Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: 2 points Chandler One Point for you to Ponder

Author: Anthony Cozzie

Date: 09:16:53 01/16/05

Go up one level in this thread


Please find the error in one or more of these statements Chandler:

1. A tablebase access takes at least 2-3 ms, even on a 15K SCSI hard drve.

2. Most engines (except for Diep ;) have evaluations that take a few
microseconds or so.

3. Therefore a tablebase access is on the order of 1000X slower than calling the
evaluation function.

4. Since most of the time is spent in the evaluation function, this means if the
programmer called tablebase_probe() at every node it would run 1000X slower,
that is 1K nps instead of 1M nps in Zappa's case.

Note that we are talking about positions with (say) 8 pieces on the board; for
example this KRPPPvKRP position.  White wants to keep his passers together and
win, Black hopes to split them and (possibly) draw.

[D]8/8/r6k/6p1/6P1/5P1P/R6K/8 w - - 0 1

When a lot of people (experts especially) think something, it is usually good to
at least consider their position.  They might be wrong anyway, but they probably
have at least some justification for their opinion.  I get the feeling that you
never read anything I write, but I keep trying for the sake of world (or at
least forum) peace anyway :)

anthony



This page took 0.01 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.