Author: Uri Blass
Date: 11:24:47 11/23/05
Go up one level in this thread
On November 23, 2005 at 14:00:00, chandler yergin wrote: >On November 22, 2005 at 05:20:59, Uri Blass wrote: > >>On November 22, 2005 at 05:17:44, Lin Harper wrote: >> >>>On November 22, 2005 at 02:35:15, Alexander Ermolenko wrote: >>> >>>>Why max Hashtable Limit for Fruit is 1024 ? >>> This does'nt answer your question, but I just as an experiment, set up a >>>position and analyze it with just 1mb hash. Check the nps. Now reset the same >>>position and set your hash to the maximum. Check the nps again. See a >>>difference? >>>There appears to be a trade-off between speed and hash size. I'm no expert, but >>>it seems to me that optimal playing strength will be had with a moderate hash >>>size. >>> regards >>> Lin >> >>I think that this is the wrong way to test hash. >>You should test time to solution and not nodes per second. >What solution? You can choose test positions and find the time to find the right move and compare. Here are some examples from the gcp test suite [D]1r5r/3b1pk1/3p1np1/p1qPp3/p1N1PbP1/2P2PN1/1PB1Q1K1/R3R3 b - - 01 bm Nxg4 [D]b3r3/q2B2k1/3Q2p1/1p5p/3pP3/5P2/1p4PP/5RK1 b - - bm d3+ [D]r1bq1rk1/pp2bp1p/2p2np1/3p2B1/3P4/2NQ2N1/PPP2PPP/4RRK1 w - - bm Rxe7 [D]2r1rbk1/5pp1/bq5p/1pnBPN2/6Q1/N7/5PPP/R2R2K1 w - - bm Bxf7+ [D]2r1r1k1/pp1nbpp1/4pn1p/q3NN1P/P1pP1B2/2P5/1PQ2PP1/R3R1K1 w - - bm Nxg7 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.