Originally Posted by grond
There was no 680x0 compatible with all other 680x0. The 68000 is the least compatible of them all as it cannot execute 68020+ code.
When talking about compatibility, of course it should be specified what it is relative to. What I meant was compatibility with plain 68000 code like most games. Of course 020, 030 etc. has their own compatibility issues like self-modifying code. The Apollo core could be more compatible with plain 68000 code than 030 and 040, but hardly as compatible as a real, but overclocked 68000. Nothing can be more compatible than the reference.
If the reference is Amiga software as a whole where some parts are software written for 68000, other parts are 020+ code etc., then perhaps the Apollo core could be generally more compatible if it resolves some of the compatibility issues between 68k versions.
It might end up becoming a price issue whether there would still be a market for something like the Zeus68k. People who just want to play WHDLoad games does not need something as fancy as the Vampire so if they could save some money, the Zeus68k might be all they need. Also, I think the Zeus68k works with the Indivision? That could also be a deciding factor as long as something similar is not available for the Vampire.
Originally Posted by eXeler0
Why are you quoting me on select parts ;-)
In the next sentence I clearly stated you could do an *exact replica* of the original CPU.
Yes it is possible, but I don't think it will happen since it is very tedious work to make sure something is 100% identical instead of just 99.9%. And why would you want to if you can use genuine 68000 CPUs running at 50 MHz? People with a reconfigurable FPGA CPU-core will probably rather have a 040/060 style CPU like the Apollo if it can achieve good compatibility with 000/020 code. But of course it would still be nice to have a 100% true 68000 core as well and be able to switch between the cores, either through reflashing or perhaps some other switching mechanism if the FPGA has space for both cores simultaneously.
Anyone know if the 68000 has been 100% reverse-engineered on a transistor level (and not just empirically)? I know the 6502 has, but it is also much simpler.