From: Mike M. <mi...@mu...> - 2004-10-22 23:28:28
|
Stephen torri wrote: > I believe the accuracy can only be improved by finding any bugs in what > I modified. Run it on your machine and tell me the results: Hmm, "Illegal instruction". Ah, it's running on a 486-class machine. Let me try it on a higher-spec machine (Pentium+): CPU Frequency: 1809482476.0000 hz (1809.4825 Mhz) # of ticks in a 1 period: 1809482380 ticks CPU Frequency: 1809481374.4000 hz (1809.4814 Mhz) # of ticks in a 10 period: 18094813648 ticks On the 1.8 GHz machine. So this approach may have some merit on Pentium+ machines. And it would definitely fix conflicts that users (like me) see when running the chrony NTP daemon that persistently updates the clock. But my area is not engine matters... I will pass that to Thibaut or Miguel or Michael or someone else who knows better. -- -Mike Melanson |