Author: Jonas Bylund
Date: 02:26:19 10/16/03
Go up one level in this thread
On October 14, 2003 at 07:40:54, James T. Walker wrote: >On October 14, 2003 at 02:38:45, Jonas Bylund wrote: > >>On October 13, 2003 at 19:15:52, James T. Walker wrote: >> >>>On October 13, 2003 at 14:57:35, Jonas Bylund wrote: >>> >>>>On October 13, 2003 at 14:42:45, Djordje Vidanovic wrote: >>>> >>>>>On October 13, 2003 at 14:36:09, Jonas Bylund wrote: >>>>> >>>>>>On October 13, 2003 at 14:19:14, Christophe Theron wrote: >>>>>> >>>>>>>On October 13, 2003 at 13:09:03, Charles Roberson wrote: >>>>>>> >>>>>>>> >>>>>>>> You make the statement that Diep is a positional engine and you chose it based >>>>>>>>on that. So, why did you run G/5 matches? At G/5 tactics and search depth >>>>>>>>is crucial. >>>>>>> >>>>>>> >>>>>>> >>>>>>>I would like to bring to your attention that tactics and search depth are >>>>>>>crucial at any time controls in chess. >>>>>>> >>>>>>>Showing dimishing returns from increased search depth is so difficult that in >>>>>>>practice there is little difference between blitz and long time controls. >>>>>>> >>>>>>>If engine A gets a beating at blitz, expect it to get the same beating if you >>>>>>>repeat the match with long time controls. >>>>>>> >>>>>>> >>>>>>> >>>>>>> Christophe >>>>>> >>>>>>Now that is not my experience at all, some engines do seem to be much better at >>>>>>long time controls than at blitz and also the opposite is the case, however it >>>>>>seems that engines that do better at blitz TC's don't have the same margin of >>>>>>difference. >>>>>> >>>>>>Jonas >>>>> >>>>> >>>>>Jonas, >>>>> >>>>>this _is_ an interesting issue, I admit. However, a very quick glance at the top >>>>>section of the SSDF list will tell you that the best blitzers are up there, and >>>>>the games played by the SSDF are tournament control games. I can draw a simple >>>>>conclusion here. Naturally, there might be some conspicious exceptions... Could >>>>>you please name a program that does extremely poorly at blitz and extremely well >>>>>at longer time controls? >>>>> >>>>>Rgds. >>>>> >>>>>Djordje >>>> >>>> >>>>Maybe not extremely poorly, but Gandalf is an example of a top engine which do >>>>not do well in blitz games compared to how well it does in classic and longer >>>>time controls. (Gandalf is still one of my preferred engines for LONG analysis) >>>> >>>>If you look at the playchess.com rating lists you will find that the top rated >>>>engines are Deep Fritz 7 and Shredder 7.04. Now Shredder is another example of >>>>an engine that, atleast used to do much better at long time controls, it seems >>>>that margin has shrunk considerably, but still remains an issue. If it wheren't >>>>for dual amd's and xeon's Shredder probably wouldn't be a first choice for many >>>>of the users on playchess.com. >>>> >>>>My point about shredder is that i don't think it would top any blitz lists on >>>>equal hardware. >>>> >>>>The point that CT makes is in theory the "right" approach i think, but in >>>>reality things looks different. >>>> >>>>Regards >>>>Jonas >>> >>>Here is my "Top 10" from my blitz database of over 19000 games. See any >>>surprises? All games on equal hardware using auto232 and almost all are G/5min. >>> >>>1 Fritz 8 2489 2127 >>>2 Shredder 7.04 2489 1171 >>>3 Fritz 7 2474 3844 >>>4 Shredder 7 2470 1140 >>>5 Shredder 7.04C 2468 328 >>>6 Chess Tiger 15 2462 2130 >>>7 Chess Tiger 14.0 2456 2895 >>>8 Hiarcs 9 2450 722 >>>9 Hiarcs8 Bareev 2440 116 >>>10 Gambit Tiger 2.0 2425 838 >> >>What hardware, books, ram etc.? >> >>A surprise to me is you have different ver. of the same engine, also where is >>the same engine list using long time controls for comparison??? >> >>Jonas > >Why shouldn't I test all versions since they are different? Try the SSDF list >for long time controls on equal hardware. I see no sense in duplicating their >work (SSDF). If you want the entire list it includes Rebel 12, Ruffian, Fritz >5, Yace, Gandalf etc. etc. All engines I own and am personally interested in. >I see very little difference between my Blitz list and SSDF. (There may be some >exceptions) I have said several times here that I believe the SSDF is wasting a >lot of time testing at 40/2hours since NOBODY plays that time control anymore. >I believe there would be very little difference in the list if the games were >played at G/60min. >Jim Well that's kinda' point of it all, not to second guess any results based on bias, thus i asked if you had a longer TC list you could post, so we could actually see if there were any detectable differences between the engines in your list... The reason i was surprised to see same engines on the same list, is that i think it is pointless, but hey that's only my opinion. "lot of time testing at 40/2hours since NOBODY plays that time control anymore." Someone forgot to tell the SSDF and Kurt Utzinger that :) Jonas
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.