Author: Uri Blass
Date: 17:06:21 05/18/02
Go up one level in this thread
On May 18, 2002 at 05:01:30, Ian Aston wrote: >On May 18, 2002 at 02:23:18, Uri Blass wrote: > >>On May 17, 2002 at 15:11:58, Ian Aston wrote: >> >>>On May 17, 2002 at 13:15:19, Uri Blass wrote: >>> >>>>On May 17, 2002 at 12:24:27, Brian Katz wrote: >>>> >>>>>On May 16, 2002 at 19:03:49, Jorge Pichard wrote: >>>>> >>>>>>On May 16, 2002 at 19:00:52, Jorge Pichard wrote: >>>>>> >>>>>>>These 6 games were played with Nimzo8.ctg and 64 MB hashtable. Average time >>>>>>>control 4 minutes per move. Final score 3.5 to 2.5 in favor of Hiarcs 8. I would >>>>>>>like to continue , but these games are too l o n g. >>>>>>> >>>>>>>I would like to know if other people can get better results using other Opening >>>>>>>Books? Even at shorter time control. >>>>>> >>>>>>PS: I also have Blitz games if you need to compare them later. >>>>>> >>>>>>Pichard. >>>>> >>>>>I recently ran an engine tournament on my 350 MHz Pent II. >>>>>Time control 1/1 hash tables set at 8. I have 384 SD RAM. >>>>>I thought Hiarcs would excel at the longer time controls rather than the shorter >>>>>time controls. This was with the Hiarcs 8 using the CD book. This was before I >>>>>downloaded the new upgrade. Fritz 7 used it's own book of course. >>>>>I was quite suprised with these results. >>>>> >>>>>game 1-1 Eatontown 2002 >>>>> >>>>> 1 2 3 4 5 >>>>>1 Hiarcs 8 **** 0½1½ ½11½ 1111 110½ 11.5/16 >>>>>2 Shredder 6.02 1½0½ **** 0100 001½ 1111 8.5/16 >>>>>3 Shredder 6 ½00½ 1011 **** ½½½1 0010 7.5/16 >>>>>4 Fritz 7 0000 110½ ½½½0 **** 1101 7.0/16 >>>>>5 Crafty 18.14 001½ 0000 1101 0010 **** 5.5/16 >>>>> >>>>> >>>>>In longer time controls Fritz 7 seems to come out quite a bit ahead of Hiarcs 8. >>>>>Too soon to tell. Any similar results out there? >>>> >>>>I have not Hiarcs8 but I am not surprised because Hiarcs was also in the past >>>>relatively better in blitz. >>>> >>>>I see no reason to expect Hiarcs to do better at longer time control. >>>>It has not special search rules for long time control or special evaluation to >>>>know things that cannot be discovered by search(for example fortress positions). >>>> >>>>Uri >>> >>>On the other hand, the programmer says that the Hiarcs 8 search is capable of >>>reaching 1-3 iterations deeper than Hiarcs 7.32. Surely this must count for >>>something at longer time controls. >> >>I do not see why only at long time control. >>It should help at all time controls. > > >Perhaps I should have said, the article also states, that the Hiarcs 7.32 search >was prone to exploding at plies of 10 or over. If the change in the shape of the >search tree has fixed this problem then it should help with long time controls. >Anyway, I suppose we will find out when it is included in the SSDF ratings and >we can see the difference between HIarcs 7.32 and Hiarcs 8. The comparison of Hiarcs8 is with other top programs. I know that Hiarcs7.32 had bugs for long time control and it could explain the fact that hiarcs7.32 earned less from time but I see no reason to believe that without bugs Hiarcs earns more from time than programs like Fritz7. 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.