On Sat, Apr 25, 2015 at 10:01:15AM +0200, Anders Carlsson wrote: >Hm, does this mean the TI$ in BASIC is based on the simple timers (less >accurate?) found in both 6522 and 6526, and the TOD clock is something >else? Yes. Actually, I made a mistake in my previous post by claiming that the PAL/NTSC detection does not affect anything. Now that I think more carefully, the CIA timer interrupts for RS-232, tape, and for the regular $ea31 interrupt must be at a fixed rate, so they must be adapted for the different 6510 clock frequency in the C64 (17734472/18 Hz on PAL and 14318181/14 Hz on NTSC). On the PET, the 6502 always runs at 1MHz. The timer interrupt should be 60Hz. It is not 50Hz in PAL models, and I guess it is not synchronized with the 6567 screen updates in NTSC models either. AFAIU, the C64 KERNAL is very close to the VIC-20 one, and very little software depends on the ToD clock. GEOS is one example. I wonder if anyone ever measured the inaccuracy of the TI$ clock, especially during disk, tape, and RS-232 operations. (And this should be without considering software that disables interrupts or changes the interrupt frequency.) Marko Message was sent through the cbm-hackers mailing listReceived on 2015-04-25 12:00:06
Archive generated by hypermail 2.2.0.