Many Faces is a strong traditional program.  It evaluates about 100 to 150
positions/sec in the middle game, and up to 500 positions/sec in the
endgame.  This is on one processor of a 2.3 GHz core duo.  It's for version
12 (not released).  Version 11 is slower.

 

I plan to make it faster, but right now I'm working on making the search
more efficient. 

 

Here is output from a middle game position against gnugo (at move 80).  Each
iteration only takes about twice as long as the previous one.  This one is
125 evaluations per second.  The tactical search evaluates 220377 nodes per
second.

 

Search 18-36 seconds (10000 max evals),   0.3 secs in life reading level 5,
life reading 35 life()

New best 2432 (strat 352)d16  d16 d18 e18 c19 

New best 2691 (strat 497)d18  d18 

New best 2741 (strat 447)m13  m13 d18 b17 

New best 2994 (strat 151)q11  q11 q12 

Iteration 1 complete 20 moves in  2.46 secs.  Total evals 304, search life()
269, 

Final value is 2994 for q11 : q11 q12 

New best 2994 (strat 151)q11  q11 q12 

Iteration 2 complete 20 moves in  5.59 secs.  Total evals 698, search life()
663, 

Final value is 2994 for q11 : q11 q12 

New best 2898 (strat 151)q11  q11 q12 m13 o16 

Iteration 3 complete 20 moves in 11.56 secs.  Total evals 1420, search
life() 1385, 

Final value is 2898 for q11 : q11 q12 m13 o16 

New best 2831 (strat 151)q11  q11 q12 p10 o12 

Iteration 4 complete 20 moves in 25.74 secs.  Total evals 3223, search
life() 3188, 

Final value is 2831 for q11 : q11 q12 p10 o12 

Pass val 40.2, best val 56.6

 25.7 secs. life() 3223 (125/s), searchevals 1124, Nodes 1851 ( 72/s), full
gen 72, Q gen 1603, Moves 1848( 72/s), tree nodes 114/114, list 11472/11472,
tac nodes 5672726 (220377/s), evalopj 483514 (  9%), life 5153443 ( 91%) [
fixgralive 13538182 (239%) miai 3823111 ( 67%) tvpot 3669677 ( 65%), group
270464 (  5%), conn 342851 (  6%), eye 604303 ( 11%) ]

 

David

 

From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of mingwu
Sent: Monday, January 14, 2008 5:45 PM
To: computer-go@computer-go.org
Subject: [computer-go] On average how many board updates/sec can top Go
programs do these days?

 

Hi,

I read on the web, and some other places that most Go programs can only
evaluate "a dozen" of moves per second.  Is this still true today on a
typical machine, say, single 2GHz CPU, 2GB memory?

And if this is still true, how can we make it faster?

To make the question more precise, I define a board updates as: suppose the
program places a new move on an existing board, and then update all the
blocks, dragons, eyes, connections, territory ... info, and output the
evaluation as a score ( e.g. B leads by 15.5 points).

I also read that UTC programs choose a move by running lots of simulations,
are their update speed any faster? Or they evaluation lots of boards, but
for each move they only calculate some very simple information (to me, it
will still be a surprise, because to evaluate a Go board, one at least have
to know the life/death of each dragon, that would require lots of
computation, which I think is the primary reason that why Go programs is
slow compared to Chess programs that can evaluate thousands or even millions
of boards per second). 

I'm just curious, any info / thoughts / comments is appreciated.






_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

Reply via email to