Waiting some time!

[PokeMon]

Yes - that should be right. Anyway the 5.5 million calculation was completely wrong.
This brings me to following approach with more easy calculation:

Code: Select all

            ld      c, N
    loop   djnz loop
            dec c
            jr nz, loop

Which is only 7 bytes long, runs approx. C * 1ms and avoids stack usage (which should better avoided when calculating every byte).
Only the first iteration of B is undefined - giving a tolerance of about +/- 500us and a granularity in ms steps.
In SLOW mode you have to multiply with about factor 4, so it is C * 4ms +/- 2ms, giving an all in all running time of maximum 1 second with C=0.

[antoniovillena]

Wow. It's awesome. Good optimization. You can eliminate the unknowed B by changing the first instruction with "defb 1, N" and with A!=13. That loads B with $10 (opcode of djnz).