Designed a working 272K bankswitched board.

[CPUWIZ]

Should I manufacture these, would anyone want to take advantage of this sort of thing?

 

It works like this:

 

You have a fixed bank of code from $8000-$BFFF (16K) and you have 16 x 16K banks at address $4000-$7FFF, accessable by simply touching address lines $BFC0 through $BFCF. The bank's data is accessable within a cycle after touching the address line.

 

Please ignore my little jumper wire, I hastily moved a VIA before sending it off, without noticing it crossing one of the INTERLOCK lines.

[mos6507]

I think you reach a point of diminishing returns with ROM banking. I think this much ROM really needs at least a little RAM to help balance it out.

[gdement]

I'm not a 5200 programmer, so my wishes don't matter.

 

But what about having the fixed portion be a RAM chip? The programmer can always copy frequently-used code to the fixed bank RAM, but has the option of doing more volatile things with it as well.

[Shawn]

Well this board is sure gonna make BBSB a little easier to get a "copy" of

[CPUWIZ]

I'm not a 5200 programmer, so my wishes don't matter.

 

But what about having the fixed portion be a RAM chip? The programmer can always copy frequently-used code to the fixed bank RAM, but has the option of doing more volatile things with it as well.

 

If you can show me how you would implement a non-existing WE line, I'd love to add RAM to a cart.

 

This board is not made to handle BBSB.

 

I figured it was a shot in the dark, most people aren't really that interested in the 5200.

[Curt Vendel]

I think there is interest, I'm actually disappointed there isn't more coding for the 5200 done since its an 800 with a different memory mapping and its got just as many coding tools as the 2600, just needs more interested coders.

 

 

What we need to see are several ports never available -

 

Adventure on the 5200 (yes I know we have ADV2 coming....)

Yars' Revenge

Caverns of Mars

 

and there are a TON of others just begging to be done on the 5200.

 

 

Nice board CPU!

 

 

 

Curt

 

I'm not a 5200 programmer, so my wishes don't matter.

 

But what about having the fixed portion be a RAM chip? The programmer can always copy frequently-used code to the fixed bank RAM, but has the option of doing more volatile things with it as well.

 

If you can show me how you would implement a non-existing WE line, I'd love to add RAM to a cart.

 

This board is not made to handle BBSB.

 

I figured it was a shot in the dark, most people aren't really that interested in the 5200.

[+Allan]

I think there is interest, I'm actually disappointed there isn't more coding for the 5200 done since its an 800 with a different memory mapping and its got just as many coding tools as the 2600, just needs more interested coders.

 

 

What we need to see are several ports never available -

 

Adventure on the 5200 (yes I know we have ADV2 coming....)

Yars' Revenge

Caverns of Mars

 

and there are a TON of others just begging to be done on the 5200.

 

 

Nice board CPU!

 

 

 

Curt

 

I'm not a 5200 programmer, so my wishes don't matter.

 

But what about having the fixed portion be a RAM chip? The programmer can always copy frequently-used code to the fixed bank RAM, but has the option of doing more volatile things with it as well.

 

If you can show me how you would implement a non-existing WE line, I'd love to add RAM to a cart.

 

This board is not made to handle BBSB.

 

I figured it was a shot in the dark, most people aren't really that interested in the 5200.

 

Caverns of Mars was already converted to the 5200. It's a great game.

 

Great job, CPUWIZ. I hope to program something one day to take advantage of them.

 

Allan

[jbanes]

272K?!?

 

New tagline: "More memory than Carter has little pills!"

 

All hail the (first?) 2 megabit cart for the 5200! I expect a clone of Super Mario Bros. 3 by noon tomorrow.

[Atariboy2600]

Do this mean the Star Wars Domark's version can port or convert the Domark's Star Wars Atari 8bit to the 5200?

Star_Wars_Domark.zip

[kenfused]

I'm not a 5200 programmer, so my wishes don't matter.

 

But what about having the fixed portion be a RAM chip? The programmer can always copy frequently-used code to the fixed bank RAM, but has the option of doing more volatile things with it as well.

 

If you can show me how you would implement a non-existing WE line, I'd love to add RAM to a cart.

 

This board is not made to handle BBSB.

 

I figured it was a shot in the dark, most people aren't really that interested in the 5200.

I still would prefer to have it use only one eprom. If there is not enough inputs/outputs left in the gal/pal, an AND or other 74xx IC could probably be used on a couple of the address line inputs to free up an input or two. This could then allow the gal to be programmed for other bank switching schemes and make things such as a nintendo multicart possible ( donkey kong, dk jr, popeye, and ?), or a pacman multi, all without reprogramming the games. ALso the 27128 is probably going to add a little more to the cost than an LS chip. Never-the-less I would probably use it if I got to the point of needing it. Just my 2 cents.

 

RAM would be hard to do unless read or writes can be at different addresses as is typically done on the 2600. If someone can make the read/writes work normally at the same address without the R/W line (I think someone has some scheme for doing this on the 2600?), then a lot of conversions would be possible.

[CPUWIZ]

Yeah yeah, I know you would prefer a single chip design. However 27C080's are a lot more expensive than a 27C020 and a 27C128 combined.

[classics]

2600 style cartridge ram access hardly seems worth the effort on a machine that has 16k of ram already available.

 

Cart ram doesn't seem to me like it would be very useful for porting existing software, and for new or completely re-written software 16k can go a long long way, if you are careful with it.

 

Nice PCB though, I like the xegs style banking. I have a 512k cart that only uses two chips (and can do BBSB ) but it switches 32k banks, which means a lot of wasted space duplicating fixed areas of code.

 

Steve

[kenfused]

Yeah yeah, I know you would prefer a single chip design. However 27C080's are a lot more expensive than a 27C020 and a 27C128 combined.

yea, but if I want a 64K cart, a 27512 and 27128 are more expensive than a 27512. Of course I haven't written anything yet to use either :-)

[Lord Thag]

Yeah, I think it would be a good idea. Of course, it will just take a dedicated programmer. You could do some AMAZING stuff with that much memory, but of course, it's all about the willingnes to do it. The fact that it's available might get some enterprising person to write us a really big RPG or whatever on a cart like that.

 

What really holds the 5200 back is the lack of low-cost alternatives to the standard sticks. A cheap adaptor for the 5200 would sell very well. The Redemptions are great, but pretty pricey, and (I think) it's the main reason people stay away.

[CPUWIZ]

Yeah yeah, I know you would prefer a single chip design. However 27C080's are a lot more expensive than a 27C020 and a 27C128 combined.

yea, but if I want a 64K cart, a 27512 and 27128 are more expensive than a 27512. Of course I haven't written anything yet to use either :-)

 

Now you are even stepping into the 28pin versus 32pin chip problem. I am sure people would pay $2 more for a cart using the kind of board I made.

[supercat]

If you can show me how you would implement a non-existing WE line, I'd love to add RAM to a cart.

 

I don't know what wires you get on the 5200, but I did a RAM cart on the 2600 which allows code to 'simply' write to RAM. Does the 5200 run at a solid chroma/2 clock rate (with the video stealing whole cycles), or does it do goofy things like the 7800?

[tjb]

I think there is interest, I'm actually disappointed there isn't more coding for the 5200 done since its an 800 with a different memory mapping and its got just as many coding tools as the 2600, just needs more interested coders.

 

 

What we need to see are several ports never available -

 

Adventure on the 5200 (yes I know we have ADV2 coming....)

Yars' Revenge

Caverns of Mars

 

and there are a TON of others just begging to be done on the 5200.

 

 

Nice board CPU!

 

 

 

Curt

 

 

I agree that there doesn't seem to be much interest in 5200 development compared to 2600 development. For what its worth I'm trying to do my part. I'm currently working on an Asteroids clone for the 5200 and home computers. Here's a short video:

 

rast09162006.zip

 

Getting back to the cart, I for one could use the extra memory for sprite data among other things. I could create enough Asteroid frames to smoothly rotate the Asteroids. Great work!

 

-tjb

[kenfused]

 

I don't know what wires you get on the 5200, but I did a RAM cart on the 2600 which allows code to 'simply' write to RAM. Does the 5200 run at a solid chroma/2 clock rate (with the video stealing whole cycles), or does it do goofy things like the 7800?

Not sure exactly what the question is, but it does not change clock speeds from 1.79 to 1.19? like the 7800 does when the TIA and RIOT? are accessed.

On the 2600, are there any side effects at all to the RAM access. All addressing modes work, etc?

[mos6507]

That asteroids game looks excellent!

[mos6507]

I didn't read closely enough to realize this was for the 5200. I also wasn't aware that the 5200 cart port didn't have the read/write line. Man, that's lame. They should have kept the signals the same as the Atari 8-bit, which does have it.

[supercat]

All addressing modes work, etc?

 

Yup. I'm pretty certain mine is the only cartridge ever to manage that feat on the 2600; the same approach should work on the 5200 if all memory cycles are two chroma clocks.

[Bruce Tomlin]

I still would prefer to have it use only one eprom. If there is not enough inputs/outputs left in the gal/pal, an AND or other 74xx IC could probably be used on a couple of the address line inputs to free up an input or two. This could then allow the gal to be programmed for other bank switching schemes and make things such as a nintendo multicart possible ( donkey kong, dk jr, popeye, and ?), or a pacman multi, all without reprogramming the games. ALso the 27128 is probably going to add a little more to the cost than an LS chip. Never-the-less I would probably use it if I got to the point of needing it. Just my 2 cents.

I suggest doing it the 7800 way, with the fixed bank being the last bank. Then it's just a matter of tri-stating the address outputs and using pull-up resistors. Then if you want to stick in an '080, it'll be a 1 megabyte cart.

 

RAM would be hard to do unless read or writes can be at different addresses as is typically done on the 2600. If someone can make the read/writes work normally at the same address without the R/W line (I think someone has some scheme for doing this on the 2600?), then a lot of conversions would be possible.

That is probably the most annoying thing about the cartridge slots for most machines from that era. Nobody had the vision to think that a cartridge could contain more than just an EPROM, and add the extra signal or two needed to support writes. Using special addresses for writes is a pain in the butt. I think the Intellivision was really the only significant pre-1983 game console system that could do this, and that was because of its non-standard bus. (Techinically, the Channel F could too, for the same reason.)

 

The Famicom and SG-3000 both came out in 1983, and the 7800 in 1984. (Did the SG-3000 have a WE line in its cartridge slot?)

[supercat]

I suggest doing it the 7800 way, with the fixed bank being the last bank. Then it's just a matter of tri-stating the address outputs and using pull-up resistors. Then if you want to stick in an '080, it'll be a 1 megabyte cart.

 

Given that humungous EPROMs are cheap, is there any advantage to having a shared bank, versus simply duplicating code in all the banks where you want it? Having RAM shared among various banks is useful, since what's written in one may be read in another, but having using a 512Kx8 chip to hold 16x16K banks and a 16K fixed bank is no better than using the same chip to hold 16x32K banks, where 16K of data is duplicated on half of each of 16 banks.

[Bruce Tomlin]

I suggest doing it the 7800 way, with the fixed bank being the last bank. Then it's just a matter of tri-stating the address outputs and using pull-up resistors. Then if you want to stick in an '080, it'll be a 1 megabyte cart.

Given that humungous EPROMs are cheap, is there any advantage to having a shared bank, versus simply duplicating code in all the banks where you want it?

Absolutely. The advantage is that you only have to solder one EPROM chip, not two. "Humungous EPROMs are cheap" is exactly the reason why this is better. (Unless your name is Hadden.) No matter how cheap the chip is, one of them is probably going to cost less than two of them. Do you really think someone is going to miss that extra 16K between 256K vs 272K when only one game was ever larger than 32K?

 

Basically, you have to have one bank be fixed, unless you want to 1) add a reset circuit to the cartridge and 2) require the programmer to download trampoline code to RAM. (okay, you could do two bank zones, but that would make the mapper way too complicated) Why use two EPROMs when you can use just one?

but having using a 512Kx8 chip to hold 16x16K banks and a 16K fixed bank is no better than using the same chip to hold 16x32K banks, where 16K of data is duplicated on half of each of 16 banks.

Did you even look at the first message in the thread? The original design was a 27128 + 16 x 16K.

[Bryan]

272K?!?

 

New tagline: "More memory than Carter has little pills!"

 

All hail the (first?) 2 megabit cart for the 5200! I expect a clone of Super Mario Bros. 3 by noon tomorrow.

 

The board I designed for MULE takes up to 512K. I'm sitting on a huge stack of them since I had 200 made.

 

 

 

-Bry

Edited September 24, 2006 by Bryan