Author: Ian Osgood
Date: 10:47:10 12/14/99
Go up one level in this thread
On December 14, 1999 at 04:06:36, Daniel Clausen wrote: >Hi > >On December 13, 1999 at 23:10:31, Tom Kerrigan wrote: > >>On December 13, 1999 at 18:39:25, Dann Corbit wrote: >>>If you are running other threads of execution on your system, it won't be very >>>accurate, since clock() returns time slices given to the process only, not wall >>>time. However, if you start the process in high priority, it will be pretty >>>close. Unfortunately, high resolution timers are not very portable, and you >>>usually end up writing them one by one for each new system. >> >>If we were talking about UNIX, I would agree with you. But instead we're talking >>about the Mac OS, which is so bloody retarded that I would be surprised if it >>could tell how much time a "process" got. It doesn't even have preemptive >>multitasking, and it most certainly can't assign priorities to processes. >> >>(Of course, I'm talking about OS9 and not OSX. OSX rocks.) > >I ran the program with all extensions turned off and nothing else >running. (except for Finder of course) So I think I'm on the safe side here. >I'd love to change this clock() thingy with something more accurate, but I >didn't find anything better so far. Of course there has to be something >more accurate, otherwise profiling wouldn't be fun. =) > >And yes, MacOSX will rock. :) > >Kind regards, > -sargon FWIW, I compiled TSCP for the Mac implementing get_ms() using TickCount() *1000/60 and got nps results for the same processor speed similar to yours (+/- 5%). Ian
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.