Author: Vincent Diepeveen
Date: 11:07:12 11/10/01
Go up one level in this thread
On November 10, 2001 at 10:51:57, K. Burcham wrote: >as you can see in this position, with 1500 mhz using shredder5, we can get >values for kns, and total nodes for a posted time. > >so these values for 58 minutes are: 307 kns, and 1,087,429,783 nodes. >what if we used a multiplier of 13.5. 13.5 x 1500 mhz = 20000 mhz. >using this value of 13.5, we can divide 58 minutes by 13.5 = 4.29 minutes. >lets say that Qe5 was a winning move. >so from this we could say that with shredder5 on 20 gig, could find Qe5 >in 4.29 minutes. usually a long standard time control for 4 minutes. Actually you forgot to take into account improvements in shredder positionally, and a bigger hashtable. >what about blitz. 20 x 1500 mhz = 30000 mhz. 58 divided 20 = 2.9 minutes. >what about bullet. 40 x 1500 mhz = 60000 mhz. 58 divided 40 = 1.45 minutes. >way too much time for blitz and bullet. >i know there are many variables here. and the future will get crazy with >the machines that will play chess. i read the other day that we might have >24 volts running to our fish bowl. that the water molecules with the current >passing through will actually do the switching. the chiphead that gave this >theory, said with this molecule switching, everything as we know it, as an >enduser would be instant. > >anyway i am changing my way of thinking about these mhz increases, and how our >depth changes with each. it seems that it will take an enourmous amount of >todays mhz to get shredder5 to play Qe5 in a 3/0 or a 3/3 game. >i conclude that with todays cpu, as we know them, even with smaller dies, we >cannot achieve the necessary mhz for shredder5 to find Qe5. > > > [D] r2r2k1/1R2qp2/p5pp/2P5/b1PN1b2/P7/1Q3PPP/1B1R2K1 b - - > > > 15.01 15:31 +3.21 1...Rab8 2.Rxe7 Rxb2 3.Rf1 Rxd4 4.Re4 Rxe4 5.Bxe4 Kg7 >6.g3 Be5 7.Kg2 f5 8.Bd5 Kf6 (296.898.463) 318.8 > 15.03 58:55 +3.22++ 1...Qe5 2.Bc2 Bxc2 3.Qxc2 Rxd4 4.g3 Rad8 5.Rxd4 Rxd4 >6.Kg2 Rd2 7.gxf4 Qe1 8.Rb8+ Kh7 9.Qxd2 Qxd2 10.Kg3 g5 11.c6 (1.087.429.783) >307.5 > 15.03 77:16 +3.57++ 1...Qe5 2.Bc2 Bxc2 3.Qxc2 Rxd4 4.g3 Rad8 5.Rxd4 Rxd4 >6.Kg2 Rd2 7.gxf4 Qe1 8.Rb8+ Kh7 9.Qxd2 Qxd2 10.Kg3 Qc3+ (1.420.005.345) 306.2 > 15.03 104:53 +3.80 1...Qe5 2.Bc2 Bxc2 3.Qxc2 Rxd4 4.g3 Rad8 5.Rbb1 Rd2 6.Qb3 >Qxc5 7.Qf3 Bc7 8.Rxd2 Rxd2 9.Qe3 Qxe3 (1.920.599.184) 305.1 > >please dont post where your tunnel ram and two fours running Klotz finds Qe5 in >.0000001 second. that is not what this post is about. it is about a given >position, with a certain program finding a better score, with a better move, >in a shorter time. >kburcham
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.