I'm sorry, it's late for me 
At the time you wrote I was already in bed 
I only had to change "alloc" instead of "malloc" and it works perfectly (I am not quite useful at coding
)
with the same code (and the same openmsx) it gives 108 ticks (at 50 Mhz, MSX2 NMS 8245) ... slow?
at 50 Mhz sounds a bit slow 
...perhaps tomorrow I will be awoke
don't worry, MSX can wait 
Hi again
I said I would try on a real MSX, and ... of course I do it
I've used the same code on ML-G3, at 50 Hz, and the result was 111/112 ticks (a bit more than two seconds for 4095 moves) so, it is not good for BDS-C
anyway, if you want to try another code, I will be glad to do it
see you
sounds to me like IX points to stack base of current function and is a pointer to a pointers list containing stack bases of underlying functions.Correct. But that has nothing to do with debug support. It's a normal way of setting up re-entrant code.
What you could try is to make all variables static, with the exception of the ones in storre(). Static variables are not created on the stack (and thus accessed via IX/IY), but in the bss segment.
[edit]Ah, wait there are no local variables in the code, except for a few in the main() routine. Then my last suggestion was not very helpful Then the best thing to do is to get rid of the recursion![/edit]
