Ian Collier wrote: > Er, the multiple interrupt effect is the *whole point*! The purpose of > my program is to see how many times the multiple interrupt occurs, and > thus determine how long the Sam holds the interrupt line for.
Ah, I never realised that it could/would be called multiple times, but probably because almost all of the other interrupt handlers I write are long enough to mask the effect! It took me ages to track down my scroller problem, but I thought the multiple calls was the problem that had been uncovered, not realising that reducing the active interrupt time was really what needed doing. So reducing it to 20us (120 cycles) will bring it into line with what a real SAM will do, and the adjustments to take the time to get to the interrupt handler also needs to be taken into account. I guess there might still be some quirks with programs that are very sensitive to it until the instruction timings includes RAM contention (for the patch the start of the main area of the display is being draw on the next line). <snip> > so you can actually have an interrupt with no clear bit in the status > register (but not in SimCoupe). So on a real SAM the interrupt handler could be called without a bit reset in the status flags? Or did I read that wrong? Si
