Port T80 changes from ZX Spectrum/Amstrad, remove async clocks driving CPU

[MP2E]

Thank you for your comments, I reworked all my changes from last time, and spent some time porting the T80 changes over from the ZX Spectrum core. I made sure to preserve R800 support, which is still working from my testing.

I think I correctly made the CPU clocks synchronous this time, but it's possible I misunderstood something again

[MP2E]

Oh I forgot to mention: there are async clocks left for me to remove, and I still am doing that. Just wanted to make sure I'm not doing something wrong first

[sorgelig]

Although i've merged the PR, all CEs are wrong! Every CE must be active only ONE master clock cycle. If there is 5MHz CE, then only every 4th cycle must be active.

My T80pa wrapper has protection against CE being active more than 1 cycle - that's why it may be still work (i didn't test), but overall logic is now wrong because i see trueClk is used in other places.

[MP2E]

Ahh, in hindsight, that does seem obvious. It seems to work, but I will fix the CEs so that they only stay enabled one cycle

[sorgelig]

and those CPU ce_p/ce_n must be shifted against each other. for example CE_p is active at cycles 0,4,8,12... and CE_n is active 2,6,10,14...

[sorgelig]

Btw, now MSX got compiled in 8mins. Before it was 33mins! So you can see how much easier now the project for Quartus to compile.

[MP2E]

Yes it's much better to develop on with the faster compile times :D

[sorgelig]

Converting async clocks to CE won't exactly will work the same. So after you fix CE signals you need to test the core against the possible new bugs.

[MP2E]

Will do! I'll make sure to test many programs after converting each clock

[gyurco]

It would be cool if you can cooperate with Luca (KdL) in this change. He knows the best what needs to be tested.

[MP2E]

I agree, thank you for the suggestion!