Author: Robert Hyatt
Date: 11:54:16 01/01/04
Go up one level in this thread
On January 01, 2004 at 10:07:49, Peter Berger wrote: >On January 01, 2004 at 09:54:45, Martin Andersen wrote: > >>On January 01, 2004 at 09:46:50, Peter Berger wrote: >> >> >>> >>>Assume one computer with 500Mhz, time control is 1 min/move. >>> >>>a.) Ponder=Off >>> >>>Both engines get the full 500MHz when it is their move, and don't do anything >>>when its the other's move -> fair conditions >>> >>>b.) Ponder=On >>> >>>Both engines get half the power of the 500MHz (let's call this 250Mhz) most of >>>the time -> this would be fair conditions too. >>> >>>But it can happen that one engine doesn't have a move to ponder. So the other >>>engine can think, while the first one sits idle. If the thinking one only got >>>250MHz this would still be fair, but the engine will suddenly get twice the >>>power for this move, namely the full 500MHz. -> unfair conditions >>> >>>Peter >> >>Ah, so if you limit the cpu usage to a max of 50% for each program, that >>would be fair conditions ? >> >>Martin. > >All else being equal, and in case you can really limit the CPU time and are not >thinking about setting process priority only, I think the answer is yes: then >ponder=on matches on a single processor computer would be fair too. > >I can still think of some possible technical issues, but then I can do the same >when it is about ponder=off matches. > >Peter I know of _NO_ operating system that will let you do that. IE run a process for 50% of the time, and sit idle for 50% of the time. It is simply against any logical approach to process scheduling. It is easy to control this if there are multiple ready processes on the machine, but not if there is just _one_. That one will run all the time.
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.