I have been trying for the past few hours slow down the CPU on bus access (more time in 8 MHz mode like the previous V1 booster).. While this works, it actually works out slower running at 36MHz than the V1 running 32MHz. This is because the asynchronous clock needs to be synchronised which adds delays into the logic. I have tried all sorts to avoid this but it just isn't going to work
For example the benchmark below.. is 36MHz but with 8/36 switching...

- 836.jpg (55.82 KiB) Viewed 3820 times
Versus the V1 32MHz booster...

- ste32mhz.jpg (52.21 KiB) Viewed 3820 times
It is pretty close in both cases, but even so, its still not totally stable.
<rant>
These problems really just come down to trying to speed up the on-board ROM access. This access is all done in the 8MHz domain, so syncing it to the faster clock domain, adds delays. Without the delays, it does actually run fractionally faster, but it is very unstable. Switching to 8MHz to keep it all happy is the way to go, but still ends up being slower than it really should be. So neither method is a "good" option.
Another issue is that we have the ACIA timings to complicate things even more, the CPU has to down clock to 8MHz.. But this is why I like the SEC CPU, as you have to emulate those timings yourself manually, but these are free running that no switching about frequencies is needed at all. So makes a design a whole lot better.
Of course in terms of future booster designs, the ROM decoding will be done by the PLD logic all in the faster clock domain, so these problems would not be apparent on a proper booster like the STFM V2.2. This STE booster really needs its own ROM and SEC CPU, but that's more work and expense.. I'm not really happy about how these STE boosters are coming together, but again, I just don't have time to build a "proper" full blown booster for the STE. I don't even have time to do the STFM one!
It is why I just think all this time is better spent on developing a new ST machine and new hardware to fit it. I have gone through all this development work countless times with various boosters over the years, I dare say 90% of the time is fighting problems which should not be there in the first place. As said before, it is just not a constructive use of time... While it is more work initially to create a new motherboard, upgrades for it will ultimately take considerably less time to develop.
</rant>