-on chip set known as ANTIC and CTIA for which he holds a patent.[2] These chips would be used for the Atari 8-bit family of home computers and the Atari 5200 video game system.

I was mistaken, he was the technical director of the CTIA/GTIA design at least, but George McLeod and Steve Smith did the actual circuit design work.

 

In any case, it had no real bearing on the A8 design languishing or lack of something newer, more powerful (and more expensive). Atari lagged a bit on cost-reducing the A8 as it was and lagged in filling the gap between the 400 and 800 in price/functionality. (unless you resorted to 3rd party upgrades -the keyboard was a huge issue, much less than the 32k RAM limit on the 400) Atari did TONS of R&D work, a lot of impressive stuff too, but it lacked direction and focus as much as or more than the business end of things. With more focus, they should have had the Atari 3200 working and on the market by 1981 alongside a cost-reduced Atari 600-ish replacement for the 400 AND progress towards a new 16-bit computer design. (or possibly a fast 8-bit one ... if they could get decent yields on their licensed 6502C at 3.58 MHz, that'd be pretty significant especially if you throw in some local, zero wait state SRAM masking Zero page and then some)

 

I, myself did not think too much of the XL and XE upgrade lineage because the graphics hardware, etc remained the same - which dates back to it's 1978? architecture. It really needed something better to outclass the C64 with its 16? multi-coloured sprites.

Unlike coin-op videogames - they simply stuck in? more and more chips to handle more and more sprites.

However the XL/XE computers were of course cheaper to produce and could therefore be sold at a low price - but you couldn't really add the selling point - that it's graphic architecture dates back to 1978/1979.

 

The A8 hardware was older, but aside from the C64 really wasn't that superior, ESPECIALLY as a computer. As a game machine it still had build quality and overheating issues (depending on revision and climate) and horrendously slow disk load times (tape loads more modestly so -fastloaders aside). As a general purpose computer, the 800, 1200XL, or 800XL was superior: faster CPU, faster disk drive/interface, more standard/compatible disk format, better build quality, better technical support. Hell, the chipset was even more cost effective in some respects (chips ran cooler and all ran of +5V single rail power and I/O) and even without further component integration it was only Atari's higher quality standards, profit margins, and lack of in-house IC manufacturing (Synertek would have been a good investment there -Atari already used them to second-source most parts) that gave CBM its advantages. (a more advanced system would be at an even BIGGER disadvantage using the same business model -marketing and price point is where the C64 won, plain and simple)

 

In terms of graphics, the A8 had some useful aspects too: the linear (true) bitmap modes, 256 color palette, 9 or 16 low res color modes. And like the C64 (but unlike the Colecovision, MSX, SG-1000, or other TMS9928 based systems) you had hardware scrolling and 4 colors per character. (with that 5th extra color select option -shame they couldn't share the sprite colors and swap out 3 or 4 colors though, make a 7 or 8 color mode rather than 5 -DLIs aside) The C64's greater popularity drove much more aggressive software development than the 64k era A8 did, so the disparity in graphics and sound quality is far greater than the technical limitations. (if the A8 had been nearly as popular, you'd have seen far, far more impressive games taking advantage of the added memory and programming experience -and game design experience -most pre 1984 games were NOT using anywhere near the full graphics potential of the system and many used lower resolution and/or color modes to work within more limited ROM and RAM limits, not to even start on effective use of programming tricks and more advanced game design ideas developing in the mid/late 1980s)

Most late 80s and early 90s European A8 games were pretty lazy ports too compared to their C64 ... or even Spectrum and Amstrad counterparts.

 

Additionally, the C64 also has 8 sprites, not 16. (each of which can be 1 color and 24 pixels wide or 3 colors and 12 pixels wide -or rather 2 global sprite colors + one unique per sprite)

 

 

On the computer end, Commodore's chroma/luma monitor was nice too. CTIA/GTIA output similar separate chroma/luma signals too with a compatible pinout on the video connector even (RF only models aside -for 5-pin DIN monitor cables at least ... ie most common ones that use composite video as chroma). Odd that Atari didn't produce a nice monitor to compliment their high-end 800 or 1200XL before Commodore did ... or just ... at all. (you had to use a commodore monitor to get the best from your Atari ... )

Link to comment
Share on other sites

With the popularity of the C64 and it's vast numbers sold - creating truly a mass market - this of course provides a monetary reward incentive that pushes the creativity with the software development - which you see present on the spectacular C64 games that stand out from the rest as they explore the limits? of this particular hardware.

 

If you had the same number of people developing what they can on the A8 hardware there can be ways around the limitations of the A8 hardware to a certain degree and also exploit what is not possible on other hardware.

What was found to be possible on the 2600 - should have also been used on the Atari 400/800 etc hardware. While I do not understand the complexity and details of what these kind of techniques entail - what they produce on screen is immediately apparent to anyone seeing it in action. Though it requires a programmer who has a better understanding of how the hardware works and how to work around it's limitations. The limits are still there - but you can fuzz the limits to some degree like some magician.

 

I am always puzzled by the 7800 design - to me it never appeared to be the wonder machine it was suppose to be - while it does provide many more multi-coloured sprites - it's playfield (background) graphics seem to be below the A8 standard. I can't seem to find a 7800 game which shows this is not the case at all? Not including a decent sound chip seems to confirm how little they (Atari) wanted to spend on it's chips? Thereby not giving this system hardware a decent chance at living long and prosper.

They certainly did not have the foresight to see that extensive playfield graphics would be the norm at that time - today I'd guess that 7800 developers don't have this shortage of memory problem anymore? Like with the A8 hardware today - the limit is whatever you want to work with, if you take advantage of bank switching via flashcart.

 

Harvey

Edited by kiwilove
Link to comment
Share on other sites

If you had the same number of people developing what they can on the A8 hardware there can be ways around the limitations of the A8 hardware to a certain degree and also exploit what is not possible on other hardware.

What was found to be possible on the 2600 - should have also been used on the Atari 400/800 etc hardware. While I do not understand the complexity and details of what these kind of techniques entail - what they produce on screen is immediately apparent to anyone seeing it in action. Though it requires a programmer who has a better understanding of how the hardware works and how to work around it's limitations. The limits are still there - but you can fuzz the limits to some degree like some magician.

True, but most of those really trick-heavy 2600 games used a LOT of ROM and sometimes RAM expansion. (there's also some tricks you can do with a lot less ROM if you have a little more RAM) The same goes for the A8 both hypothetically and in reality for a lot of effects and even just making the most of the 'standard' features. (given the memory used by the full res full color modes with multiple character sets and animated sprites)

 

Porting C64 games to the A8 would be a lot worse than having unique games optimized for the machine OR games co-developed or heavily redesigned to cater to the pros and cons of the system. Ports of some NES games might actually be easier given how 8-pixel wide sprites were more common there and GTIA allows attaching sprites somewhat like the Amiga to make 3-color (2-bit) sprites with 2 per scanline rather than four. (I believe Crownland does this, hence the colorful sprites and flicker when more than 2 are on the same line -very good use of sprite multiplexing in that game, fairly simple level design but not bad and it shows off more possibilities for games working around similar optimizations)

 

With the A8 it's generally not worth using CPU driven GTIA tweaks/tricks (racing the beam a la TIA) and best to rely on ANTIC's DMA functionality with the possible exception of sprite handling (which I'm pretty sure is mostly CPU heavy normally). CPU intervention through display list interrupts are obviously useful and a lot easier to do than on the 2600 (interrupts rather than meticulously hand-timed code) for palette swapping effects on the playfield and sprites.

 

 

 

I am always puzzled by the 7800 design - to me it never appeared to be the wonder machine it was suppose to be - while it does provide many more multi-coloured sprites - it's playfield (background) graphics seem to be below the A8 standard. I can't seem to find a 7800 game which shows this is not the case at all? Not including a decent sound chip seems to confirm how little they (Atari) wanted to spend on it's chips? Thereby not giving this system hardware a decent chance at living long and prosper.

They certainly did not have the foresight to see that extensive playfield graphics would be the norm at that time - today I'd guess that 7800 developers don't have this shortage of memory problem anymore? Like with the A8 hardware today - the limit is whatever you want to work with, if you take advantage of bank switching via flashcart.

 

The 7800 has no playfield, only sprites. It also lacks a fixed number of sprites as they're list driven and not hard-coded (more akin to what the Jaguar's Object Processor does and some arcade machines). You can simulate a playfield by using sprites or, ideally make a more animated, detailed background entirely out of sprites rather than just chaining sprites together like a character map. (LOTS of potential for parallax effects, though you hit a limit with memory bandwidth at some point and sprite vs BG detail is always traded off AND CPU time -outside of vblank- gets eaten up with MARIA saturating memory access -with MARIA maxed out, the CPU will be waiting more than half the time)

I'm pretty sure RIOT is mapped to the same bus as MARIA, so using its 128 bytes as local memory for the CPU isn't so useful either. (if MARIA didn't force wait-states for RIOT accesses, that would be really useful for zero-page operations -the 6502 has few registers and needs a lot more memory access than something like a Z80 to do the same things efficiently, and it does register-like operations in its first 256 bytes of address space, so the 128 bytes on RIOT could be pretty useful for that)

 

Also, even with lots of ROM, limited RAM is still an issue (pending added expansion there) and as it was a few games used 8 or 16 kB of SRAM on-cart to work around limitations of the onboard 4k. (Summer and Winter games carts actually used 32kB SRAM chips but only mapped 16 kB to the system -using 2 8k chips would have been more costly at the time and would have required an odd/longer PCB and cart case -this was 1987 or 1988, probably '87 at least for design/manufacturing decisions being made) One of the workarounds for the lack of palyfield is to make one BIG 'sprite' (a 4 color framebuffer, basically) and use the CPU to software-blit onto that. That also simplifies the display-list-lists and reduces MARIA bandwidth used (compared to filling the background with sprites) but it requires RAM expansion, especially if you want double buffering. (160x192 with 4 colors -2-bits per pixel- takes 7.5 kB, so you'd use most of the 16kB up with double-buffering -or even just partial added buffering for a better looping scroll area to have off-screen)

 

The framebuffer 'trick' is also one of the workarounds that makes cross-platform development for the 7800 a lot easier. (the 7800's list-based sprite system was unique among consoles and computers; no native character modes, not X/Y positioned hardware sprites like the C64, NES, or SMS, and a lot of bandwidth and added programming to try to approximate all of that) Using a framebuffer background made it a lot easier, partially given computer platforms often used framebuffers and partially because software blitted 'characters' are easier to approximate than tiling the whole background with sprites. (it's also NOT like NeoGeo sprites, those are more conventional just big and plentiful)

 

 

 

The 7800 probably could have used 16 kB of DRAM without cost drawbacks if they'd had more design time, but as it was it was rushed for the planned 1984 release (and to be fair, a good effort for the level of experience the GCC engineers had). That rushed nature is also why MARIA is a separate chip and not more carefully designed to integrate TIA functionality with minimal wasted board and chip space. (had the 7800 been DESIGNED to be released in 1985, let alone 1986 it would have been a very different and much more powerful machine) Those compromises also precluded adding a POKEY onboard.

 

^though with all that said, some of those short-term compromises could have been better addressed if they'd come out with an expansion module. (the cart slot had all the signals needed, so a passthrough cart would work, like the High Score cart already planned in 1984 just with a POKEY and more RAM onboard -easily could have added 8k in 1986, probably 16k, push it out to '87 and 32k was probably fine) And Tramiel was trying to go cheap on stuff, but loading costly components onto carts rather than a one-time upgrade is kind of against that idea. (and newer models could just integrate the added hardware on a more consolidated board) Also a MUCH better idea than the XEGS ... and the POKEY onboard could easily include a keyboard connector AND SIO port. (that means potential for cheap tape software too -big deal if they managed a European launch sooner ... it would also have worked around some of Nitnendo's licensing restrictions)

Edited by kool kitty89
Link to comment
Share on other sites

I have only a vague idea what might have happened with the release of the 7800 or what went on between the three aforementioned companies. I was just going by some of what Leonard said at CommVEx in July (the video is posted online). He did seem to have trouble remembering some of the stuff that went on 30 years ago during Q&A, so I wouldn't be surprised if there were some errors. I also might have misunderstood some of it. He focused primarily on Commodore stuff, as that was the theme of the gathering.

 

It would have been great to hear more. If only he had more time...

Link to comment
Share on other sites

  • 1 month later...

 

 

In hindsight (or Mike Katz's own view in 1988), the complicated partnership might have been the better idea especially if Katz stayed around (Sam Tramiel probably would have ruined everything if Katz was gone). Mike Katz and Jack Tramel leaving had a huge negative impact on post 1988 Atari Corp -also remember Katz joined Sega of America in 1989 and was instrumental in establishing the Genesis before Tom Kalaniske took over in 1991. (Mike and Jack also fit wonderfully together in terms of business sense, ethics and just general non-nonsense practicality and at times being too honest and not charismatic enough -the main fact Katz cites for his leaving Sega, being unable to placate Japanese management or stroke their ego satisfactorily -not 'warm and fuzzy enough' per his own words) Sega would have almost certainly been dragged down with Atari had the deal gone through AND the bad shift in management occurred as it did historically. (between the DRAM price inflation and stagnation of 1988 and Jack Tramiel leaving followed by Mike Katz a few months later, the winter of 1988-1989 marked the beginning of Atari Corp's downward spiral)

 

 

First off, great points about all of the hardware. We should remember that Yamaha wasn't willing to sell Atari Corp the YM2151 at the time of the ST's development because they weren't selling it to any [consumer] company they considered a competitor which is one of the reasons why the ST got stuck with that crummy YM2149. Yamaha had their own MIDI computer they debuted around the same time as the ST [i think it was actually presented to the public and on sale first].

 

As for POKEY making it into the ST, your points made make sense. It probably would've required a 6502 in there to manage it. That's what Atari Coin/Games did with a lot of their arcade titles…they had the 6502 behaving as a Sound CPU managing POKEYs, YM2151s, and TI speech synthesis chips while the the main CPU was usually a 68000 or 68010. From the specs I've seen, Atari Games never cranked up the Mhz on their POKEYs although prior to the death of Atari Inc, there were efforts to get various A8 chips cranked up to twice their normal clock speed when people like Tod Frye and Jerry Jessop were trying to convince Warner not to go with GCC's 7800. Having the ST use a 6502 as a sound CPU - amongst other features - would've been interesting and predated the Sega Genesis/Mega Drive doing the same but with the Z80 [gross] while in the Genesis/Mega Drive mode and then calling the shots when the Power Base Converter was plugged in to play the Z80 based SMS titles. Of course, adding POKEYs, 6502s, or more expensive YM chips would've increased the ST's prices which was counter to the "RBP" goal. What a shame.

 

Now to your point about Michael Katz and Jack Tramiel being simpatico, I think that's a bit rosy. Listen to Katz's interview with the Antic Podcast. He took the job not only to do video games but to also do electronic games. This is why the division was known as "Atari Entertainment Electronics" during his tenure [that and they obviously couldn't use the "Atari Games" moniker]. He wanted to compete against Worlds of Wonder's Lazer Tag - and Photon - with a similar product that Midway - of all companies - had designed. According to him, that was part of the employment agreement but Jack essentially screwed him and Midway over. As a kid back then who had Photon, I had really wished at the time that Atari had produced their own version [which is why this was so surprising to find out Katz wanted to do such a thing]. Hell, they could've tied it in with Atari Force. Katz also alluded to Jack basically believing he could get away with selling the XE Game System in the place of the Sega Genesis which is really eye opening. But that would've been the start to the "Entertainment Electronics" part. I think he said he wanted to do interactive toys like what Mattel was doing with Captain Power [or Axlon for that matter]. Oh what could've been. Needless to say, Katz should've been given as much room as possible. His later success at Sega of America clearly illustrates that.

Link to comment
Share on other sites

 

 

First off, great points about all of the hardware. We should remember that Yamaha wasn't willing to sell Atari Corp the YM2151 at the time of the ST's development because they weren't selling it to any [consumer] company they considered a competitor which is one of the reasons why the ST got stuck with that crummy YM2149. Yamaha had their own MIDI computer they debuted around the same time as the ST [i think it was actually presented to the public and on sale first].

That seems unlikely to be the main reason. The YM2151 would've just been too expensive and IF Atari wanted to use SOME FM sound chip in place of the YM2149, the YM2203 would make perfect sense as it shares identical I/O functionality and backwards compatibility while also being cheaper than the YM2151. (or at least should be cheaper and a less high-end part than the 2151; the OPL would be the other cheaper option, but I doubt cheaper than the 2203 and less useful given the lack of I/O and use of 2-op rather than 4-op FM -yes, 9 rather than 3 channels, but not near as nice, plus the 2203 includes the YM square wave channels too)

 

The YM2149 was better in both being cheap and having General Instruments manufacturing compatible chips (so Atari could pick between which vendor was cheaper at the time and had a second source if shipments were delayed).

 

Given how high-end the YM2151 was in 1985, Atari was probably better off looking into the likes of Ensoniq or such for competitive pricing. Anything in that range would be more an add-on but potentially a cartridge type arrangement rather than a more costly midi box. (the existing ST cart slow has enough address space to support the Ensonique DOC used in the Apple IIgs, but an external audio mixing line would be needed -adding audio in/out lines on the cart slot would've been nice, especially with the YM's 3-channel output for potential stereo effects through an external device)

 

Still, simple DMA sound would be the cheapest enhancement and probably something omitted more due to time to market than any material cost. (though not including it in the 1040ST or 520STf in 1986 or at least the MEGA models in 1987 was more a mistake, same for omitting scroll registers in the SHIFTER -a bigger problem than lack of blitter and the main two things that even lazy Amiga games tended to show off) Scrolling's also useful for business applications, especially in a single-tasking environment with full-screen applications. (scroll most of the page and just re-draw the static boarder and mouse cursor -nice for paint programs too)

 

 

 

 

 

 

As for POKEY making it into the ST, your points made make sense. It probably would've required a 6502 in there to manage it. That's what Atari Coin/Games did with a lot of their arcade titles…they had the 6502 behaving as a Sound CPU managing POKEYs, YM2151s, and TI speech synthesis chips while the the main CPU was usually a 68000 or 68010. From the specs I've seen, Atari Games never cranked up the Mhz on their POKEYs although prior to the death of Atari Inc, there were efforts to get various A8 chips cranked up to twice their normal clock speed when people like Tod Frye and Jerry Jessop were trying to convince Warner not to go with GCC's 7800. Having the ST use a 6502 as a sound CPU - amongst other features - would've been interesting and predated the Sega Genesis/Mega Drive doing the same but with the Z80 [gross] while in the Genesis/Mega Drive mode and then calling the shots when the Power Base Converter was plugged in to play the Z80 based SMS titles. Of course, adding POKEYs, 6502s, or more expensive YM chips would've increased the ST's prices which was counter to the "RBP" goal. What a shame.

I suggested the same thing in a thread a few years ago, that along with dual POKEYs in the ST and similar things (including omitting both ACIAs and the hitachi keyboard scanner MCU in favor of a parallel keyboard interface) but there were a bunch of practial counters to it, particularly including that a DMA sound circuit should be much cheaper than any of those options and more powerful. (in fact, the same criticism comes up from Genesis/MD programmers who have any hardware design knowledge -the Z80 is a massive waste in there and even gimped as a poor-man's less-than-cost-effective PCM sound driver; adding a DMA sound circuit and dropping the Z80+SRAM into the Power Bace Converter should have cut costs and made the MD easier to get good sound out of -and no Z80-68k bus arbitration logic needed, either ... smaller, simpler, cleaner board and the only backwards compatible hardware being in the VDP and I/O ends of things)

 

There's probably a better argument that the Apple II to Macintosh transition should've been backwards compatible (use the 6502 bus for I/O and video handling, let the 68k run without wait states) especially since up to 1984, all Apple IIs had still been plain old 1 MHz systems, making that much more potential for cycle stealing bandwidth without totally halting that little 8-bit processor. (you could easily manage 3 MB/s on the 6502 bus without asserting wait states, PLUS the 6502 is also extremely fast on interrupts, so if they DID go lazy and omit DMA sound, it'd be pretty reasonable to just throw in a bare DAC port for the CPU to write to manually with an interval timer setting the sample rate -still a waste though, better to use DMA sound and make use of that 1 MHz CPU for coprosessing/IO handling)

 

The Apple II was a much more costly/higher margin product than any of Atari Inc or Atari Corp's machines, so a more costly arrangement like that makes a lot of sense. (same for Tandy using a 4 MHz Z80 + 6 MHz 68000 in the Model 16 ... same WOULD have made sense if IBM had gone for a Z80 for the PC rather than an 8088 and dropped in a 68000 later on instead of upgrading to a 286 -a 5.37 MHz Z80 would be faster/nicer for a lot of stuff than a 4.77 MHz 8088 anyway, even with a primitive bank-switching scheme like the TRS 80 Model 2 used ... even a 3.58 MHz Z80 might've been preferable in some ways -sticking with the NTSC-centric timing IBM used)

 

 

But back on topic: the ST in large part was to be a Better, color-capable Macintosh for a rock-bottom price. And honestly, hardware-wise, the DMA sound of the mac was the only technical shortcoming of the 520ST. (the CPU was faster and wait-free, the monochrome mode was higher res, the standard monitors were larger -though H/V beam pitch/height control pots would've been nice to hide overscan, it had 2 color graphics modes and a nice palette, and the overall base system was more capable than a Mac or PC/AT short of the latter's standard hard drive -Atari should've gotten on the ball with a SCSI hard-drive interface via the DMA port much sooner too, particularly in time for the MEGA ST a couple years down the road -I understand the pricing wasn't right for the 520 or 1040 in Atari's business model, but the Mega should've fit even a luxury priced drive)

 

 

 

 

Now to your point about Michael Katz and Jack Tramiel being simpatico, I think that's a bit rosy. Listen to Katz's interview with the Antic Podcast. He took the job not only to do video games but to also do electronic games. This is why the division was known as "Atari Entertainment Electronics" during his tenure [that and they obviously couldn't use the "Atari Games" moniker]. He wanted to compete against Worlds of Wonder's Lazer Tag - and Photon - with a similar product that Midway - of all companies - had designed. According to him, that was part of the employment agreement but Jack essentially screwed him and Midway over. As a kid back then who had Photon, I had really wished at the time that Atari had produced their own version [which is why this was so surprising to find out Katz wanted to do such a thing]. Hell, they could've tied it in with Atari Force. Katz also alluded to Jack basically believing he could get away with selling the XE Game System in the place of the Sega Genesis which is really eye opening. But that would've been the start to the "Entertainment Electronics" part. I think he said he wanted to do interactive toys like what Mattel was doing with Captain Power [or Axlon for that matter]. Oh what could've been. Needless to say, Katz should've been given as much room as possible. His later success at Sega of America clearly illustrates that.

Interesting, but that doesn't sound quite right. Firstly though, I will say that I meant that Katz and Jack Tramel worked well together on a management/business level, not that they LIKED each other per se. Both were capable of dealing with a critical, no-nonsense hard business style from what I've seen/read and given Katz didn't leave until AFTER Jack had stepped down as CEO, I can't think that he was pressured into leaving (or outright fired as he was at Sega) for failing to fall in line and humor upper management's egos. (I could see him leaving in disappointment over the loss of the Genesis contract, or in disappointment in what Sam Tramiel was doing, or just general wear and tear over many factors -he did leave to take an extended vacation from the industry as a whole that was only cut short at the behest of Sega) I also meant that, regardless of their relationship, Atari Corp never regained the level of management ability shown during the Jack+Katz days. (either on the computer or games end of things)

 

From Jack's point of view it really seems foolhardy to judge the XEGS that way ... the ST and STe perhaps, and the Panther must have been on the drawing board in 1988 as well if not wirewrapped (it was in down silicon the following year following revision and refinement from Martin Brennan) but the XEGS made no sense there (and honestly, it made very little sense next to the 65XE selling for less -a Deluxe gaming bundled 65XE would've made more sense and simplified mass production) A very stripped down STe might have made a decent budget-market console for 1989, but really would've made more sense back in '87 had they had a SHIFTER with at LEAST hardware scrolling at the time. (with scrolling you can get away with a lot more without needing massive pre-shifting tricks and thus make do with around 128 kB of DRAM rather than 512 kB ... scrolling SHIFTER with DMA sound would've done it, along with scrapping the mess of computer-specific I/O hardware in favor of something cheap and simple ... a 256x200 32 color/shade mode would've been nicer too, better pixel size for NTSC and better color without more CRAM and just one bit for high/lo intensity -ie 3-3-3-1 RGBI)

 

 

Now ... a 7800 expansion module in 1987 would probably have been a better idea than the XEGS too, probably easier to expand on their existing budget market than a 16-bit console too. 32kx8-bit SRAMs were getting cheap by that time (cheap enough that Epyx put them on carts that only needed 16k) so something like 32k+POKEY would've been nice, maybe with SIO and keyboard ports too. (XEGS style keyboard port would be useful there) Sacrificing 4 kB of that RAM to mask the existing SRAM block would be a nice option if they planned to upgrade the base unit into a Super 7800 type system. (if they wanted to keep it add-on only, it'd be better to just bank switch the 32k SRAM into the existing 48k address space) 32k would be enough to load tape based games into and/or use framebuffer based 4 color backgrounds, or multiple 3 color playfields even. (and software sprites or -more likely/useful, software driven character modes on those playfields that take less time per frame to build than CPU+DMA overhead of MARIA sprites each 160x192 3-color plane would be 7.5 kB or a bit more if you use overscan to help with scrolling -hiding character updates in the boarders) Probably make Summer and Winter games both launch titles for the expansion system and cancel the more costly versions with onboard SRAM. (but beef the new versions up with some nice POKEY tunes)

 

You'd also have a system that (for cart games at least) was definitively superior to the 65XE. (better graphics, equally fast CPU -with similar free CPU time in framebuffer displays, and better sound via TIA+POKEY) No low res 9/16 color modes, though, for what those're worth.

 

If they did release an all in one Super 7800 (or 7800Plus) or whatever, using a nicer arrangement for the TIA/MARIA video would be worth it, even if it amounted to a literal manual switch. (composite video connectors like the XEGS used would be nice too)

 

A cost cut, merged TIA+RIOT would be nice too. (arguably a better cost-cutting measure than the single-chip late-gen 2600 Jr, unless it was cheaper to just re-use that chip and disable the 6507 when using MARIA)

 

Hell, given how late the 7800 was introduced overseas, it might make more sense to go straight for the 32k+POKEY equipped model there. (that might actually have a chance at competing alongside the Master System in the budget end of the market there, especially with SIO tape drives used for budget software)

 

 

The $200 price tag the XEGS was touting in 1987 would be much more justified for a deluxe bundled 7800+Expansion module (or integrated all in one 7800EX) with pack-in games showing off the new capabilities ... maybe include Ballblazer as a built-in game too. (nice POKEY-compatible game that'd be confusing to re-release separately)

 

 

Another bonus to drawing more attention to the 7800 is the sprite/list system is similar to what the Panther and Jaguar later used, so if Atari was keen on sticking with that (rather than say, dropping it in favor of a blitter based system -or in the Jaguar's case, blitter without OPL) it'd give developers more room to get comfortable with using the novel/different architecture. (I'm not so sure they should've gone with Panther, but this would at least give it more merit ... designing the Panther around 4 of those fairly cheap 32kx8-bit SRAMs rather than the rather costly 8kx8-bit 35 ns -386/486 cache class- SRAMs, it'd have been far more useful ... maybe needing an added latch to make it work right, or clocking the graphics chip and CPU a bit slower, but totally worth it -using an 8 MHz pixel clock like the ST is kind of nasty on NTSC TVs anyway, blurry and leaving a big boarder at 320x200 ... 12.5 MHz or 13.3 MHz like the Jag used would be nice, or something more universally NTSC/PAL-compliant ... plus SRAM makes interleaved bus sharing so so SO easy it's a horrible waste not to take advantage of) I'm also assuming the 16 MHz Panther used a simple 8 MHz dot clock and not more variable (with the 32.22 MHz master clock, you'd get a nice 320x200 screen at 1/5 speed -6.44 MHz, just below the MegaDrive's 320 wide mode, and 1/6x would give a nice 5.37 MHz if you wanted to exactly match the SNES or MD low-res mode -useful for ports using common pixel art for similar aspect ratios)

 

^Honestly, I think they'd have been wiser to invest more in the ST chipset at the time, possibly with re-spinning it as a game system as a secondary goal. (putting more effort into making the 1040STe really nice and a serious upgrade -like 16 MHz CPU, 256 color chunky/packed pixel modes, maybe a 16 MHz blitter with fast block/line fill and move for chunky/packed pixel operation, and overscan support for the SHIFTER with more variable dot clock/resolution modes) And honestly, adding that to the SHIFTER should be a simpler job, chip-design wise than the Panther by a good margin. (hell ... even with the 1989 STe as it was, if they'd turned around and spun a 16 MHz, 256 color capable, chunky-pixel arranged system for 1990 ... something cheaper/simpler by far than the TT030 -ie about as cheap as the existing 1040STe- it'd be pretty nice ... hell, given how little the BLiTTER was actually used, and given how decent a 16 MHz 68000 would be at software blits for 8-bit chunky pixels, they could save the space/cost and omit that entirely -the TT030, unfortunately did not use chunky pixels and thus had less than elegant handling of its added 256 color VGA monitor modes)

Link to comment
Share on other sites

That seems unlikely to be the main reason. The YM2151 would've just been too expensive and IF Atari wanted to use SOME FM sound chip in place of the YM2149, the YM2203 would make perfect sense as it shares identical I/O functionality and backwards compatibility while also being cheaper than the YM2151. (or at least should be cheaper and a less high-end part than the 2151; the OPL would be the other cheaper option, but I doubt cheaper than the 2203 and less useful given the lack of I/O and use of 2-op rather than 4-op FM -yes, 9 rather than 3 channels, but not near as nice, plus the 2203 includes the YM square wave channels too)

 

The YM2149 was better in both being cheap and having General Instruments manufacturing compatible chips (so Atari could pick between which vendor was cheaper at the time and had a second source if shipments were delayed).

 

Given how high-end the YM2151 was in 1985, Atari was probably better off looking into the likes of Ensoniq or such for competitive pricing. Anything in that range would be more an add-on but potentially a cartridge type arrangement rather than a more costly midi box. (the existing ST cart slow has enough address space to support the Ensonique DOC used in the Apple IIgs, but an external audio mixing line would be needed -adding audio in/out lines on the cart slot would've been nice, especially with the YM's 3-channel output for potential stereo effects through an external device)

 

 

I doubt they would've went with Ensoniq considering how much bad blood there was between Jack Tramiel and Bob Yannes then. Remember, when Bob and others left MOS/Commodore over a pay dispute with the C64, they approached Steve Ross of Warner to create a keyboard/computer upgrade for the 2600. That made Tramiel so livid that he tried to sue them for patent infringement just because they were using a 6502 in the project.

 

Now years later, Atari Corp was going to use an Ensoniq chip in the Panther but Sam Tramiel was calling the shots then. Or maybe that was just another attempt to screw them over kinda like what his father did to Midway over the Lazer Tag competitor Michael Katz had licensed and was going to release through Atari Corp until Jack reneged on the deal [most likely just to show Katz he was the one in charge].

Edited by Lynxpro
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...