was there a reason why atari limited the memory size of 2600 games

[carmel_andrews]

Following on from a similar thread where it mentioned that theoretically you could have 512k or more 2600 games, bought me to ask the question

 

Was there a reason why atari limited the memory size of 2600 games (cartridges)

 

Now, i remember a something mentioned in the Scott Cohen book 'Zap' where he said that Atari could only get technology (games) out in the marketplace at a 'consumer friendly' price point

 

I don't quite buy that argument as, companies like coleco and mattel were producing games for their systems that had bigger memory and were'nt selling for that much more then 2600 games

 

And as companies like arcadia/starpath proved with things like super charger...if you make games with larger memory requirements you can end up with better games...and not some of the dross that ended up on the 2600

 

perhaps if atari invested more in the 2600 by developing games with larger memory requirements, the 2600 might not have suffered as badly from the crash becasue atari would have been developing better 2600 games and also encouraged other software developers to take the same lead

 

what do you peeps think

[SeaGtGruff]

what do you peeps think

From what I've heard and read, the 2600's design *was* primarily geared toward keeping the price down, and it was expected to have a fairly short lifespan in the marketplace, to be replaced by something else-- something bigger and better-- within a few short years. Atari reportedly had no idea in those early days that the 2600 would end up having such an incredible lifespan.

 

Also, in those earliest of days when it was being designed, 4K probably seemed like a *lot* of memory-- consider how many of those early games were 2K cartridges. The whole gamut of ideas about what a video game could be and could do hadn't really evolved yet, so they may not have had much of an idea how or why a video game would ever need to be more than 4K.

 

The 6507 could address 8K, and only a few addresses (comparitively speaking) are used from the first 4K of address space, so it seems like they should have been able to restrict the TIA and RIOT to the first 1K or less of address space, which would theoretically have left 7K or more of address space for the cartridge slot.

 

However, the connections between the chip selects and the address lines must be taken into consideration, too. It takes only one address line to distinguiish between the upper 4K and the lower 4K of an 8K address space, so if you "sacrifice" pin A12 for that purpose, you can have a 4K cartridge area, and the other 4K could be used for other things-- onboard registers, onboard RAM, or even onboard ROM.

 

Michael

[kskunk]

Was there a reason why atari limited the memory size of 2600 games (cartridges)

 

In the 80s there were 2600 games larger than 4K, thanks to bank switching.

 

Given ROM prices in 1977, 4K was just a practical size. The NES came out in 1983 and had a 32K limit (8 times bigger), even though ROM prices were actually closer to 1/16th the cost by then. Of course bank switching was used on the NES too.

 

Newer consoles always support more detailed and colorful graphics, which require more space. So it's logical that newer consoles have bigger ROMs. It just takes bigger ROMs to push newer hardware to the limit. We've seen games that pushed the 2600 close to the limit in 16K and less. (I'm sort of a Solaris fan...)

 

perhaps if atari invested more in the 2600 by developing games with larger memory requirements, the 2600 might not have suffered as badly from the crash becasue atari would have been developing better 2600 games and also encouraged other software developers to take the same lead

 

The 2600 hardware was really antiquated by 1983. By 1983 we had NES quality hardware on the market. No amount of ROM would allow the 2600 to deliver that kind of generational leap.

 

And despite technical factors, the business issues were just a much bigger problem. Even if Atari had not been so irrationally exuberant, they could not have stopped all the 3rd parties who were eagerly overinflating the video game bubble.

 

- KS

Edited July 11, 2009 by kskunk

[kool kitty89]

Was there a reason why atari limited the memory size of 2600 games (cartridges)

 

In the 80s there were 2600 games larger than 4K, thanks to bank switching.

 

Given ROM prices in 1977, 4K was just a practical size. The NES came out in 1983 and had a 32K limit (8 times bigger), even though ROM prices were actually closer to 1/16th the cost by then. Of course bank switching was used on the NES too.

 

Bankswitching requires additional hardware on the cartridge which increases cost, right? (in addition to the ROM costs)

 

BTW, what does the 12-pin ribbon cable running from the cartridge slot/switch board to the motherboard have to do with things? (I'll have to open my system up again, but I think that's the only connection from the cartridge to the main/motherboard, the other 12 pins could only be going to the switchboard, assuming they're all used)

 

Also, I seem to remember refrenced on the NES cartridges mentionin that it supported more than 32 kB of game ROM. (somthing like 48 kB I think, and I'm not confusing it with the 7800) But maybe this was referring to the overall cartridge address space, not just game ROM. (ROM/RAM for video and main RAM expansion, plus SRAM battery save mechanisms)

 

Edit: found this: http://nocash.emubase.de/everynes.htm The 49,128 bytes figure is including 8 kB reserved for SRAM and 8,164 bytes for "Cartridge Expansion Area" (?). The PPU addressing isn't included with this.

 

Newer consoles always support more detailed and colorful graphics, which require more space. So it's logical that newer consoles have bigger ROMs. It just takes bigger ROMs to push newer hardware to the limit. We've seen games that pushed the 2600 close to the limit in 16K and less. (I'm sort of a Solaris fan...)

 

I find Pitfall 2 rather impressive personally (not sure how much it actually pushes the hardware, but the multicolor sprites, animated water and continuous letalone dynamic soundtrack), though Solaris is definitely impressive. (I don't have either yet though)

 

 

And despite technical factors, the business issues were just a much bigger problem. Even if Atari had not been so irrationally exuberant, they could not have stopped all the 3rd parties who were eagerly overinflating the video game bubble.

 

Didn't Atari's semi-monopoly on ROM ICs give them some control over 3rd parties?

Edited July 11, 2009 by kool kitty89

[kskunk]

Bankswitching requires additional hardware on the cartridge which increases cost, right? (in addition to the ROM costs)

 

Yes, but it's a small percentage of ROM cost. On small batches (under 100K cartridges), it's an extra 7 cents per-unit for a TTL latch. On larger batches you can just roll it into your mask ROM for a one-time cost of $3-5K and then it's free per-unit.

 

Pitfall II integrated its bank switching into a mask ROM chip. (Along with a lot of other stuff.)

 

With the benefit of hindsight, it seems like Atari made the right call. Not many games needed more than 4K, and most of the ones that did came out long after Atari should have had moved on to the next gen.

 

BTW, what does the 12-pin ribbon cable running from the cartridge slot/switch board to the motherboard have to do with things?

 

Putting bank switching in the console requires a larger cartridge slot and larger ribbon cable, and thus larger cartridges that require larger shells made from larger tools and larger PCBs. It increases the cost for everything on the off chance a game might use it. (Which must have seemed unlikely in 1977!)

 

Also, I seem to remember refrenced on the NES cartridges mentionin that it supported more than 32 kB of game ROM. (somthing like 48 kB I think.

 

You're correct, but only 32KB program ROM is available without additional decoding hardware. In other words, to map the other 16KB to additional program ROM requires extra logic that is no different than bank switching logic. (But maybe the software guys will like it more than bank switching.)

 

Didn't Atari's semi-monopoly on ROM ICs give them some control over 3rd parties?

 

I'm not sure what you mean. There were dozens of semiconductor companies in the late 70s/early 80s that offered mask ROM services.

 

- KS

[kool kitty89]

Yes, but it's a small percentage of ROM cost. On small batches (under 100K cartridges), it's an extra 7 cents per-unit for a TTL latch. On larger batches you can just roll it into your mask ROM for a one-time cost of $3-5K and then it's free per-unit.

Thanks, that's less significant than I'd have guessed.

 

 

BTW, what does the 12-pin ribbon cable running from the cartridge slot/switch board to the motherboard have to do with things?

 

Putting bank switching in the console requires a larger cartridge slot and larger ribbon cable, and thus larger cartridges that require larger shells made from larger tools and larger PCBs. It increases the cost for everything on the off chance a game might use it. (Which must have seemed unlikely in 1977!)

 

I wasn't asking about that, I was asking whether the 12-pin ribbon cable was part of the limitation/reason for the 4 kB window for cartridge rom. I'd assume so as this apears to be the only connectivity between the cartridge slot and motherboard, so they'd have to be the 12 address pins. It came up in another thread that all 24 pins on the 2600 cartridge slot were connected and used, but (while mu current techincal knolege is rather limited) it seems a bit odd that you'd need all additional 12 pins connected to the switchboard.

 

That said, had they expanded the cartridge bus any I'd think it most reasonable to just go up to 13-bit and address as much of the 6507's 8 kB space as possible. (inless all of the remaining 4 kB is already reserved/used)

 

Didn't Atari's semi-monopoly on ROM ICs give them some control over 3rd parties?

 

I'm not sure what you mean. There were dozens of semiconductor companies in the late 70s/early 80s that offered mask ROM services.

 

- KS

 

I remember severl mentions of such in older threads I was reading through on atariage (I seem to remember some of it was in the context of Nolan Bushnell, but I'm not sure), while I can't seem to find any of those discussions atm, I recently had a related discussion on sega-16:

http://sega-16.com/forum/showthread.php?t=672&page=20

 

Atari didn't hold any sort of patents on rom technology that was used for the cartridges. Atari did however, own most of the major RAM manufacturers that provided the chips inside those cartridges. Even so, you had Time Warner running a business they knew nothing about. If someone wanted to make a game, all Time Warner saw was $$$$. There was no quality control and Atari helped distribute the product to retail outlets in exchange for a royalty check.

 

If someone wanted to make a cartridge outside of Atari's manufacturing plant, they usually had to find a RAM chip manufacturer (most were located inside Silicon Valley) outside of the US. So it wasn't just some hack creating illegal software for the 2600, it was mostly Atari allowing anyone with a game and cash in hand to make a game for their system.

 

The only legal way to manufacture cartridges on the Master System and NES, was to (not so obviously) reverse engineer the lockout technology used inside the cartridges for those systems. So, the likelyhood of someone making an unapproved cartridge became pretty slim. Color Dreams found a way and Tengen would have, if they didn't make a mistake that showed they were illegally disabling the lockout.

 

Yes, Nintendo licensed some pretty crappy games. But they never went overboard like the sales people did at Atari. The NES has 2.5 times more consoles sold than the 2600 did, and the 2600 probably has a couple hundred more titles.

 

You should check out this book It really sheds some light on the early days of the industry, and it goes into deeper detail than what you'll find in Game Over. Honestly, it's a great read!

Edited July 12, 2009 by kool kitty89

[cwilkson]

Just a quick note to clarify the 2600 cart port usage. There are 13 adress lines, 8 data lines, a +5v power line, and 2 ground lines for a total of 24. The dual ground lines were to support the planned but unused built in ROM feature of the console. Note that there is no clock, or chip enable, or read/write line. The decision to omit these was based on a cost savings of 50 cents in the edge connector. (And it was later considered by the designers to be a huge mistake. ) These missing control lines limit cartridges to ROM only without the addtion of extra logic in the cart. And because there is no cart enable line, one of the 6507's address lines is sacrificed for that function. That's why it's a 4kB ROM space max instead of 8kB.

 

BTW, which 2600 model uses a ribbon cable to connect the cart port to the motherboard? Every one I've seen has the cart edge connector directly soldered to the motherboard.

[Wickeycolumbus]

BTW, which 2600 model uses a ribbon cable to connect the cart port to the motherboard? Every one I've seen has the cart edge connector directly soldered to the motherboard.

 

AFAIK, none of them do. I think he is talking about this, there appears to be a red ribbon cable connecting the switches to the main board.

[Polybius]

It's all about the money!

 

 

thats why Atari was releasing 2k games untill Activision Started releasing their own games and putting Atari to shame

 

 

Healthy competition keeps the business world going.

 

 

!!!!!!!!!!Long live Activision, the world's First 3rd party Game devloper!!!!!!!!!!

[Polybius]

BTW, which 2600 model uses a ribbon cable to connect the cart port to the motherboard? Every one I've seen has the cart edge connector directly soldered to the motherboard.

 

AFAIK, none of them do. I think he is talking about this, there appears to be a red ribbon cable connecting the switches to the main board.

 

Maybe on the Juniors, as all the other ones have their switches directly soldered to the board.

[kskunk]

The decision to omit these was based on a cost savings of 50 cents in the edge connector. (And it was later considered by the designers to be a huge mistake. )

 

50 cents is more than I expected to drop 2-4 pins off a connector. I guess the small pinout allowed them to move to a standard/high volume connector in place of a custom/pricey connector.

 

I've been involved in cost-down efforts before, and usually there are dozens of 50 cent savings to be had. Many of them just plain suck from an engineering 'correctness' perspective. It's very hard to predict which one(s) you'll regret later, but if you don't relentlessly cut every cost you end up losing in the market.

 

I doubt that management regretted the lack of a read/write line much.

 

- KS

[kool kitty89]

AFAIK, none of them do. I think he is talking about this, there appears to be a red ribbon cable connecting the switches to the main board.

 

Oh crap, you're right, I've been making an idiot out of myself... I should have taken another look inside one of my 2600's... (they're both in our storgage room at the moment na dwe've been doing some reorganizeng and it's kind of burried ATM). The cable on my Heavy Sixer is blue though, intestering that my light sixer (pretty sure it's a Hong Kong one) has what seems to be a nicer (and incompatible) cable. (the heavy's is a single peice flat plastic ribbon cable with flat aluminum pins, the light's has indivitual insulated wires fused together with a plastic conntector at either end)

 

Anyway, thank you very much for that correction, that makes a whole lot more sense.

 

Just a quick note to clarify the 2600 cart port usage. There are 13 adress lines, 8 data lines, a +5v power line, and 2 ground lines for a total of 24. The dual ground lines were to support the planned but unused built in ROM feature of the console. Note that there is no clock, or chip enable, or read/write line. The decision to omit these was based on a cost savings of 50 cents in the edge connector. (And it was later considered by the designers to be a huge mistake. ) These missing control lines limit cartridges to ROM only without the addtion of extra logic in the cart. And because there is no cart enable line, one of the 6507's address lines is sacrificed for that function. That's why it's a 4kB ROM space max instead of 8kB.

 

Thanks for the nice explanation.

 

So, had they not reserved the pin for the second ground they could have used the full 13-bit address bus. (or added a read/write line)

How did they manage to include on-cart RAM expansion w/out a read/write line?

Edited July 13, 2009 by kool kitty89

[cwilkson]

So, had they not reserved the pin for the second ground they could have used the full 13-bit address bus. (or added a read/write line)

How did they manage to include on-cart RAM expansion w/out a read/write line?

There were different methods used. The simplest was to use yet another address line. So for example addresses $000~0FF would be used as the write port and addresses $100~1FF would be the read port, using A8 as the read/write line. Additional logic in the cartridge would be used to time the necessary signals locally.

[kskunk]

The dual ground lines were to support the planned but unused built in ROM feature of the console.

 

Interesting stuff! Do you know the details of how the built-in ROM feature was planned to work?

 

- KS

[cwilkson]

The dual ground lines were to support the planned but unused built in ROM feature of the console.

 

Interesting stuff! Do you know the details of how the built-in ROM feature was planned to work?

 

- KS

I don't know the exact details, but the basic idea is that one was a true ground and the other was an input from the cartridge. Let's call that the NO_CARTRIDGE line. Without a cartridge plugged in, the NO_CARTRIDGE line would be connected to +5V through an internal resistor. With a cartridge plugged in, it would be shorted to ground. So this signal could be used as an enable line for the internal ROM.

[carmel_andrews]

Perhaps for the 2600 to compete with the NES etc (higher mem capacity games and all) they could have released a 2600 with a 6502 and removed the limitation on accessing the cartridge port (which i guess is something to do with the 6507 only accessing so much memory)

[cwilkson]

Perhaps for the 2600 to compete with the NES etc (higher mem capacity games and all) they could have released a 2600 with a 6502 and removed the limitation on accessing the cartridge port (which i guess is something to do with the 6507 only accessing so much memory)

I think that's what you get with the 7800.

[carmel_andrews]

No, because you have additional hardware (maria)

 

I was thinking just a plain vanilla flavoured 2600 but with a 6502 and 64k games (or 48k even)

 

would adding more ram to the 2600 m/b improve the gfx as i read somwhere that a stock 2600/vcs can only draw/see half the screen

[cwilkson]

No, because you have additional hardware (maria)

 

I was thinking just a plain vanilla flavoured 2600 but with a 6502 and 64k games (or 48k even)

 

would adding more ram to the 2600 m/b improve the gfx as i read somwhere that a stock 2600/vcs can only draw/see half the screen

(first I appologize if I ramble. I haven't slept much for the last 2 weeks because of CAX. It was fun, but I'm exhausted now. But I'll try to answer this now before I finally sleep.)

 

Not in 2600 mode. That's what I meant. I'm not sure how the 7800 memory map works, but I know it's running a 6502 and MARIA is disabled for 2600 games. I'm not sure what the article in question was saying. But here's my very abbreviated take. It's not really appropriate to talk about a "screen" for the 2600. It is a scanline based architecture. It only draws one scanline at a time. In the simple case, you load the graphics registers, wait for the line to be drawn, then reload the registers with new data. Etc. In more elaborate cases, you load the registers exactly one clock cycle before they are used. If one must refer to graphics memory you can think of the 2600 as having less than 5 bytes. Plus a few primitive hardware tools for manipulating those bytes. The 6507 has to constantly update these registers, at *exactly* the correct time with single clock cycle precision to draw an interesting display. So the biggest bottleneck is processor cycles. You can't do anything during the active picture because you're handfeeding the audio and video to the TV. More ROM/RAM lets you make bigger games that are more complex logically. And by creatively paging that memory in/out addresses $000-$ff you can get a few more processor operations. But really because of the just-in-time nature of the TIA (i.e. no frame buffer) you really need to increase the clock speed of the 6507. (or 6502)

 

64k and larger games with oodles of RAM are not that hard to make inside the cartridge, from a hardware standpoint. They are just expensive for us hobbiests because we're only making a few. But for a company with $$$ to make custom chips by the millions it's not that hard or expensive. (Writing the games of course is another matter.) Sadly, I just don't think the 2600 has the horsepower to compete with the NES. Which is understandable....I would consider them 2 generations apart.

[BigO]

Super Atari: http://www.baroquegaming.com/projects/supe...cal_details.htm

Been around awhile, but seems relevant to this discussion.

Edited July 14, 2009 by BigO

[+Random Terrain]

By 1983 we had NES quality hardware on the market.

Do you mean the knowledge that NES quality hardware existed in the world or that people in the USA could run out and buy their own NES? For any young newbies out there reading this thread who may not know, the NES wasn't available nationwide in the USA until 1986:

 

http://en.wikipedia.org/wiki/History_of_th...tainment_System

 

http://www.nintendoland.com/nes/history.htm

[mos6507]

I was thinking just a plain vanilla flavoured 2600 but with a 6502 and 64k games (or 48k even)

 

They could have done that in 1980 or 81. That would have been a wise move, probably, however it would have alienated the original installed base of 2600 users. Remember that videogames were really just out of the starting gate by then. Even the 2600 was a big purchase decision. People were not really ready for the hardware upgrade treadmill which is so routine today. If they were going to offer a next gen system it was going to have to be truly next gen and we all know how that played out.

 

The funny thing when interviewing the Atari guys is how many of them (thought not all) don't really have the same sentimental attitude towards the 2600 that we do. We look at the 2600 as an icon of success but to them it was also an icon of mismanagement. That's because they saw all the great stuff that was on the horizon that didn't come out. The 2600 wasn't supposed to be Atari's best foot forward in 1982 or 1983. It just wound up that way by default since Warners had no good strategy for how to follow it up. Companies like Activision and Imagic made lemons into lemonade which everyone else just surfed on.

[mos6507]

64k and larger games with oodles of RAM are not that hard to make inside the cartridge, from a hardware standpoint.

 

If you want to talk historically, it was a big leap at the time. I don't think anyone really knew you could manage to get RAM in a cart without the read/write lines back in '77. And based on what Larry Wagner said, I think the general thinking at the time was that banking hardware was going to be too tricky to implement until he came up with a way to do it with a "couple of ICs" as he said. So what seems obvious to us now required some technical breakthroughs.

[+batari]

64k and larger games with oodles of RAM are not that hard to make inside the cartridge, from a hardware standpoint.

 

If you want to talk historically, it was a big leap at the time. I don't think anyone really knew you could manage to get RAM in a cart without the read/write lines back in '77. And based on what Larry Wagner said, I think the general thinking at the time was that banking hardware was going to be too tricky to implement until he came up with a way to do it with a "couple of ICs" as he said. So what seems obvious to us now required some technical breakthroughs.

Bankswitching hardware is sort of trivial. I wonder if any engineer really had trouble with it.

 

It's not just the R/W line but the lack of phi2 that makes adding RAM to a lot harder than just switching banks. I'm not sure what cwlikson meant by "not that hard" but it is a lot harder than bankswitching. Certainly it's not an amazingly difficult problem but it is not trivial.

 

The lack of R/W line is easily worked around by having separate address for reads and writes. But, no phi2 signal means that you have to use accurately timed circuitry to simulate phi2 to hit the small window where data is valid for writing between cycles. Worse, there are actually two windows you need to hit as some instructions write on the first cycle and some write on the second.

 

Alternatively you can latch the address bus on writes and wait for the address to change, then grab the data and use the latched address to know where to write it. This method requires much more hardware than regular bankswitching.

 

Also, neither of the above RAM methods can be gleaned from simply examining 6507 datasheets. Atari had to do experiments to figure out the timing.

 

The document here shows some of the research done. Although this was done for the 2600 computer add-on, as I understand the research ultimately resulted in the SARA chip:

http://www.atarimuseum.com/archives/pdf/vi...esign_notes.pdf

[Nukey Shay]

The funny thing when interviewing the Atari guys is how many of them (thought not all) don't really have the same sentimental attitude towards the 2600 that we do. We look at the 2600 as an icon of success but to them it was also an icon of mismanagement. That's because they saw all the great stuff that was on the horizon that didn't come out.

Perhaps being screwed over one too many times also plays a bit on this lack of sentiment toward the 2600?