Author: blass uri
Date: 11:21:34 03/03/00
Go up one level in this thread
On March 03, 2000 at 12:51:28, Ed Schröder wrote: >On March 02, 2000 at 18:28:31, Robert Hyatt wrote: > >>On March 02, 2000 at 05:12:10, Ed Schröder wrote: >> >>>On March 02, 2000 at 04:45:46, Jeremiah Penery wrote: >>> >>>>On March 02, 2000 at 03:58:13, blass uri wrote: >>>> >>>>>On March 02, 2000 at 01:42:43, Ed Schröder wrote: >>>>> >>>>>>Nowadays programs hit a deeper ply depth than DB. See the log files >>>>>>on the IBM site. >>>>>> >>>>>2)I understood also from hyatt's posts that you need to add 3-4 plies to the >>>>>depth in the logfiles of the IBM site to get the real depth. >>>> >>>>Actually, it's more than that. For this search: >>>> >>>>depth score time PV >>>>11(6) 48 T=204 Qa7... >>>> >>>>You have a depth of 11 PLUS *six* plies of hardware depth. So a total depth of >>>>_17_. That seems to be about the average depth of most of their searches. >>>>That's a bit beyond PC programs. And we know they were extending probably more >>>>than PC programs also. >>> >>>Is there an official statement of this? >>> >>>It's all not so clear. >>> >>>We also don't know which or how many extensions are done. The same >>>applies for most chess programs BTW. We can only guess. >>> >>>Ed >> >> >>This is official from the DB team, two different members confirming the >>notion that (x) is additional plies searched only by the hardware. I told >>vincent for several years that when they said "10" plies they meant something >>different than he did. This was common knowledge for deep thought. I just >>never thought to re-state it since it was mentioned so many times at the ACM >>events they participated in. >> >>11(6) suddenly becomes frightening, doesn't it? :) > >Maybe, maybe not :) > >Nowadays chess programs have a branch factor of 2-2.5. The branching factor is not important because the meaning of branching factor is different for different programs. DB has a branch >factor of 4. Every doubling in PC speed we are almost hitting one ply >deeper, coming closer and closer and eat the thing for dinner in the end :) > >Alarming doesn't it? :) > >Ed The question is what is the meaning of one ply deeper. The meaning is not the same if you do more extensions. The meaning is not the same if you do more pruning. We cannot compare between programs by the number of plies they see forward because the meaning is not the same. The only way to compare is by playing games. Deeper blue is dead and does not play so I do not see the point of all the discussion about it because we cannot compare between it and the top programs of the next years because we will never have data. The games against kasparov do not help much because we need comparison of results in the same time. If another program will lose in 2004 against kasparov it may be because kasparov of 2004 is a better player and not because deeper blue was better. 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.