Sorgelig wrote:On computers where 68000 is used usually cycle accuracy is not required. It's more important how correct every instruction is executed. Minimig is pretty compatible with most Amiga 500-1200 software, so the CPU is not an issue.
Exactly! As you are saying, "usually
" cycle accuracy is not required. But in the case of the ST it is. Trust me, I know what I am talking about.
And yes, I am aware that the Minimig is very compatible with most Amiga software. Precisely because it seems that the 68000 core in the Minimig is good enough for the Amiga, but not so much for the ST. For starters, the ST uses some features of the 68000 that the Amiga does not, such as Bus Error exceptions. And that is, precisely, one of the 68000 features more difficult to emulate accurately.
It is well known that the ST requires higher 68000 accuracy than the Amiga. We know that from software emulators even before we had FPGA "clones".
toromand wrote:Google-ing around for ST FPGA, apart from MIST, brings Suska up. They claim pretty accurate ST implementation ...
It is not cycle accurate. It is not even the developer's goal to be cycle accurate. He is more interested in implementing a Falcon (68030 level ST) compatible core, which is quite a task.