Author: Uri Blass
Date: 05:21:28 01/17/02
Go up one level in this thread
On January 17, 2002 at 08:11:19, Harald Faber wrote: >On January 16, 2002 at 05:32:48, Uri Blass wrote: > >>On January 16, 2002 at 00:55:22, Harald Faber wrote: >> >>>On January 15, 2002 at 17:33:07, James T. Walker wrote: >>> >>>>On January 15, 2002 at 13:36:33, Harald Faber wrote: >>>> >>>>>On January 15, 2002 at 10:44:45, James T. Walker wrote: >>>>> >>>>>>On January 15, 2002 at 01:26:54, Harald Faber wrote: >>>>>> >>>>>>>On January 14, 2002 at 20:02:32, James T. Walker wrote: >>>>>>> >>>>>>>>On January 14, 2002 at 13:04:18, Harald Faber wrote: >>>>>>>> >>>>>>>>Hello Harald, >>>>>>>>What time control are you playing? >>>>>>>>I don't know what Gandalf 5 will do at 40/2hours but I'm >>>>>>>>waiting for SSDF to test it. >>>>>>>>Jim >>>>>>> >>>>>>>You could have found out yourself. ;-) >>>>>>>You can make a tattoo on your skin: >>>>>>> >>>>>>>"Harald Faber is only testing 40/120+60" >>>>>>>:) >>>>>>>Add.: On two PCs of course with ponder=on! >>>>>> >>>>>>I'm sorry to hear that. I was hopeing for better results at the longer time >>>>>>controls. I almost never test at longer than G/2hours anymore. That's done >>>>>>well enough by SSDF. By the way, what does the above 40/120/60" mean? I read >>>>>>it as 40 moves in 120 minutes /60 seconds?? >>>>> >>>>>:-))) >>>>>No, these " are only the closing quotation marks that began with "Harald... >>>>> >>>>>40/120+60 means 40 moves in 120 minutes plus 60 minutes for the rest of the >>>>>game. So the games cannot last longer than 6 hours which is absolutely enough. >>>> >>>>I agree with that !! I now use G/2hours as the max time control and I believe >>>>it's enough. :-) >>>>Jim >>> >>>If I were sure that the results stay the same, I'd do that too. :-) >>>But a problem is that some programs suffer from blitz time control. E.g. >>>Shredder plays worse with g/180 than with 40/120+60 although the total playing >>>time is the same. The time management within the first 40 moves seems to be >>>significant different. >> >>programs should play better at g/180 because at g/180 they can choose if to use >>more than 2 hours for the first 40 moves and for 40/120+60 they have not the >>choice. > >But as you don't know how many moves the game will last, you will probably move >faster. See: >40/120+60, first 18 moves out of book. Leaves 22 moves in 120. There are SOME! >programs which use >10min per move up to move #35 and play the remaining time >faster. This is also awaiting some p.b. hits. >g/180: I have seen Shredder move much faster (certainly because there is no >implementation to use more time for the moves up to #40), and I have operated >Century 4 in the Thueringen tournament last year where Century rarely used more >than 6min for a move while in 40/120, as I already said, take about 10 or more >minutes. You can see it in the annotations I provide in my game databases. > >>Does shredder use less time for the first 40 moves >>when you tell it g/180? > >Yes, definitely. > >>Is there a program that always use the rule of using more time at slower time >>control? > >What do you mean exactly? Comparing g/120 with g/180 or g/180 with 40/120+60? > >>I suggest to test programs in the following situations with no opening book > >Personally I find tests without opening book senseless. > >>1)200 moves/200 minutes+1 minutes per game >>2)g/180 >>3)10 moves/120 minutes+1 minutes/game >> >>T(1) is the time that is used for the first move in case 1 >>T(2) is the time that is used for the first move in case 2 >>T(3) is the time that is used for the first move in case 3 >> >>How many programs have T(1)>T(2)>T(3)? >> >>Uri > >For all programs I have seen you will get it vice versa, T(1)<T(2)<T(3). My point was testing the time management of programs and not testing the programs and this is the reason that I suggested no opening book. The fact that you get the wrong result for all programs that you have seen suggests that no programmer worked on logical time management. Uri
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.