Hi, just for curiosity. I'm wondering how much, in terms of speed, can be pushed FPGA z80 cores nowadays without doing sw emulation. and, of course without respect the original sequence of states of the original state machine.
Maybe one could get a FPGA z80 core @100Mhz? or more?
And doing some optimization how much can required can T-states can be reduced?
This came up recently in the Spectrum forums, and it seems that the Spectrum Next can already have an 28MHz program as long as you don't use any external connections like ROM cartridges, or printer, or keyboard, or any other external stuff.
This doesn't mean it's just as easy to do this for the MSX, but it may be possible, in theory.
For modifying vhdl of one chip msx or terasic de0/de1/de2 or nano boards, the crystal and the sdram (and fpga) is fixed, so their speed limits the speed up. Just guessing but I think on these boards I think 50 to 100mhz z80 is the limit?
If we use a different fpga chip and better memory, I guess it might be possible to design a faster z80, but may be requires more modern cpu design, it would require cache etc and even architectural change that it won't be a z80 anymore.
In software emulation, what is the speed equivalent in mhz are we getting from a z80 emulator (running on a 2ghz pc for example)?
Maybe I'm wrong but I think there's no speed limit with a Z80 FPGA (depend of the FPGA).
In an MSX system, clock should be slow down every I/O or MEM access.
I project to make such tests but I don't know when I will have the time.
i need to clarify a bit: the question was not how much the msx system could be speeded up. the question is (almost in the beginning) how much the z80 fpga could be made faster in terms of both:
- pure Mhz
- architectural changes (like the r800 does for bus timings for example) including some more advanced pipeline or phases overlapping....
