30 replies to this topic

[HP Atari King of Michigan]

Before the 5200 became a reality, there was this system titled the Atari 3200 System X. Just after the 2600 came out, Atari was already reaserching a more powerful system. First came this sytem called the System X, or the CX-3200. Atari also made a Sears counterpart called the Sears Super Arcade II (The Super Arcade I was Sears' version of the Intellivision). But programming games for this system proved to be too hard for the folks at Atari. So they decided to scrap this system and try something else. This time they would look towards making their 8-bit computer lines (the 400 and the 800 respectively) and converting them into a gaming system. They did away with the keyboard and gave it a streamlined wedge look, and initially named it the Atari Video System X, to separate it from the 3200. It was also called the P.A.M. (Personel Arcade Machine). This would later become the 5200 SuperSystem and you know the rest.

[82atari5200]

HP Atari King of Michigan, on Thu Oct 6, 2005 5:19 PM, said:

Before the 5200 became a reality, there was this system titled the Atari 3200 System X.  Just after the 2600 came out, Atari was already reaserching a more powerful system.  First came this sytem called the System X, or the CX-3200.  Atari also made a Sears counterpart called the Sears Super Arcade II (The Super Arcade I was Sears' version of the Intellivision).  But programming games for this system proved to be too hard for the folks at Atari.  So they decided to scrap this system and try something else.  This time they would look towards making their 8-bit computer lines (the 400 and the 800 respectively) and converting them into a gaming system.  They did away with the keyboard and gave it a streamlined wedge look, and initially named it the Atari Video System X, to separate it from the 3200.  It was also called the P.A.M. (Personel Arcade Machine).  This would later become the 5200 SuperSystem and you know the rest.

Could you tell us something we didn't know...

[HP Atari King of Michigan]

82atari5200, on Thu Oct 6, 2005 9:18 PM, said:

Could you tell us something we didn't know...

Like, maybe, if anyone has seen one of these upclose at conventions or visited Curt Vendel, for instance...

[Tempest]

I don't think anyone has a 3200. AFAIK they only had two games done for it before it was scrapped (a Mouse Trap type game and something else). I've asked about it before, but there doesn't seem to be much info on it other than it was a 10-bit system (like the Intelli) and apperently wasn't that big of an improvement over the 2600.

Tempest

[TheRedEye]

HP Atari King of Michigan, on Fri Oct 7, 2005 8:42 AM, said:

82atari5200, on Thu Oct 6, 2005 9:18 PM, said:

Could you tell us something we didn't know...

Like, maybe, if anyone has seen one of these upclose at conventions or visited Curt Vendel, for instance...

Like, maybe, you could have asked "has anyone seen one of these upclose at conventions" in your original post instead of assuming we could read your mind

[Curt Vendel]

I only have a mockup console and a solid resin joystick, as for electronics, I've yet to run across any schematic or electronics for the 3200... yet



Curt

Tempest, on Fri Oct 7, 2005 10:08 AM, said:

I don't think anyone has a 3200.  AFAIK they only had two games done for it before it was scrapped (a Mouse Trap type game and something else).  I've asked about it before, but there doesn't seem to be much info on it other than it was a 10-bit system (like the Intelli) and apperently wasn't that big of an improvement over the 2600.

Tempest

[Brad2600]

Don't people use the search function anymore?

[Atariboy]

Kepone, on Fri Oct 7, 2005 10:07 PM, said:

Don't people use the search function anymore?

Did you use it before posting that? Found about 4 pages of topics relating to "Atari 3200" when I did a search just now, but the ones I checked had no relevent information (Except this one), and none of the topics appeared to be related to the search. Perhaps if I had checked 4 pages of different threads over the span of a few hours, something actually about it would've appeared.

Edited by Atariboy, Sat Oct 8, 2005 7:47 PM.

[~llama]

beyond actually seeing one, what's known about the technical aspects of the 3200? What kind of processor, video hardware, etc. did it have? I've always thought it'd be cool to write a 3200 emulator and write homebrews for it, but then again I have a sort of warped concept of what's "cool"

[HP Atari King of Michigan]

[quote name='TheRedEye' date='Fri Oct 7, 2005 3:45 PM']
[quote name='HP Atari King of Michigan' date='Fri Oct 7, 2005 8:42 AM'][quote=82atari5200,Thu Oct 6, 2005 9:18

Like, maybe, if anyone has seen one of these upclose at conventions or visited Curt Vendel, for instance...
[right][/right]
[/quote]

Like, maybe, you could have asked "has anyone seen one of these upclose at conventions" in your original post instead of assuming we could read your mind
[right][/right]
[/quote]

Point taken...

[Laner]

TheRedEye, on Fri Oct 7, 2005 2:45 PM, said:

HP Atari King of Michigan, on Fri Oct 7, 2005 8:42 AM, said:

82atari5200, on Thu Oct 6, 2005 9:18 PM, said:

Could you tell us something we didn't know...

Like, maybe, if anyone has seen one of these upclose at conventions or visited Curt Vendel, for instance...

Like, maybe, you could have asked "has anyone seen one of these upclose at conventions" in your original post instead of assuming we could read your mind

True. And just a suggestion, HP - a lengthy historical rundown every time you start a new topic isn't necessary.

Edited by Laner, Thu Oct 13, 2005 2:16 PM.

[Jubal]

"True. And just a suggestion, HP - a lengthy historical rundown every time you start a new topic isn't necessary."

it may not be necessary, but it is appricated by this noob =)

[KulorXL]

The 3200? I've got info on that from a Retro Gamer magazine.
::opens::
"And then came Project Sylvia, also known as Super-Stella and the Atari 3200. It retained compatibility with the 2600 while offering new possibilities through enhanced sound, graphics and an improved 10-bit processor. It also looked a lot like sandwich toaster, but that's not what killed it. When game designers were presented with the new machine to begin experimental development they found it was simply too difficult to program -- and remember, these people were accustomed to 2600 development which as good as required them to program the TV's electron beam directly! The machine was scrapped, meaning Atari still had nothing but it's ageing[sic] 2600 as its[sp] main product in the escalating console war."
Wow, their grammar and spelling are terrible. I suppose that really didn't help...
I've also got a picture of the thing, which I can scan in once I get my scanner working.

EDIT: Nevermind, it's just this one: http://www.atari-com...ypen/cx3200.jpg

Edited by KulorXL, Tue Dec 13, 2005 5:42 PM.

[Draikar]

Thanks for the photo KulorXL, kinda cool looking system like most Atari proto systems that never made it. I wish they made some for retail...

[supercat]

KulorXL, on Tue Dec 13, 2005 6:41 PM, said:

"And then came Project Sylvia, also known as Super-Stella and the Atari 3200. It retained compatibility with the 2600 while offering new possibilities through enhanced sound, graphics and an improved 10-bit processor. It also looked a lot like sandwich toaster, but that's not what killed it. When game designers were presented with the new machine to begin experimental development they found it was simply too difficult to program -- and remember, these people were accustomed to 2600 development which as good as required them to program the TV's electron beam directly! The machine was scrapped, meaning Atari still had nothing but it's ageing[sic] 2600 as its[sp] main product in the escalating console war."

If I were designing a "2600-plus console", here's what I'd do...

-1- Have an unused TIA register enable "deluxe" mode

-2- In deluxe mode, divide phi by 2 rather than 3.

-3- Add new registers for the ball color, missile colors, and right half of the playfield. Add new addresses to write only to the new addresses or only to the old (the existing addresses will write to both).

-4- Double the effective clock rate for the audio chain, but extend the dividers (AUDF0/AUDF1) to eight bits. When writing in non-deluxe mode, strip off the top 3 bits.

-5- Add a new player/set of missiles (only enabled in deluxe mode)

-6- Increase the period of the horizontal position counters to 168; stuff eight cycles during the eight pixels before scan line start except following an HMOVE. When in non-deluxe mode, keep the 8-pixel blanking even though it's no longer functionally needed (in deluxe mode, eliminate it).

-7- Add a vertical counter which can provide automatic VSYNC. Make the counter readable (add data bus drivers to D0-D5, but only enable them in deluxe mode).

-8- Add some new NUSIZx options, including the ability to 'clone' the player but not the missile, and to have more than three players (see #9) or back-to-back players (see #10)

-9- Add an option to auto-load a delayed sprite shape register immediately following the display of a sprite, and an option to store to both regular and delayed sprites simultaneously. Combined with the phi/2 clock speed, this would allow an 80-pixel hires sprite.

-10- (Getting really fancy here) Allow "auto-load" mode to be turned on for a sprite which would cause the GRPx to be snatched from the bus every fourth cycle. Thus, a sequence:

  sta AUTON0
  LDA Bitmap0,x
  LDA Bitmap1,x
  LDA Bitmap2,x
  LDA Bitmap3,x
  LDA Bitmap4,x
  LDA Bitmap5,x
  LDA Bitmap6,x
  LDA Bitmap7,x
  LDA Bitmap8,x
  LDA Bitmap9,x
  sta AUDOFF0

could display an 80-pixel wide picture using just one sprite.

I would think those changes could allow easy compatibility with the existing 2600 cartridges (the only one that might break something would be #6), but they should be fairly inexpensive yet powerful. Obviously a couple decades too late, though.

[Curt Vendel]

If you'd like to see more photo's, plus the actual text that Retro gamer took off of my site:


http://www.atarimuse...via/sylvia.html

Did they at least credit my site where they plagerized their content and photo's from?

I have the epoxy mockup controller and the empty mockup shells.


Curt

KulorXL, on Tue Dec 13, 2005 6:41 PM, said:

The 3200? I've got info on that from a Retro Gamer magazine.
::opens::
"And then came Project Sylvia, also known as Super-Stella and the Atari 3200. It retained compatibility with the 2600 while offering new possibilities through enhanced sound, graphics and an improved 10-bit processor. It also looked a lot like sandwich toaster, but that's not what killed it. When game designers were presented with the new machine to begin experimental development they found it was simply too difficult to program -- and remember, these people were accustomed to 2600 development which as good as required them to program the TV's electron beam directly! The machine was scrapped, meaning Atari still had nothing but it's ageing[sic] 2600 as its[sp] main product in the escalating console war."
Wow, their grammar and spelling are terrible. I suppose that really didn't help...
I've also got a picture of the thing, which I can scan in once I get my scanner working.

EDIT: Nevermind, it's just this one: http://www.atari-com...ypen/cx3200.jpg

[Retro Archaeologist]

Curt Vendel, on Tue Dec 13, 2005 11:37 PM, said:

If you'd like to see more photo's, plus the actual text that Retro gamer took off of my site:


http://www.atarimuse...via/sylvia.html

Did they at least credit my site where they plagerized their content and photo's from?

I have the epoxy mockup controller and the empty mockup shells.


Curt

KulorXL, on Tue Dec 13, 2005 6:41 PM, said:

The 3200? I've got info on that from a Retro Gamer magazine.
::opens::
"And then came Project Sylvia, also known as Super-Stella and the Atari 3200. It retained compatibility with the 2600 while offering new possibilities through enhanced sound, graphics and an improved 10-bit processor. It also looked a lot like sandwich toaster, but that's not what killed it. When game designers were presented with the new machine to begin experimental development they found it was simply too difficult to program -- and remember, these people were accustomed to 2600 development which as good as required them to program the TV's electron beam directly! The machine was scrapped, meaning Atari still had nothing but it's ageing[sic] 2600 as its[sp] main product in the escalating console war."
Wow, their grammar and spelling are terrible. I suppose that really didn't help...
I've also got a picture of the thing, which I can scan in once I get my scanner working.

EDIT: Nevermind, it's just this one: http://www.atari-com...ypen/cx3200.jpg

Well at least your site is listed in the Links , can't tell if your mentioned for the photo can't read the langauge the page is in.

RA

[sukotsu9]

supercat, on Tue Dec 13, 2005 5:14 PM, said:

If I were designing a "2600-plus console", here's what I'd do...

I'm aware of supercat's programming savvy (Strat-O-Gems anyone?), and I know others on the AA forums are plenty talented. What sometimes amazes me is why 20 years ago a project would be scrapped because it's "too difficult to program for" but nowadays you can get some wicked hacks, tricks & homebrews from an un-official web community. Is it the interveneing 20 years of technological immersion that has advanced our programming capabilities? When I read about the old days (I did BASIC & Pascal back then), I feel like I'm reading about the Cro-Magnon era.

But I'd still rather they released the 7800 earlier than a super-2600 (System X).

[supercat]

sukotsu9, on Wed Dec 14, 2005 7:10 PM, said:

But I'd still rather they released the 7800 earlier than a super-2600 (System X).

I'll admit I've not tried actually programming the 7800 yet, but as someone with a C64 background the 7800 seems unnecessarily complicated and limited. Things I'd like to see:

-1- Allow for 320 mode to be used on lines with a nice number of colors. The cheapest way to do this might be to allow separate palette entries to be specified for the left and right half of each pixel (so if four palette entries are programmed as black-white, white-black, black-black, and white-white, that would reduce by four the number of available colors other than black and white, and would only allow black and white to appear "paired", but would still allow much more versatility than the existing system. Having a 320x5 line buffer would be better, but would cost more.

-2- For each item, include an address mask and address match register; if any bit in an item's address disagree with the address match register, and the corresponding mask bit is set, don't draw the item. This feature would eliminate the need to pad display items in memory beyond a power-of-two size, and would thus allow display "zones" to be larger.

-3- (Someone tell me if the system already does this) Use phi1 clocks alone whenever possible for display fetches, to avoid slowing down the main CPU. My suggested implentation would be to allow a display list to specify one of three modes: (1) lazy mode--always use phi1 only; if stuff doesn't get displayed in time, it doesn't get displayed; (2) burst mode--always use both phi1 and phi2; (3) mixed mode--use phi1 for the first part of the scan line, and switch to burst mode after a certain amount of time. Depending upon the number of items to be displayed, any might be best. Ideal would be to have the system use burst mode unless/until the number of remaining memory fetches fell below the number of remaining cycles, but that would seem tricky. Closest implementation I can think of to that would be to use an extra line of display buffering. Use lazy mode whenever there's a full line of data all set to go out to the display. If things start to fall behind, switch to burst mode to finish off the current line, then switch back to lazy mode once it's done. This would yield optimal performance (no CPU lag unless there are more than 114 fetches on a scan line, but able to accommodate up to 220 if needed).

-4- Something better than TIA sound.

[Bruce Tomlin]

As long as we're bitching about the 7800, my number one wish would be a 256 pixel wide mode. 160 is just way too low, and 320 isn't bad, but it's high enough to cause artifacting problems and takes more memory. Also, 256 pixels wide is just about right to get squarish pixels. And better use of the color registers such that higher resolution modes wouldn't ignore half of the available color registers.

Number two would be to somehow keep the Maria from slowing down the CPU so badly. It's too bad they didn't put DRAM in there on a dedicated video bus. I'm pretty sure that 16Kx4 DRAMs were around then, and it wouldn't have taken any more space on the board than one of the SRAM chips. And any unused DRAM could be used as secondary storage. This is the main strong point of the Colecovision.


Back on topic, the "too difficult to program for" part was probably mostly because it would be a completely new CPU instruction set that came from out of nowhere. The 1610 in the Intellivision was unique enough already, but Mattel didn't come up with it, GI did.

[supercat]

Bruce Tomlin, on Thu Dec 15, 2005 1:33 AM, said:

Number two would be to somehow keep the Maria from slowing down the CPU so badly.  It's too bad they didn't put DRAM in there on a dedicated video bus.  I'm pretty sure that 16Kx4 DRAMs were around then, and it wouldn't have taken any more space on the board than one of the SRAM chips.  And any unused DRAM could be used as secondary storage.  This is the main strong point of the Colecovision.

There are pros and cons to having the video share the bus, but it should be done in such a way as to minimize overhead. The 6502 only needs the bus half the time, so a properly-designed video system can sneak in most or all of its cycles when the 6502 isn't using it. The Apple II, Commodore VIC-20, and Commodore 64 all use this approach; I would guess that the Atari 8-bits and the Commodore PET probably did as well but I don't really know. Using this approach would require moderately fast RAM/ROM (probably 200ns for reliable operation at 3.58Mhz) but I don't think that speed was outrageous when the 7800 was in development. Had bus interleaving been implemented, the MARIA could have fetched 114 bytes/line without slowing down the main CPU at all.

On a related note, I think using an 8Kx8 RAM instead of a couple 2Kx8's would have been good also.

[Bruce Tomlin]

supercat, on Thu Dec 15, 2005 3:08 AM, said:

There are pros and cons to having the video share the bus, but it should be done in such a way as to minimize overhead.  The 6502 only needs the bus half the time, so a properly-designed video system can sneak in most or all of its cycles when the 6502 isn't using it.  The Apple II, Commodore VIC-20, and Commodore 64 all use this approach; I would guess that the Atari 8-bits and the Commodore PET probably did as well but I don't really know.  Using this approach would require moderately fast RAM/ROM (probably 200ns for reliable operation at 3.58Mhz) but I don't think that speed was outrageous when the 7800 was in development.  Had bus interleaving been implemented, the MARIA could have fetched 114 bytes/line without slowing down the main CPU at all.

Yes, you're right, not sure why I shied away from mentioning that. Maybe they couldn't use it because of Atari 2600 compatibility reasons. BTW, the original Macintosh also used bus interleaving, but it was mostly due to most 68000 instructions taking two cycles and not needing the bus for the second cycle.

Quote

On a related note, I think using an 8Kx8 RAM instead of a couple 2Kx8's would have been good also.

I just know that Atari had a good deal on 2Kx8 parts. Atari was notoriously cheap when it came to chips, except maybe when they were trying to monopolize fab production of mask ROMs. Tramiel only made things worse.

Edited by Bruce Tomlin, Thu Dec 15, 2005 12:22 PM.

[djmips]

regarding the 2600+, I'd want to add the facility to allow smooth horizontal scroll of the playfield. I'd also like to increase the horizontal playfield resolution in deluxe mode.

With regard to why certain console designs end up being ill fitted for their mission, I've seen a pattern repeat itself over the years. The people responsible for designing and building the console hardware are not the same people who program the games, and there can be a lack of communication.

[supercat]

djmips, on Sun Dec 18, 2005 1:13 AM, said:

With regard to why certain console designs end up being ill fitted for their mission, I've seen a pattern repeat itself over the years. The people responsible for designing and building the console hardware are not the same people who program the games, and there can be a lack of communication.

When I designed the 4A50 bankswitching method (which hopefully I'll get back to working on once Strat-O-Gems is out) I tried to combine my knowledge of what a programmer would find useful with my knowledge of how hardware works. Some of the stuff in there is IMHO pretty neat.

I think my favorite feature is having accesses certain addresses in RIOT RAM select the a RAM bank specified on the data bus. For example, the bank-switchable area at $1Exx is associated with a few addresses including $FD and $FF. If address $FF holds the value $12 and $FD holds the value $34, the instruction "NOP $FF" will cause hardware addresses $1200-$12FF to be visible at address $1E00-$1EFF, while the instruction "NOP $FD" will cause hardware addresses $3400-$34FF to be visible there. "INC $FD" will cause addresses $3500-$35FF to be visible there and update address $FD for next time. Note that writes in the $01xx address range will not trigger these bank-switch locations.

[~llama]

Wait, wasn't the 7800 designed by GCC? I'm pretty sure they're programmers... *cough* Ms. Pac-Man *cough*

I tried messing with 7800 programming but couldn't really understand DLL's and all that stuff. I'm going to tackle it over the summer when I don't have college to worry about.