 No.05355
Tafoid Administrator
Dec 31, 2009, 22:45
|
If I had to guess, the 'internal divide by 15' function is at fault. 0.134u4 first showed the 400khz speed and I assume, the first version that was so slow. Previous version held a 6mhz speed for the i8035. |
|---|---|
 No.05437
etabeta Developer
Jan 6, 2010, 18:55
edited on: Jan 6, 2010, 19:36
|
these slow downs are related to some s2650 change happened between 0.134 and 0.134u1 (I just tested this by applying the ACCESS VIOLATION fix to u1). we probably need couriersud to have a proper fix as an additional note: regression is due to irq handling. I'm going to check if the changes were correct with some s2650 expert... |
|
No.05451
Haze Senior Tester
Jan 7, 2010, 18:37
|
well, if you use pulse line, there is a chance the interrupt can be missed. If you use hold line, it will be held until the interrupt is taken. static INTERRUPT_GEN( quasar_interrupt ) { cputag_set_input_line_and_vector(device->machine, "maincpu", 0x0, HOLD_LINE, 0x03); } It could just be that the interrupt handling in the driver is wrong, for IRQs PULSE_LINE is pretty uncommon anyway (it's usually used with NMIs which the hardware has to take regardless) |
|
No.05454
etabeta Developer
Jan 7, 2010, 23:12
|
Fixed by Quench (which told me how to fix this half an hour before Haze's comment ;) ) |