The Falcon's confusing history

[czietz]

As you might know since my find of Atari ASIC schematics (https://www.chzsoft.de/asic-web/) last year, I'm browsing through the contents of the Atari HQ CD-ROM and further files available on the Internet that contain historical data from Atari.

Currently, I'm researching a bit into the history of the Atari Falcon. Maybe there's someone -- an insider even? -- who can confirm my assumptions or provide further details.

Interestingly, there is a technical specifications document for a "Falcon", dated December 1991 which describes some kind of powerful machine: optional 68040 CPU, FastRAM (like in the TT), VME bus etc. As you know, the released Falcon has none of those. The design team for this "Falcon", lead by John Horton, was with Atari Dallas, Texas. But then there is a specification document for the "Sparrow", also dated December 1991! The "Sparrow" is often considered to be the Falcon prototype (see for example http://www.atarimuseum.com/computers/16 ... index.html) and indeed its specifications much better match what we know as Falcon030. The "Sparrow" design team was with Atari Israel, e.g. Moshe Segal.

My interpretation of this is that Atari in 1991 had two competing design teams (one in Texas and one Israel), each working on a new computer -- and finally the "Sparrow" team won and their machine was released as Falcon030, whereas the computer originally designed as "Falcon" never made it to the market.

Then, there is the Falcon Microbox for which I have bill of materials, netlist and PCB placement plan, dated January 1993. Prototypes of this machine are known (http://www.computerhistory.org/collecti ... /102672949). It seems marginally improved from the released Falcon030 but still nowhere near the "Falcon" of late 1991.

[bid]

Wowser!!

What you seem to have discovered, is that the "Falcon" codename was for the 68040 microbox with build in card slots, and the true next gen machine.

Sparrow, is what the ST Cased 68030 cut down, baby falcon, which was released ... possibly using up IC's and parts from inventory.

I've personally never liked the Falcon030 (haters gonna hate), and this justifies (to myself) my view, that Falcon030 is indeed the 'sparrow', and the the (true) 'falcon' was never released.
- I'm pretty sure the true falcon was not going to use VME, but ISA type slots tho? There is a prototype board on Atari Museum website.

I've always wondered how the (abomination) that was the Falcon030 came about, and I was shocked to my core by a ST cased 68030 with no easy upgrade path ... especially after the rumours, and the amazing prototypes shown in the 90's for amazing machines.
- I remember the 'What communucates now communicates' advert, and being highly confused and not at all impressed by what I saw.

I'll bet that Falcon was canned, and sparrow was a method to use up remaining stocks and just fold up the computer side of the business.
- AFAIK the sparrow was originally a daughterboard mounted onto an ST, and IMHO the "Falcon030" that we have today, is just an ST with a 68030 accellerator slapped on it (not a great accelerator either!?!).
- I mean the MSTe was 16Mhz with cache, and take a look at Dave Small's 68030 SST accelerator!

I think that the slapping on of a 68030 to an ST bus, with maybe the original spec graphics chip. The Motorola 68030 was like a modular system, so Atari cant take credit for the DSP ... Its exactly the same in the NeXT and almost any other 030 based system - Motorola designed a wonderful modular system. Just add sound, shifter and other bits you want.

I think Atari should have gone the whole hog, and made a Modern PC, like a mobo and add-on cards we have today. This would pretty much me like the Hades, but price comes in volume, and I feel that the management at Atari were VERY poor. Here you have an amazing company, with the most gifted and ingenious engineers, with basically pricks at the helm. Crap marketing, bullying, ideas from the 80's and authoritarian regime.

Its obvious that the skilled Atari engineers built these machines, and prototyped things like ST-Book on a block of wood. STylus was a working prototype, and we see this type of thing in every hospital now on trolleys, the ST-Book, portfolio, VR-Headset, even the ST itself. A work of art! ... I love examining and admiring every detail of these machines, as there is nothing like this today, and its clear a lot of thought went in.

Apart from the sparrow. As far as I am concerned, it is a Kludge, to just use the parts up. Im guessing it got manufactured, as in accountancy terms, it was cheaper than selling back 2nd hand IC's or opening a landfill.

Appologies, if I have upset anyone!

[calimero]

@czietz thank you very much for findings! I am also very interested how Falcon come to life. I did not know that there were two times, not that Moshe Segal (first time to hear for him) was responsible for Falcon as we know it. Two separate teams explain why we have so much different Falcon "successors".

@"Falcon in ST case" - as far as I read: Atari management was on the edge to NOT RELEASE ANY Falcon by 1992.!
Atari ST case was compromise, to cut cost down, otherwise Atari management probably would not release any Falcon on market. Everything at Atari Corp. was already serve only to one goal: Atari Jaguar.

It would be very interesting to know if Texas Falcon was working prototype and if it could run TOS programs?!? Also to know some more information about hardware and capabilities. IT is interesting that both prototypes manage to get to silicon, on REAL motherboard! Really amazing! If we could only know if all these prototypes was able to run some software...

And one more note: I opened sometime ago thread "Falcon designer" in hope to find who was responsible for Falcon that come to market. I am sure that Marty Goldberg and Curt Vendel have most complete picture about this topic but unfortunate they are both quite silent for sometime...

BTW @big you have "[/i][/color]" in your sign...

beautifully said "admiring every detail of these machines, as there is nothing like this today, and its clear a lot of thought went in."

[calimero]

I just try on google: Moshe Segal atari
and there is no results regarding Atari Corp.
Also Moshe Segal is not mentioned in http://mcurrent.name/atarihistory/trame ... ology.html

[calimero]

replay form thread: REWRITING HISTORY - 'Jack Tramiel loaned Amiga $500.000'

It beggars belief that the upgrade path in terms of CPU and Motorola's own work in this regard, really only requires a proper specification to be made and executed in a timely fashion. Its clearly shown by the original aftermarket boards, that 020, 030, more ram, SCSI, IDE etc etc were all possible in a low cost manner. Atari could have easily just incorporated them into a new machine. Even the early TT shows how a 16MHz chip is replaced with a 32MHz daughterboard. I believe that Sparrow was the same, after the real Falcon was canned, I'll bet engineers built it, and practically begged for it to be released as some sort of consolation.

1) 68020 ST successor start as early as 1986. by Shiraz Shivji but it was finished by Roy Good (TT) after many delays. We can see that main problem of scaling ST architecture was in ST custom chips and timings required from memory: TT shifter have 64bit access to ST RAM, motherboard works on max 16MHz... but fast ram is added to enable full speed of 32MHz 68030. If they stick to original 68020 16MHz I believe that there is no need for fast ram at all.

2) Falcon030 is also stretch of ST architecture. While ST and TT does not suffer from slow down in case of video RAM access (like original Macintosh) Falcon030 does.

Faster RAM, like EDO Ram, in Falcon would be really beneficial.

And most of all, I would really love to see how "other" Falcon would save this problem on technical/design level. In what aspect it would be better than Falcon030. For me it is really impressive that Atari Corp. made real hardware: chips and motherboard of "other" Falcon.

[shoggoth]

I'd say the TOS sources somewhat confirms all this. All "falcon"-related stuff is labelled "sparrow" in the TOS4 sources, but there's some obvious "things to come"-thinking in the code, especially on the VDI side.

[calimero]

I'd say the TOS sources somewhat confirms all this. All "falcon"-related stuff is labelled "sparrow" in the TOS4 sources, but there's some obvious "things to come"-thinking in the code, especially on the VDI side.

You already informed us that you found in source code that TOS were prepared for 8bit chunk mode and 24bit mode, right?

Maybe this was features of "canned" Falcon?

[shoggoth]

You already informed us that you found in source code that TOS were prepared for 8bit chunk mode and 24bit mode, right?

Maybe this was features of "canned" Falcon?

It's somewhat prepared for 8-bit chunky and 32-bit truecolor, yes. They were most likely using MATRIX cards internally when developing that stuff (there's a partial MATRIX driver in TOS4, it seems). The SuperVidel driver (ab)uses this in the TOS4 driver (that's why 32-bit/8-bit modes "work" without NVDI on such hardware).

Wether or not these are leftover features intended for the "canned" falcon - I don't know. They clearly had some other future machine in mind however.

[Ektus]

As for the "modern" design, that came later with the Milan. Baby-AT board, socketed CPU, ISA and (at the time) modern PCI slots, but at the cost of compatibility. Had the Milan come in 1992, with a SuperVidel-like graphics card, things might have gone differently. But that's ancient history now...


Regards
Ektus.

[Mindthreat]

While it's been talked about elsewhere and just my 2-cents in.... it still would have been greatly fascinating had they actually implemented the Jaguar chipset with the Falcon together as planned and we would've had an amazingly capable system.

[calimero]

As for the "modern" design, that came later with the Milan. Baby-AT board, socketed CPU, ISA and (at the time) modern PCI slots, but at the cost of compatibility. Had the Milan come in 1992, with a SuperVidel-like graphics card, things might have gone differently. But that's ancient history now...

Atari ST was FIVE years (at least) ahed of PC crapware - software wise and hardware wise. (please searach for thread: Programs born on Atari, Amiga, Mac...)
Atari Falcon was couple years ahead of PC because of DSP.

There was lot of room for PC NON compatible computers back in late 80s and early 90s but only if Commodore and Atari would keep pace with Jack Tramiel quote: "I refuse to produce a clone that is behind the times. I am producing products for the most sophisticated consumer in the world."

Atari ST was exactly that: Computer much advanced than IBM compabilities from 80s, software on ST was so much advanced than DOS crapware, BUT there was no "next step" for Atari ST users for almost decade!

[calimero]

It's somewhat prepared for 8-bit chunky and 32-bit truecolor, yes. They were most likely using MATRIX cards internally when developing that stuff (there's a partial MATRIX driver in TOS4, it seems). The SuperVidel driver (ab)uses this in the TOS4 driver (that's why 32-bit/8-bit modes "work" without NVDI on such hardware).

here are some more information from 1987. (I believe) interview with Shiraz Shivji:
http://st-news.com/issues/st-news-volum ... az-shivji/

1) future high end computer
"The Atari Transputer computer will be a very fast one,
with incredible graphic capability and it will be expandable. It
will be basically a computer with 5 mega RAM, 1 of which
dedicated to video tasks; it will have 4 graphics modes with 32,
16, 8 and 4 bit per pixel, 8 bit for each colour beam (R, G and
B), an 8 bit tag and 256,000 colours to choose from, thanks to
the Inmos Colour Lookup Table chip with a new blitter;"

- @ shoggoth 8bit and 32bit colors stuff you found is only in TOS 4, not in TOS 3?

2) possible TT/UX that was never released
" Q: You said before that the "number cruncher" is a new
project and it is no more based on 68020 but on the 68030.
A: The fact is that we wanted to expand within the Motorola
family and we had chosen the 68020 with math co-processor and
Memory Management Unit (MMU) to get hardware protected multitask-
ing in order to safely run UNIX; but Motorola delayed the
availability of such MMU and so a couple of weeks ago we have
started the project again with 68030 that has MMU on chip and is
actually available to us in some samples; it will be a 4 MIPS
box with expansion slots and connection via DMA ports to existing
ST."

3) Enhanced ST (STe?) should have TT video chip:
" Q: What about the Enhanced ST?
A: This is a 68000 machine and is almost ready. Actually I'm
waiting for the video chip and I can only tell you that its video
resolution will go from 1280 x 960 in b/w to 640 x 480 in 16 cols
and 640 x 240 in 256 cols"
---

not directly Falcon related, but more of retrospective what Atari was doing in 1987/88.

[mikro]

"The Atari Transputer computer will be a very fast one,
with incredible graphic capability and it will be expandable. It
will be basically a computer with 5 mega RAM, 1 of which
dedicated to video tasks; it will have 4 graphics modes with 32,
16, 8 and 4 bit per pixel, 8 bit for each colour beam (R, G and
B), an 8 bit tag and 256,000 colours to choose from, thanks to
the Inmos Colour Lookup Table chip with a new blitter;"

- @ shoggoth 8bit and 32bit colors stuff you found is only in TOS 4, not in TOS 3?

This has nothing to do with TOS. The ATW didn't use TOS at all, it was run by HeliOS.

[calimero]

"The Atari Transputer computer will be a very fast one,
with incredible graphic capability and it will be expandable. It
will be basically a computer with 5 mega RAM, 1 of which
dedicated to video tasks; it will have 4 graphics modes with 32,
16, 8 and 4 bit per pixel, 8 bit for each colour beam (R, G and
B), an 8 bit tag and 256,000 colours to choose from, thanks to
the Inmos Colour Lookup Table chip with a new blitter;"

- @ shoggoth 8bit and 32bit colors stuff you found is only in TOS 4, not in TOS 3?

This has nothing to do with TOS. The ATW didn't use TOS at all, it was run by HeliOS.

you did not read entire text and I did not quote MAIN part:

": Mr Shivji, hi again; would you give me more details about
the transputer based machine?
A: The Atari Transputer computer will be a very fast one,
with incredible graphic capability and it will be expandable. It
will be basically a computer with 5 mega RAM, 1 of which
dedicated to video tasks; it will have 4 graphics modes with 32,
16, 8 and 4 bit per pixel, 8 bit for each colour beam (R, G and
B), an 8 bit tag and 256,000 colours to choose from, thanks to
the Inmos Colour Lookup Table chip with a new blitter; you will
be allowed to plug 12 extra Transputer to get more speed.
Transputer hardware environment is called Perihelion and is a
project by AmigaDos indipendent author Tim King from Cambridge;
the new machine will run, apart from Occam, Transputer own
language, C language and will have GEM environment."

[Ragstaff]

You can get GEM as a GUI for DOS and I believe other operating systems too. Maybe he just meant GEM would be the GUI?

[bid]

While it's been talked about elsewhere and just my 2-cents in.... it still would have been greatly fascinating had they actually implemented the Jaguar chipset with the Falcon together as planned and we would've had an amazingly capable system.

This is exactly what I was expecting! ... and although Flare seems to be unrelated to Blossom, I did think that something cool in terms of unparalleled video and audio processing power was just around the corner.

My 2 cents is that I really hate the "Jack Tramel is the father of the ST" type stuff. I work in business, run a business, and I really don't think Jack understood the 16/32 technology, its upgrade path, nor how to brand and market machines. Similarly with the C64, there was a team, or indeed one or more expectational individuals who thought way beyond others/the curve.

An example of this type of individual is Gary Kildall, who created GEM. Kildall was ripped off by MS-DOS, and hence when Microsoft copied his code, the PC Contemptables became FAT compatable with the Atari; not the other way round. Hence Atari would read PC disks and vice versa.

Kildall was a genius, in the truest sense of the world. Similarly, look at Chuck Pedal, Shiraz, and perhaps some other luminaries that worked at Atari. Atari was a mecca for these types of people, and remember all types of hardware and software gurus worked there once, including Steve Jobs.

Gem certainly, in my view, was always superior to Microsofts buggy and (in my view) still beta Windows 'Gui For DOS'. A dreadfull kudge by any standard, in which they tried to hide DOS in ME edition, and I suppose NT did start to have some of the trappings of a real operating system. But again much of NT was ripped off code, including the boot-loader which is still in use, highly buggy and I forget what its ripped from, but look into it.

GEM however, is a truely elegant OS, and it is a shame that Atari did not continue to develop ... Gary died, but the system he created is truely a wonderful, lightweight, fast OS that is modular, suitable for platform change, and tight. Hence the near perfect MIDI timing, and you know, I dont think I have experienced many OS crashes in 25 years ... and then only from software that does not stick to the rules.

I am sure that engineers wanted to update TOS, and create new hardware. Its clear that in 86-87 they were wanting to do this, and yes the 020 was crap. But where was the hardware. Sure, there was enough profits for capital expenditure. Why, even on the forum, people are creating accelerators, using the IDE tos loader code in 2.06, and look at what Dave Small did. Did he have Atari's budget.

To this day, Atari sticks out as a company that was badly managed, and the hardware idea (ST) was so great, that it took the Tramels nearly 10 years to finally kill it. The ST was the golden egg. A computer that beat the PC many times over, and as pointed out, even the crippled Falcon (aka Sparrow) was ahead of the PC.

I dont buy the lack of market for Atari computers. CLab found one, and Atari computers were purloined for industry (I got my TT from a knitting factory, where it ran an industrial machine), in the automotive industry they ran the show. I've always been frustrated into being effectively forced to use a Intel chip, sucking my electricity and pumping out heat, and an OS (Windows) that I hate. I really cant understand how a computer monopoly formed, and I've always thought that something lightweight and accessible may one day return, when the madness ends!

We have some good accounts of the ST's development, from dadhacker, and the new info is very interesting. However, I suspect that wheras, Jack 'flew in' to annoy the hell out of the ST dev team, and threaten them. That once Jack and sons had complete authoritarian control, it was fairly easy to kill hardware projects, or cripple them. If not, then lawsuits, arguments and direct war with US distributors, insanely poor marketing, communication and killing the US market dead would help some. You can hear it in the voice of employee's on the old BBS threads. These guys were smart!

Jags faults, I sure were fixable. But, no, I don't want to hear any more. I am paying for this. Lets ship it. Again, what must it have been like to be a developer? ... To be honest, I would like to hear from developers that met the company / a Tramel. I'd be curious, as I would image that they would not be under same NDA or compromise agreements?

[AdamK]

GEM was not that great. Maybe in 1985, yes, but world moved on and GEM froze. I think DRI didn't want to invest in it after they lost Apple suit, and Atari did not have manpower to push it forward.

To this day, Atari sticks out as a company that was badly managed, and the hardware idea (ST) was so great, that it took the Tramels nearly 10 years to finally kill it. The ST was the golden egg. A computer that beat the PC many times over, and as pointed out, even the crippled Falcon (aka Sparrow) was ahead of the PC.

No not really, by any count. ST line was introduced in 1985 and officially ended in 1993. That is 8 years tops, but really, after 91 (92 tops) ST line was dead in the water, and Falcon did nothing to help it.

Why Tramiels were slow to update ST? My money is on the theory, that after they invested into retail chain that flopped, they lost most of the capital and had no real money to invest. And as charming as he is Jack had trouble finding any 3rd party investor.

[Cyprian]

No not really, by any count. ST line was introduced in 1985 and officially ended in 1993. That is 8 years tops, but really, after 91 (92 tops) ST line was dead in the water, and Falcon did nothing to help it.

Why Tramiels were slow to update ST? My money is on the theory, that after they invested into retail chain that flopped, they lost most of the capital and had no real money to invest. And as charming as he is Jack had trouble finding any 3rd party investor.

the same story is with Commodore.
After 1987 none of them product didn't reach a quarter of A500 popularity.
E.g. look at A1200 - it was released in 1992 and it was just a little bit upgraded Lorraine from 1983 - exactly the same blitter, sound, a bit upgraded video (only added 3 bitplanes and extended HAM mode) and memory management. I would say after amiga acquisition in 1984 Commodore lost their "invent"

[calimero]

Kildall was a genius, in the truest sense of the world. Similarly, look at Chuck Pedal, Shiraz, and perhaps some other luminaries that worked at Atari. Atari was a mecca for these types of people, and remember all types of hardware and software gurus worked there once, including Steve Jobs.

Howard Rheingold have great articles about Atari Inc.

http://www.rheingold.com/texts/tft/1.html (not sure which chapter is most Atari related but he write about crazy stuff at Atar inc.)

Atari computers were purloined for industry (I got my TT from a knitting factory, where it ran an industrial machine), in the automotive industry they ran the show.

There was special ST in Germany in industrial case special for CNC and other purposes! (I always forget model name... :/)

Gary died, but the system he created is truely a wonderful, lightweight, fast OS that is modular, suitable for platform change, and tight.

Mike Fulton would not agree with you (on of Atari Corp. 1990 to 1995 software engineers) http://www.fultonsoft.com/tag/atari-st/

I really cant understand how a computer monopoly formed, and I've always thought that something lightweight and accessible may one day return, when the madness ends!

Nobody get fired because of buying IBM computer.
This indicate that people was STUPID (lack of experience and lack of information) back in early PC days.

I today have exactly SAME problem when I try to introduce Papyrus to PC lusers: they all give tremendous resistance to Papyrus because it is not "Microsoft Office". After some time, they start to LOVE Papyrus and to HATE Microsoft Office. But it need time.

To be honest, I would like to hear from developers that met the company / a Tramel.

You have lot of interview with employees of Atari Inc. and Atari Corp. at "ANTIC The 8bit Podcast": http://www.atari-forum.com/posting.php? ... 0&p=293931 (David Small)
THIS IS A TREASURE TROWE ! I just do not have enough time to listen them all...

[calimero]

GEM was not that great. Maybe in 1985, yes, but world moved on and GEM froze. I think DRI didn't want to invest in it after they lost Apple suit, and Atari did not have manpower to push it forward.

but how than you explain one or three man behind MagiC or Geneva? Were they Supermans?

No not really, by any count. ST line was introduced in 1985 and officially ended in 1993. That is 8 years tops, but really, after 91 (92 tops) ST line was dead in the water, and Falcon did nothing to help it.

But there was lot of life in ST even past official dismiss... ST was truly remarkable piece of equipment!

Why Tramiels were slow to update ST? My money is on the theory, that after they invested into retail chain that flopped, they lost most of the capital and had no real money to invest. And as charming as he is Jack had trouble finding any 3rd party investor.

Agree.
I have hard time to accept that some (Federate Trade) screw Jack like that! In 1989. they report 0 income after, not sure, something like 100 million $ losses (write off) on Federate Trade.

[calimero]

the same story is with Commodore.
After 1987 none of them product didn't reach a quarter of A500 popularity.
E.g. look at A1200 - it was released in 1992 and it was just a little bit upgraded Lorraine from 1983 - exactly the same blitter, sound, a bit upgraded video (only added 3 bitplanes and extended HAM mode) and memory management. I would say after amiga acquisition in 1984 Commodore lost their "invent"

Many people say that ST was "rushed" to the market.
I would say opposite: Amiga was rushed to the market by Commodore: Overpriced (compared to ST) and unstable.
After Amiga 1000 entire original Amiga team left Commodore. Jay Miner stay as consultant. Commodore Germany made cost cut version after 2-3 years of Amiga 1000! (entire Amiga 1000 was designed in same timeframe)

[Trixster]

Why was the Amiga 'unstable'?

[molofaha]

I remember the Workbench being a miserable crashy mess in the first few years but upon reflection I was _very very young_ at the time and may not be a reliable witness.

[molofaha]

Why Tramiels were slow to update ST? My money is on the theory, that after they invested into retail chain that flopped, they lost most of the capital and had no real money to invest. And as charming as he is Jack had trouble finding any 3rd party investor.

the same story is with Commodore.

I always wondered why it took both of them so long to respond to the 386 with VGA. Came out in what, 1987? The show-off 256-colour games started coming out in 1990, but it took till the end of 1992 before they managed to come out with systems that were... about as good.

[AdamK]

GEM was not that great. Maybe in 1985, yes, but world moved on and GEM froze. I think DRI didn't want to invest in it after they lost Apple suit, and Atari did not have manpower to push it forward.

but how than you explain one or three man behind MagiC or Geneva? Were they Supermans?

Probably But there was no real improvements, and nor MagiC or Geneva implement everything that GEM (and later MiNT - I'm talking about before Atari let it go) included.

No not really, by any count. ST line was introduced in 1985 and officially ended in 1993. That is 8 years tops, but really, after 91 (92 tops) ST line was dead in the water, and Falcon did nothing to help it.

But there was lot of life in ST even past official dismiss... ST was truly remarkable piece of equipment![/quote]
No, there was not. Your view is skewed. Almost all commercial software development for ST stopped. Most of the users left the platform. No mass produced new hardware since.