BGonline.org Forums

Benchmark positions for the new GNU?

Posted By: Philippe Michel
Date: Tuesday, 23 April 2013, at 2:24 p.m.

In Response To: Benchmark positions for the new GNU? (Timothy Chow)

Try rolling out some positions where rollouts with the old GNU disagreed with an XG rollout. Some such positions may be found here and here.

Not statistically significant, but it seems to prefer 8/7 (chosen by Phil Simborg at the time and not even mentioned by anybody else) over 24/23 (its clear eval. choice at every ply and the experts' preference in 2010).

The score (after 0 games) is: GNUbg 7, You 11

Move number 3: You to play 31

GNUbg	160
You	214
Position ID: 4HPkARFgjjwgWQ Match ID: cIkFAHAAWAAA

• You moves bar/22 8/7

#		Ply	Move	Equity
•	1	R	bar/22 8/7	-0.232
			0.477 0.089 0.003 - 0.523 0.262 0.024 -0.240 -0.232 0.004 0.004 0.002 - 0.004 0.005 0.002 0.013 0.022
			Full cubeful rollout with var.redn.
			1296 games, Mersenne Twister dice gen. with seed 677159275 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 2 0-ply moves and up to 6 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]
	2	R	bar/22 24/23	-0.258 ( -0.027)
			0.458 0.096 0.004 - 0.542 0.242 0.015 -0.241 -0.258 0.004 0.004 0.001 - 0.004 0.005 0.001 0.013 0.021
			Full cubeful rollout with var.redn.
			1296 games, Mersenne Twister dice gen. with seed 677159275 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 2 0-ply moves and up to 6 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]
	3	R	bar/24 8/5*	-0.290 ( -0.059)
			0.464 0.099 0.004 - 0.536 0.251 0.033 -0.253 -0.290 0.004 0.004 0.001 - 0.004 0.004 0.001 0.014 0.021
			Full cubeful rollout with var.redn.
			1296 games, Mersenne Twister dice gen. with seed 677159275 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 2 0-ply moves and up to 6 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]
	4	R	bar/22 11/10	-0.292 ( -0.060)
			0.460 0.091 0.005 - 0.540 0.243 0.019 -0.247 -0.292 0.003 0.003 0.001 - 0.003 0.003 0.001 0.011 0.018
			Full cubeful rollout with var.redn.
			1296 games, Mersenne Twister dice gen. with seed 677159275 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 2 0-ply moves and up to 6 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]
	5	R	bar/22 6/5*	-0.308 ( -0.077)
			0.465 0.079 0.006 - 0.535 0.283 0.031 -0.300 -0.308 0.007 0.008 0.002 - 0.007 0.009 0.003 0.028 0.040
			Full cubeful rollout with var.redn.
			1296 games, Mersenne Twister dice gen. with seed 677159275 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 2 0-ply moves and up to 6 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]

Messages In This Thread

• Benchmark positions for the new GNU?
  Timothy Chow -- Monday, 22 April 2013, at 1:27 a.m.
  • Benchmark positions for the new GNU?
    Bill Patterson -- Monday, 22 April 2013, at 4:35 a.m.
  • Benchmark positions for the new GNU?
    Mike Petch -- Monday, 22 April 2013, at 5:30 a.m.
    • Benchmark positions for the new GNU?
      eXtreme Gammon -- Tuesday, 23 April 2013, at 12:59 a.m.
  • Benchmark positions for the new GNU?
    Philippe Michel -- Tuesday, 23 April 2013, at 2:24 p.m.