Author: Len Spencer
Date: 14:55:49 04/28/98
Go up one level in this thread
On April 27, 1998 at 23:16:06, Christophe Theron wrote: >On April 27, 1998 at 16:57:04, Bert Seifriz wrote: > >>On April 26, 1998 at 23:35:28, Christophe Theron wrote: >> >>>On April 25, 1998 at 13:55:37, Bert Seifriz wrote: >>> >>>>On April 25, 1998 at 03:33:15, Dan Newman wrote: >>>> >>>>>On April 25, 1998 at 02:06:29, Christophe Theron wrote: >>>>> >>>>>>On April 25, 1998 at 00:06:01, Robert Hyatt wrote: >>>>>> >>>>> >>>>>>> >>>>>>There doesn't seem to be any way to turn this behavior of >>>>>CM5000 off. This makes any contest involving CM5000 and >>>>>carried out on one machine potentially unfair >>>> >>>>I would like to agree with that 100 % percent. >>>>Probably you can forget testing on ONE machine completely, >>>>because you never know how many resources program A or B >>>>get from your one machine. On this forum there are many >>>>better experts than I am, but I have never heard that anybody >>>>could be sure your 2 programs get the same chances on 1 machine. >>>>There is no way around 2 identical computers I think. Bert! >>> >>>With DOS programs, there is absolutely no problem. You can turn a DOS >>>program "off" when it is in the background. >>> >>> >>> Christophe >> >>Yes, you are right, but then you lose permanent brain function. >>I was thinking about Windows only. Bert > >You should not use permanent brain, even when using Windows programs, >when both programs run on the same computer. > > > Christophe From my experience, the only problem with permanent brain is in the beginning, where one program is still in book and the other is thinking. The thinking computer will take almost all the CPU time until the other program leaves its book, then they will both settle down to about 50-50. I have seen this with Crafty and GNU Chess playing each other through Winboard on Win 95, as well as with Crafty and ArasanX. On a 200MHz Pentium, it's like their both running on separate 100MHz machines. The only drawback is you need to keep the hash tables small enough so that no virtual-memory disk access has to occur, and TOTALLY slow down the whole shebang. But Bert is right, how can you be absolutely sure both programs are getting the same CPU time? Permanent brain or not, it's nice to be able to put 2 programs head-to-head like this, especially when you happen to be working on one of the programs involved.
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.