Computers with multiple (different) processors

[+rdemming]

The Apple IIGS was a 65816 and 6502, plus any other addon II cards (68K, Z80, 8088/x86)

 

Are you sure the Apple IIGS had both a 65C816 and 6502? I'm pretty sure the IIGS only has a 65C816 that replaces the 6502 that is used in the original Apple II. The 65C816 is backward compatible with the 6502 so no real need to have both to remain compatible with the Apple II.

 

Robert

[Osgeld]

You are correct the 816 is a 16 bit expansion of the 6502 and backwards compatible

[Nebulon]

There's the Fujitsu FM-7 with dual 6809 CPUs.

[shoestring]

 

Ah, thank you. I was trying desperately to remember this site. I spent a couple of days browsing through the whole thing. Really interesting information.

 

So many machines used a Z80 as a sound processor in concert with another main processor (68000, for instance.) I wonder what it is about the Z80 which made it so attractive versus something like a 6502 or compatible.

 

The Japanese mainly used z80 from the beginning in arcade hardware. So it was only natural for them to keep using it as a sound processor later on in conjunction with the 68000 , see Capcom CPS1 as an example. Non Japanese manufacturers tend to focus more on the 6502 because that was more popular outside of Japan. Not always true though, I'm pretty sure Data East used the 6502 for sound exclusively early on.

[towmater]

I think a better question would be what computer didn't have a second processor capability or built-in? Only TI comes to mind and I am likely to be corrected with that one. Are we defining this as any chip that uses opcodes? Even the Atari 400 had an Antic cpu.

[carlsson]

I would consider the VDP a kind of co-processor. On the other hand, systems using the CPU to generate the video like a ZX-80/ZX-81 and others might fall into your category. Still at least the 81 has a custom ULA chip instead of discrete electronics

[MarkO]

I would consider the VDP a kind of co-processor. On the other hand, systems using the CPU to generate the video like a ZX-80/ZX-81 and others might fall into your category. Still at least the 81 has a custom ULA chip instead of discrete electronics

 

I was always impressed with the ZX-81.. One Z80 CPU, One ROM, One ( 1K ) or Two ( .5K) RAM(s), and the ULA...

 

MarkO

[JamesD]

I would consider the VDP a kind of co-processor. On the other hand, systems using the CPU to generate the video like a ZX-80/ZX-81 and others might fall into your category. Still at least the 81 has a custom ULA chip instead of discrete electronics

The 9918 VDP is a bit limited and dedicated in functionality to be called a processor.

It doesn't even offer as much functionality as the Antic, which is at least programmable.

If the VDP had a blitter, drew lines, etc... you might have more of an argument, but it's still not programmable.

On a modern graphics card you can program the GPU to perform calculations besides graphics, so I think they would qualify.

 

[mr_me]

Not many were produced but the Intellivision + Keyboard Component had a CP1610 16-bit processor and a 6502 processor. There was 16K of 10-bit RAM that was shared by both processors. Software released on magnetic tape ran on the CP1610 but some, through a BASIC interpreter, ran on the 6502. I think the 6502 was also used to run the completely automated tape drive. http://spatula-city.org/~im14u2c/images/kbd/

[tschak909]

The 9918 VDP wasn't a co-processor, as it didn't run programs.

That particular chip was a real crutch for hardware developers. Even the most inept systems engineer could take a 9918, slap a 9919 for sound next to it, and a Z-80, with 16K of RAM, and 16K of VDP RAM, and make a computer (or cut the system RAM e.g. down to 2-4K for a game system) out of it, with an absolute minimum of glue logic...and so many companies did... Coleco...Sega...SpectraVideo... every damned MSX computer on Planet Earth, even the Sega Genesis could be flipped into a 9918 compatibility mode, if initialized to do so (via the Master System adapter).... Before chips like the NEC uPD7220 became available, this was the go-to chip for a quick graphic system design...

 

The TMS34010 was a serious sign post, because it gave systems designers a high speed DSP that had lots of goodies for 2D and 3D graphics accelleration. (yes..even the 34010 could do simple 3D stuff, I saw a 34010 do gouraud shading in 1985.), and paved the way for modern GPUs in systems today.

 

-Thom

[carlsson]

Well then, if we consider the ANTIC to be a co-processor of some sorts but the VDP isn't, we have some cutoff. Is the VIC-II programmable enough (you can set which line to trigger the raster interrupt, for instance) to be called a co-pro or not? Generally I think CPUs, MCUs and DSPs are within the types of chips that work with opcodes and primarily would be considered "processors" in this context.

 

Clearly though the TI-99/4(A) is not the only vintage computer without onboard or back then expansion co-pro functionality, even in the extended meaning of a co-pro. In modern days, I understand a F18A has computational power so it might qualify to a higher degree than the old VDP would. On the other hand, I'm almost certain there are CPU upgrades or expansions even for the low end Sinclairs and similar brands so eventually every computer might have the ability to plug in another CPU, even if it is highly non-standard and thus not what the said computer is known for.

[JamesD]

Well then, if we consider the ANTIC to be a co-processor of some sorts but the VDP isn't, we have some cutoff. Is the VIC-II programmable enough (you can set which line to trigger the raster interrupt, for instance) to be called a co-pro or not? Generally I think CPUs, MCUs and DSPs are within the types of chips that work with opcodes and primarily would be considered "processors" in this context.

 

Clearly though the TI-99/4(A) is not the only vintage computer without onboard or back then expansion co-pro functionality, even in the extended meaning of a co-pro. In modern days, I understand a F18A has computational power so it might qualify to a higher degree than the old VDP would. On the other hand, I'm almost certain there are CPU upgrades or expansions even for the low end Sinclairs and similar brands so eventually every computer might have the ability to plug in another CPU, even if it is highly non-standard and thus not what the said computer is known for.

When you talk about a processor, you are usually talking about running some sort of user definable program.

So the chip should have some sort of program counter, it should read a series of user written instructions, and those instructions should perform some sort of function without constant support of the CPU.

 

I don't think the VIC-II qualifies since it doesn't read instructions where the Antic clearly does, as does the Amiga Copper.

 

OTOH, should we consider offloaded functions like blitting, line drawing, etc... when they are handled by another chip?

After all, blitting is a sequence of logical memory operations performed in a loop at the command from the main CPU. The difference is that the looping, AND, OR, etc... are handled by a fixed internal hardware program.

This could be a grey area.

I don't think regular DMA memory moves that are programmed through CPU writes to registers would qualify.

But I thought about creating a blitter that could read from a list of blitter operations to perform, and have it carry them out until it hits the end of the list.

The CPU just updates the blitter list and then tells the blitter where to find it. On writing the last byte of the start address, the blitter starts processing the list.

However, it's just a list of blitter instructions loaded into the regular blitter hardware preceeded by the number of blits in the list.

Does that qualify as a program?

And line drawing is a complex program, just implemented in hardware.