Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: how many nodes for a depth 1 search?

Author: Mike Byrne

Date: 19:27:20 12/05/02

Go up one level in this thread


I set this little test up ...

starting position

[d] rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1

set threads to one -- turn off all extensions and set null moves to min  0, max
0


Crafty v19.1 SE (2 cpus)

this is a modified crafty , but I believe real crafty would behaved similarly

White(1): mt=1
max threads set to 1
White(1): sel min 0
null depth set to 0/0 (min/max)
White(1): sel max 0
null depth set to 0/0 (min/max)
White(1): sd 1
search depth set to 1.
White(1): ext in 0
one-reply extension..................0.80
in-check extension...................1.00
recapture extension..................0.80
pushpp extension.....................0.80
mate thrt extension..................0.80
White(1): ext ch 0
one-reply extension..................0.80
in-check extension...................0.00
recapture extension..................0.80
pushpp extension.....................0.80
mate thrt extension..................0.80
White(1): ext rec 0
one-reply extension..................0.80
in-check extension...................0.00
recapture extension..................0.00
pushpp extension.....................0.80
mate thrt extension..................0.80
White(1): ext push 0
one-reply extension..................0.80
in-check extension...................0.00
recapture extension..................0.00
pushpp extension.....................0.00
mate thrt extension..................0.80
White(1): ext thr 0
one-reply extension..................0.80
in-check extension...................0.00
recapture extension..................0.00
pushpp extension.....................0.00
mate thrt extension..................0.80
White(1): ext mate 0
one-reply extension..................0.80
in-check extension...................0.00
recapture extension..................0.00
pushpp extension.....................0.00
mate thrt extension..................0.00
White(1): ext one 0
one-reply extension..................0.00
in-check extension...................0.00
recapture extension..................0.00
pushpp extension.....................0.00
mate thrt extension..................0.00
White(1): book off
book file disabled.
White(1): ponder off
pondering disabled.
White(1): go
              clearing hash tables
              time surplus  30:00  time limit 1:00 (3:30)
              depth   time  score   variation (1)
              time=0.00  cpu=100%  mat=0  n=60  fh=100%  nps=10k
              ext-> chk=0 cap=0 pp=0 1rep=0 mate=0
              predicted=0  nodes=60  evals=20
              endgame tablebase-> probes done=0  successful=0
              SMP->  split=0  stop=0  data=0/32  cpu=0.00  elap=0.00

it counts 60 nodes , with 20 evals.  We know the position has 20 moves ... so
the 20 evals make sense -- where does it get the 60 nodes from -- nodes are
acounted twice in search.c and once in quiesce.c

D:\Crafty\crafty-19.1\quiesce.c(30):  tree->nodes_searched++;
D:\Crafty\crafty-19.1\search.c(51):  tree->nodes_searched++;
D:\Crafty\crafty-19.1\search.c(530):   else tree->nodes_searched++;

so in essense in some cases, it may overcount nodes in certain positions I
believe, as it appears to pass a count through the code that may not be adding
any true nodes to be evaluated  - I know Bob does't  evaluate crafty based on
his node count, but on how well it plays and I would not think he would spend a
lot time on this portion of the code ....

perhaps a better count mechanism might be in order at some point ...

all this is speculation on my part as I'm not a true programmer and Bob , I am
sure, would be better able to explain this...


Mike





This page took 0.02 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.