In theory could an Odyssey 2 emulator run on a CoCo?

[courtesi96]

Is this the real life? Is this fantasy? Could a emulator like O2EM be ported to the CoCo?

 

It would have to be in Assembly no doubt, but the Odyssey 2 is probably one of the simplest console architectures available.

 

Thoughts?

[linville]

The 8048 is clocked at the "Hi speed" rate on the CoCo, custom chips for video and audio, and more on-screen colors. The CoCo's 6809 can make-up for a lot, but even the lowly Odyssey2 would be a tall order for emulation.

[JamesD]

If it could be emulated with some degree of accuracy, you'd need a faster clock speed.
The Spectrum has successfully been emulated on the Atari, but you need an accelerator board for it to run full speed, and the Spectrum hardware might be easier to emulate.

[courtesi96]

Would a 6309 be enough?

 

As far as I know the O^2 only has 16 colors so a CoCo 3 should have that down.

[JamesD]

Would a 6309 be enough?

 

As far as I know the O^2 only has 16 colors so a CoCo 3 should have that down.

Look at the CP/M emulation running on the CoCo 3.

There is no fancy hardware to emulate and it still runs slow.

The 6803 would not be enough.

[courtesi96]

I looked at the CP/M emulator videos and my thoughts are:

 

Some percentage of slow down could be due to it running under NitrOS9. I don't know the percentage but obviously this would probably run faster under RS-DOS.

 

Because it's under NitrOS9 and the fact that writing emulators in 100% Assembly Language would be a pain I'd be curious to find out if the emulator itself is written in Assembler or more likely C - which wouldn't be as fast.

 

I've never written an emulator but the way I was imagining it is behind the scenes there would be much more complexity with CP/M due to how complex it is vs. the O^2. Yes the O^2 has unique chips but these should be (in theory) a fraction of the complexity than an operating system.

[George Phillips]

Generally speaking the host machine has to be at a minimum 10 times faster than the emulated machine to run at full speed. And that's assuming that graphics and audio emulation low overhead. To emulate another processor the host processor has to:

 

1. Read the next opcode of the emulated processor.
2. Jump to the associated routine to handle that operation.
3. Run a routine which does whatever that instruction does
4. Repeat.

It's pretty hard to do that dispatch in less than 4 or 5 instructions. And you're generally lucky to keep the opcode emulation down to an average of 5 instructions (flag registers are a big problem here). This doesn't even include detecting when the emulated processor is accessing an I/O port or memory-mapped device register. The "10 times" estimate is just an approximation and I think of it as the best one could hope for. If the host isn't 10 times faster then there's no chance of interpretive emulation. And beware the excluded middle. Just because a host is 10 times faster doesn't mean it is possible, either.

 

A dynamic emulator which translates blocks of emulated instructions into native instructions can do better. That can at least save the cost of dispatching the opcodes. And with a ROM based system that sort of translation can be done outside the emulator.

 

Even then I think the performance gap between the CoCo and Odyssey 2 is too narrow. No emulation/translation technology is going to run at full speed. At best I could see doing automated translation of most of a game and having a programmer go in and optimized the hot spots to get it running at full speed. Might be just as easy to port each game individually.

[BassGuitari]

Not sure, but I bet most APF M/P-1000 games could be ported to the CoCo 1/2 and would be virtually indistinguishable.

[courtesi96]

Thank you for all your comments. I thought it might have been more realistic than say compared to an Atari 2600.

[ColorComputerStore]

Not sure, but I bet most APF M/P-1000 games could be ported to the CoCo 1/2 and would be virtually indistinguishable.

 

Oh that would be interesting! Maybe NEC PC6600 games too as they used a 6847 as well, though different CPU.

[carlsson]

If Odyssey^2 is too much, consider Channel F.

 

The lowest rated console of them all probably is the RCA Studio II which in its original form was B&W and ran all its games through an interpreted pseudo language. Recently a Studio II simulator was released for the .. *drum roll* Intellivision, which takes advantage of the interpreted mode and implements the instructions natively instead of emulating the 1802 CPU. It brings it to decent speed. That simulator also has a mode to emulate 1802 for the few recent homebrews that don't use the interpreted pseudo language, but then it crawls. I'm sure a 6809/6309 machine could do far better than the CP-1610 does in this case, but don't expect too much speed even for that purpose.

[BassGuitari]

 

Oh that would be interesting!

The APF actually runs the Motorola 6800, the "predecessor" of the CoCo's 6809. I guess they're not binary compatible, but assembly is okay (according to Wikipedia; I have no idea--that stuff is beyond me!).

 

Many APF games look like low-res CoCo games. Even if they're not "emulated," per se, I would imagine it wouldn't be a stretch to adapt some of the games for the CoCo. They even have the same text "font."

[ColorComputerStore]

Ah, 6800...that's right! I recommend everyone read the story of the young black engineer who built it. Amazing story.

[JamesD]

Look at the CP/M emulation running on the CoCo 3.

There is no fancy hardware to emulate and it still runs slow.

The 6803 would not be enough.

That was supposed to say 6309, not 6803.

I've spent too much time on the MC-10 I guess

[JamesD]

Forum is acting odd. Posted many times for some reason. I'm unable to delete.

I thought you were a spam bot gone nuts

 

 

 

 

Ah, 6800...that's right! I recommend everyone read the story of the young black engineer who built it. Amazing story.

Carlos

Just an FYI, he's referring to the creator of the APF Imagination Machine, not the creator of the 6800.