Any chance of an 'Open Sourced' Atari platform (Any flavour of Atari)

[carmel_andrews]

Following on from Thordbg's thread about the built in OS in his emulator and given that the only 2 regularly maintained emulators now come with their own 'tweaked' versions of the Atari O.S, It begs the following question

 

Is there any chance of an Open Sourced Atari platform (not just the A8)

 

After all i beleive there now exists third party OpenSourced versions of TOS (so called EmuTOS)

 

And given that Atari++ and Altirra already come with their own versions of the Atari OS we are kind of partway there

 

And since both EmuTOS and the two A8 emulator OS's support the various ST/STE and A8 upgrades that are programmed into the emulators (i.e. VBXE, Hard drives, Memory expansions, Disk Drive mods and other such things) I think the impetus is there to get the various and all Atari platforms more opensourced, even games systems like the 2600/7800 and also Lynx/Jaguar etc...It might just lead onto more 3rd party development (both hardware and software)

 

After all, whilst i so of accept that Linux hasn't dismantled M$$$'s dominance of the desktop computer OS, It has really given M$$$ and sony/Nintendo serious competition in the desktop and mobile gaming OS markets (thanks largely to Google opting for Andriod, a linux derivative)

 

After all, Atari themselves aren't interested in hardware platforms anymore and anyway we don't really need someone like Atari to make an Atari Opensourced platform successful (after all I read somewhere that the original 'flashback' sold 500000 units and with virtually zero marketing or advertising from Atari themselves, given that flashback had to compete with the free to download software emulators for the 2600/7800, in my book that aint bad going considering the piss poor support atari gave it)

 

To get around the issue of copyrighted hardware (i.e Antic, Pokey and GTIA) I guess it wouldn't be too difficult to reproduce/replicate those chip features without neccessarily infringing copyright

Edited April 12, 2013 by carmel_andrews

[thorfdbg]

As for software, I can only comment on the 8-bits since I do not know enough about TOS and its open-source replacements to comment. For AtariOs, I would believe we are there "close enough", though some tweaks are probably still required (thus the request for beta-testing). Atari BASIC is likely the next thing that is going to happen, though this is harder to reach (re-implementing without copying), BASIC is pretty special.

 

As for the chips: There is the issue of patents, but I believe they no longer apply since if Atari patented them, this must have been at or before 1979, and thus the patents run out at this time. I don't think copyright applies to chip design (it might if the chips were designed from some type of program, such as VHDL, but they likely were not in 1979), so it might be even possible to create a custom lookalike of the chips in a legally clean way. Given that Amiga chips have been recently re-implemented in programmable logic, the less complicated Atari chipset shouldn't be much of a problem for a talented hardware designer. I'm just wondering whether that is really needed given that you can easily emulate all of the chipset in software nowadays.

 

Greetings,

Thomas

[Markdvdman]

The best option really has to be a Raspberry Pi.

 

I mean they do not seem to block people doing what they want with it, and if they paid a £5 license for every computer sold it would suit everyone. It would simply need a raspberry pi in a keyboard case. HDMI output is already inbuilt, so plug into the TV straight away. It would need simply a good OS. Would it need a HD? Possibly. A DVD drive maybe? The only issue maybe emulating the Falcon, but perhaps a pre overclocked variant to 1GHz will suffice there. An Archimedes is easily done due to it being an Arm chip and so much faster than most Archimedes too which was already a VERY FAST computrer in its day. I know as I owned an A3000!

 

It relies totally on software of course, but development cost would be so much easier than hardware surely?

 

I reckon it would make a killing and could be sold for around £100/$100 (depending where prioduced)

[+Gemintronic]

I thought we already had the Firebee?

http://medusacomputer.com/firebee.html

[Markdvdman]

I thought we already had the Firebee?

http://medusacomputer.com/firebee.html

 

The problem is that is not cheap and never likely to be mass market because of the price! A nice product mind

[The Usotsuki]

The RasPi already runs the Archie's OS, não?

[Markdvdman]

The RasPi already runs the Archie's OS, não?

 

Yes I know RiscOS as I have it on there. Also, there is a brilliant Atari 8bit emulator at full speed too. A LOT is going on with it - but it would need resources to get a full on open source multi Atari os on there!

[Darkhog]

I think it would be best to just make SOA Atari, pretty much like most NES/Famicom clones do. The only hurdle that we would need to overcome is to manufacture cart ports and remake OS/DOS code. We can use 6502 processor, as it is still manufactured (used in many "industrial" application, e.g. to control robots).

 

We could also find people who has copyright on OS/DOS code of Atari computers and talk them to release it (binary form would be sufficient). I don't think they care anymore about some old 8-bit code, even if it would be sold (too niche market).

[thorfdbg]

I think it would be best to just make SOA Atari, pretty much like most NES/Famicom clones do. The only hurdle that we would need to overcome is to manufacture cart ports and remake OS/DOS code. We can use 6502 processor, as it is still manufactured (used in many "industrial" application, e.g. to control robots).

The hurdle for the Os sources and the DOS no longer exists, that's already "all done". BASIC is still an open issue. *Still*

 

We could also find people who has copyright on OS/DOS code of Atari computers and talk them to release it (binary form would be sufficient). I don't think they care anymore about some old 8-bit code, even if it would be sold (too niche market).

 

This is unfortunately not so easy. Not probably to find some people, but actually to find the *rights holder*. The original sources went through many hands (Oss,Atari,Oss again...) and the question is, after these companies went bankrupt, to which other entitiy the rights were sold. It would not surprise me if the *actual* rights holder does not even know that he holds the rights to something that old. Pretty much the same happened for Amiga (anyone remember?) and in the end, the confusion was so big *who* actually had copyright on AmigaOs that the case went back and forth through court.

 

IOW, not a realistic option. Re-implement and avoid the trouble.

[Darkhog]

Yeah, but reimplement is harder. Still tho, even if we would use original roms, I doubt anyone will care, just like anyone don't care about NES/Fami clones (or even Genesis ones, like Retron3/upcoming Retron5).

 

The main problem would be manufacturing actual peripherals (not so machine itself) - prices of brand new atari floppies are insane ($20 for one floppy, but then again, it's ebay price). We would need to lower cost, so anyone can afford it. Because I doubt anyone would want to buy Atari clone for original retail price, even if it does have retro vide to it. And then buy peripherals of original retail price also.

[thorfdbg]

Yeah, but reimplement is harder. Still tho, even if we would use original roms, I doubt anyone will care, just like anyone don't care about NES/Fami clones (or even Genesis ones, like Retron3/upcoming Retron5).

The point is: It is not harder. It is harder to get rights on the original ROMs for reasons already indicated. As already said, as far as the Os is concerned, it's done - nothing hard at all anymore.

 

The main problem would be manufacturing actual peripherals (not so machine itself) - prices of brand new atari floppies are insane ($20 for one floppy, but then again, it's ebay price). We would need to lower cost, so anyone can afford it. Because I doubt anyone would want to buy Atari clone for original retail price, even if it does have retro vide to it. And then buy peripherals of original retail price also.

 

Original hardware should not be required anymore - all that's needed is an implementation of the SIO protocol on more modern hardware. For example, a floppy emulation for CF cards.

[The Usotsuki]

...I don't think NOACs and GOACs require any ROM code from Nintendo resp. Sega...they didn't have BIOS firmwares.

[Darkhog]

But GBA has, and from what I know, Retcon 5 is supposed to play GBA games as well. SNES also has some sort of BIOS and Retcon 3 plays SNES games AFAIK.

 

Also, no emulation is perfect and because of that we actually NEED original hardware. Let me put it this way: There ALWAYS are programs that are just not working on emulated hardware. Even with accurate emulators, like bsnes or Altirra, there always will be just "that one program" that doesn't work. And which, surprise, works with actual Atari.

 

So we need to manufacture original hardware, including 6502 (as I said, it is in use for microcontrollers, robot motors, etc. so it shouldn't be hard to obtain), other chips (such as pokey) and peripherals (cassette/floppy drives, and of course floppies).

 

Not to mention that production of Atari media and drives will help us to play our old games that we have stashed in basement .

[frogstar_robot]

But GBA has, and from what I know, Retcon 5 is supposed to play GBA games as well. SNES also has some sort of BIOS and Retcon 3 plays SNES games AFAIK.

 

Also, no emulation is perfect and because of that we actually NEED original hardware. Let me put it this way: There ALWAYS are programs that are just not working on emulated hardware. Even with accurate emulators, like bsnes or Altirra, there always will be just "that one program" that doesn't work. And which, surprise, works with actual Atari.

 

So we need to manufacture original hardware, including 6502 (as I said, it is in use for microcontrollers, robot motors, etc. so it shouldn't be hard to obtain), other chips (such as pokey) and peripherals (cassette/floppy drives, and of course floppies).

 

Not to mention that production of Atari media and drives will help us to play our old games that we have stashed in basement .

 

Any re-implementation of hardware faces the same problems emulation does. There will be cases where the behaviour of the re-implement differs from original hardware. The original hardware was TTL logic built with 70s methodologies. Any re-implement will most likely be based on fixed CMOS or an FPGA. We have a lot of the original chipmasks and schematics so have the luxury of implementing the same circuits. But the timing of those circuits will often be different if built with newer tech.

 

Both the Flashback II and the C-64 joysticks were modern reimplements of the 2600 and C-64. Both were highly compatible but not perfectly compatible. In the case of the Flashback II, there were at least three revs of the hardware and while compatibility increased with each rev it never got perfect. Software emulators are least trivial to update. FPGA implementations CAN be made trivial to update but the tech is more expensive for large production runs. A fixed function chip of course will have it's bugs forever. It also occurs to me that modern 65xx cores aren't necessarily going to be bug-compatible with legacy cores. Since the old 8-bits are "dead" platforms, modern coders can exploit things like illegal 650x instructions and hardware bugs in the support chips to the fullest. New 6502 cores won't necessarily behave the same way with illegal opcodes and bug compatibility on say a new GTIA will also be very difficult to achieve.

Edited April 15, 2013 by frogstar_robot

[thorfdbg]

<p>

Any re-implementation of hardware faces the same problems emulation does.

But does emulation face problems these days? I mean, how many titles *do not* work on emulated systems? I'm curious to know...The only problem I can think of is that emulation does not always gets the timing 100% correctly because it needs to synchronize the audio clock with the emulation clock, both of which are typically different. But in hardware, this will not be a problem since the two clocks can be locked together.

But the timing of those circuits will often be different if built with newer tech.

Why? The original hardware is always synchronous to the system clock, and every specific function takes a given number of cycles. This number is known, for almost all functions I'm aware of. So where is the problem?

Modern coders can exploit things like illegal 650x instructions and hardware bugs in the support chips to the fullest. New 6502 cores won't necessarily behave the same way with illegal opcodes and bug compatibility on say a new GTIA will also be very difficult to achieve.

None of that is difficult to achieve. The invalid opcodes are known. There are a couple of unstable opcodes you cannot depend on and that function different from revision, timing and temperature of the chips, so you cannot really use them. So do not emulate them. GTIA is probably the best invested part of the system, I don't see a problem emulating it. Code for its emulation is out there, and works pretty well.

[frogstar_robot]

But does emulation face problems these days? I mean, how many titles *do not* work on emulated systems? I'm curious to know...The only problem I can think of is that emulation does not always gets the timing 100% correctly because it needs to synchronize the audio clock with the emulation clock, both of which are typically different. But in hardware, this will not be a problem since the two clocks can be locked together.Why? The original hardware is always synchronous to the system clock, and every specific function takes a given number of cycles. This number is known, for almost all functions I'm aware of. So where is the problem?None of that is difficult to achieve. The invalid opcodes are known. There are a couple of unstable opcodes you cannot depend on and that function different from revision, timing and temperature of the chips, so you cannot really use them. So do not emulate them. GTIA is probably the best invested part of the system, I don't see a problem emulating it. Code for its emulation is out there, and works pretty well.

 

I was speaking mainly to the point of reimplementing the A8 on modern hardware rather than software emulation. Many people seem to think that hardware reimplementation of classic chipsets is somehow more real than emulation. In most respects it isn't save that a hardware re-implemention doesn't have render into discrete frames and thus can potentially avoid tiny latencies with control inputs and sound that a sw emulator most likely can't. By that I mean a hardware reimplemention can "race the beam" and have things happen with finer timing than a PC monitor's refresh rate. But apart from that.......

 

One of the ideas mentioned in a hardware redo of the A8 is dropping in a modern 650x core. It is in that context that I see illegal opcodes as being a bugaboo since these cores need only adhere to spec as far as their external interfaces go. They don't have to be same inside at all (though they likely are %95 percent identical to the 70s and 80x chips). If one were adapted and used to replace the 6502 in say an 800, I'd expect the same potential issues.

 

Though "timing" may indeed be a red herring, I highly suspect that if ANTIC-GTIA-POKEY-PIA were transposed gate for gate to an FPGA , it would need much tweaking to even start up much less run stably and compatibly. The thing can be counted on to put up multiple fights.

 

In any case, good sw emulators (like Atari++ say ) are here now and there are small inexpensive devices that can run them acceptably. Any flaws that turn up in emulation are going to be simple fixes for end users as only replacing old files with new files is necessary. A hardware re-implentation will need additional logic and UI for flashing at a minimum for updates. And that is assuming such abilities are in the device in the first place. Flashback 2s and Commodore joysticks aren't patchable. Stella on a Pi beats an FB 2 handily in that regard.

[Markdvdman]

I was speaking mainly to the point of reimplementing the A8 on modern hardware rather than software emulation. Many people seem to think that hardware reimplementation of classic chipsets is somehow more real than emulation. In most respects it isn't save that a hardware re-implemention doesn't have render into discrete frames and thus can potentially avoid tiny latencies with control inputs and sound that a sw emulator most likely can't. By that I mean a hardware reimplemention can "race the beam" and have things happen with finer timing than a PC monitor's refresh rate. But apart from that.......

 

One of the ideas mentioned in a hardware redo of the A8 is dropping in a modern 650x core. It is in that context that I see illegal opcodes as being a bugaboo since these cores need only adhere to spec as far as their external interfaces go. They don't have to be same inside at all (though they likely are %95 percent identical to the 70s and 80x chips). If one were adapted and used to replace the 6502 in say an 800, I'd expect the same potential issues.

 

Though "timing" may indeed be a red herring, I highly suspect that if ANTIC-GTIA-POKEY-PIA were transposed gate for gate to an FPGA , it would need much tweaking to even start up much less run stably and compatibly. The thing can be counted on to put up multiple fights.

 

In any case, good sw emulators (like Atari++ say ) are here now and there are small inexpensive devices that can run them acceptably. Any flaws that turn up in emulation are going to be simple fixes for end users as only replacing old files with new files is necessary. A hardware re-implentation will need additional logic and UI for flashing at a minimum for updates. And that is assuming such abilities are in the device in the first place. Flashback 2s and Commodore joysticks aren't patchable. Stella on a Pi beats an FB 2 handily in that regard.

 

A very good point. A software emulator will ALWAYS be more adept - as the software can update with the more powerful hardware over time. The Raspberry Pi even has an acceptable DOS emulator for old games, which will get faster with more understanding. I still think the Raspberry Pi as a base with all add ons would be the cheapest, most effective way. It needs to be to get people to want to buy it!