Indeed. I tried several different CPU cycles using the DOSBox CPU dropdown and the generation times (actual) vs. what Fractint reported varied wildly. My setup defaulted to cycles 3000 and the reported time and the real time were pretty close. I then tried max, which on my system is reported as approx 22000. I say approx because it varied during two different tests. the first time is was 21692 and the second was 22214. and both times Fractint reported around :09 vs actual :08.5. So I then doubled the CPU to 44000 and Fractint reported :04.39 vs. actual of :08.5. I tried 500000 and Fractint reported :00.39 vs actual :07.94. I even tried 1,500,0000 cycles and got Fractint reported :00.17 vs. :09.37 actual. Summary: CPU cycles Fractint time Actual time 3,000 :01:05.74 :01:06.37 22,000 (max) :00:08.79 :00:09.19 44,000 :00:04.39 :00:08.5 500,000 :00:00.39 :00:07.94 1,500,000 :00:00.17 :00:09.37 I hope this table comes out ok. My conclusion is that for CPU cycles, the best setting is "max". I think that the various speeds that I measured with my stopwatch show no consistent variations in speed after the max setting is exceeded. On Sun, Nov 29, 2020 at 4:28 PM Timothy Wegner <tim@tswegner.net> wrote:
I changed the cycles= command in dosbox-x.conf to cycles=max. On my system that changes cycles to 54337. Using a higher number reduces the calculation time further as reported by Fractint's tab screen, but I wouldn't trust the tab screen value anyway when the CPU is pushed to the limit. If anyone discovers other DOSBox-X parameters that should be tweaked to let Fractint run as fast as possible, please let me know.
Tim _______________________________________________ Fractint mailing list Fractint@mailman.xmission.com https://mailman.xmission.com/cgi-bin/mailman/listinfo/fractint