Position test without and with SYzygy bases

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

BBauer
Posts: 658
Joined: Wed Mar 08, 2006 8:58 pm

Position test without and with SYzygy bases

Post by BBauer »

The following position is from a recent game
Spike-Nebula TCEC Season 6, Stage 1b

[D]8/8/4R1p1/1p1n1k2/8/5K2/1P6/8 w - - [D]

Code: Select all

Stockfish-090314 without Syzygy bases:
 depth   time    nodes          nps     score   pv
 49/68	03:17	 658.825k	3.344k	+2,79	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 50/68	03:39	 752.169k	3.434k	+2,79	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 51/70	05:16	 1.161.010k	3.674k	+2,72	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 52/70	05:49	 1.306.862k	3.742k	+2,72	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 53/70+	06:40	 1.519.360k	3.792k	+2,80	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 53/70	07:32	 1.730.672k	3.827k	+2,80	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 54/70	08:31	 1.954.294k	3.822k	+2,80	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 55/70	09:26	 2.184.644k	3.859k	+2,80	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 56/70	10:35	 2.482.621k	3.906k	+2,80	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 57/75	10:39	 2.497.785k	3.906k	+2,80	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4

Stockfish-090314 with 5-piece bases SyzygyProbeLimit=5 SyzygyProbeDepth=1:
 depth   time    nodes          nps     score   pv
 49/60	00:43	 141.856k	3.263k	+2,86	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 50/60	00:49	 162.894k	3.303k	+2,86	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 51/60	00:55	 187.339k	3.369k	+2,86	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 52/60	01:08	 231.331k	3.389k	+2,86	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 53/60	01:15	 258.460k	3.431k	+2,86	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 54/60	01:25	 299.253k	3.499k	+2,86	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 55/60	01:39	 351.383k	3.517k	+2,86	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 56/60	01:52	 402.735k	3.567k	+2,86	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 57/60	02:07	 461.293k	3.630k	+2,86	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 58/72	02:48	 612.072k	3.638k	+2,86	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 59/72	04:52	 1.018.642k	3.486k	+2,86	Te6-d6 Kf5-e5 Td6xg6 Ke5-d4
 60/72+	05:51	 1.207.411k	3.439k	+2,94	Te6-c6 Kf5-e5 Tc6-c5 b5-b4 
 60/72+	06:01	 1.243.295k	3.435k	+3,02	Te6-c6 Kf5-e5 Tc6-c5 b5-b4 
 60/72+	06:12	 1.278.909k	3.434k	+3,14	Te6-c6 Kf5-e5 Tc6-c5 b5-b4
 60/72	06:19	 1.303.221k	3.433k	+3,14	Te6-c6 Kf5-e5 Tc6-c5 b5-b4

Stockfish-090314 with 5-piece bases SyzygyProbeLimit=6 SyzygyProbeDepth=7:
 depth   time    nodes          nps     score   pv
 20/32	00:01	 1.733k		1.215k	+1,72	Te6-e8 b5-b4 Te8-c8 Kf5-e5 Tc8-c4
 21/32	00:01	 2.185k		1.212k	+1,88	Te6-c6 Kf5-e5 Tc6-c5 b5-b4 Tc5-c4
 22/32	00:01	 2.191k		1.213k	+1,88	Te6-c6 Kf5-e5 Tc6-c5 b5-b4 Tc5-c4
 23/32	00:02	 2.532k		1.262k	+2,15	Te6-c6 Kf5-e5 Tc6-c5 b5-b4 Tc5-c4
 24/36	00:02	 3.380k		1.300k	+3,34	Te6-c6 Kf5-e5 Tc6-c5 b5-b4 Tc5-c4
 25/36	00:03	 4.681k		1.206k	+150,77	Te6-c6 Kf5-e5 Tc6-c5 b5-b4 Tc5-c4
 26/36	00:04	 5.526k		1.252k	+150,77	Te6-c6 Kf5-e5 Tc6-c5 b5-b4 Tc5-c4
 27/36	00:05	 6.882k		1.309k	+150,77	Te6-c6 Kf5-e5 Tc6-c5 b5-b4 Tc5-c4
 28/36	00:05	 8.153k		1.386k	+150,77	Te6-c6 Kf5-e5 Tc6-c5 b5-b4 Tc5-c4
 29/36	00:06	 10.086k	1.475k	+150,77	Te6-c6 Kf5-e5 Tc6-c5 b5-b
 30/36	00:07	 11.872k	1.532k	+150,77	Te6-c6 Kf5-e5 Tc6-c5 b5-b
 31/36	00:09	 15.146k	1.628k	+150,77	Te6-c6 Kf5-e5 Tc6-c5 b5-b
 32/36	00:10	 19.099k	1.745k	+150,77	Te6-c6 Kf5-e5 Tc6-c5 b5-b
So on my small notebook I get:
Stockfish-090314 does not find Rc6 within 10 minutes.
Stockfish-090314 with 5-piece bases finds Rc6 in 5:51 minutes.
Stockfish-090314 with 6-piece bases finds Rc6 in 1 second.
Kind regards
Top
syzygy
Posts: 5563
Joined: Tue Feb 28, 2012 11:56 pm

Re: Position test without and with SYzygy bases

Post by syzygy »

Nice position :-)

It seems SF can find Re6 in the end:

Code: Select all

info depth 57 seldepth 77 score cp 293 lowerbound nodes 3664624523 nps 15069782 tbhits 0 time 243177 multipv 1 pv e6c6 b5b4 c6c5 f5e5 c5c4 e5d6 f3e2 d6e5 e2d3 e5d6 d3d2 d6e5 c4c6 e5f5 d2d3 d5f4 d3d4 f4g2 c6b6 g2e1 d4e3 e1c2 e3d2 b4b3 b6b3 c2d4 b3b6 g6g5 b6b4 f5e5 d2d3 d4e6 d3e3 e6c7 b4e4 e5f5 e4c4 c7d5 e3d4 d5f6 b2b4 g5g4 d4e3 f6d7 c4c7 f5e6 c7c6 e6e5 c6g6 e5f5 g6g7 d7f6 b4b5 f6e8 g7e7 e8d6 e7c7 f5e5 c7c5 e5e6 b5b6 g4g3 e3f3 d6b7 c5c7 g3g2 f3g2 b7d6 b6b7

...

info depth 58 seldepth 80 score cp 700 lowerbound nodes 6688451066 nps 13864545 tbhits 0 time 482414 multipv 1 pv e6c6 f5e5 c6c5 e5d4 c5b5 d5e3 b5b4 d4c5 b4b8 e3f5 b8c8 c5d4 c8d8 d4c4 d8g8 f5e7 g8b8 e7d5 b8d8 d5b6 f3e4 b6a4 d8c8 c4b3 c8b8 b3c2 b2b4 g6g5 b4b5 c2d2 e4d4 a4c3 b5b6 c3e2 d4c4 d2e1 b8f8
Probing 5-piece tables (probedepth=1) it finds it at depth 58, so finding it at depth 57 without probing was probably lucky:

Code: Select all

info depth 58 seldepth 75 score cp 292 lowerbound nodes 607194364 nps 12975070 tbhits 4193825 time 46797 multipv 1 pv e6c6 f5e5 c6c5 b5b4 c5c4 e5f5 c4g4 f5f6 f3e4 d5b6 e4d3 b4b3 g4b4 b6d7 b4b5 g6g5 d3e3 d7f8 b5b3 f6f5 b3b5 f5g4 b2b4 f8d7 b5a5 d7b6 e3e4 b6c4 a5c5 c4d2 e4d3 d2f3 d3e3 f3h2 b4b5

...

info depth 59 seldepth 88 score cp 2274 lowerbound nodes 1300053531 nps 12210629 tbhits 8714471 time 106469 multipv 1 pv e6c6 f5e5 c6c5 b5b4 c5c4 e5f5 c4c6 g6g5 c6c5 f5e5 c5c4 e5f5 f3e2 d5f4 e2e3 f4d5 e3d3 d5f6 c4b4 g5g4 b4c4 g4g3 d3e2 g3g2 e2f2 f6g4 f2g1 g4e3 c4c5 f5e4 b2b4 e3d5 b4b5 d5e3 c5c8 e4d4 b5b6 d4d3 b6b7 d3e2 c8c3
So this is also a nice example of how 5-piece tables may help to search deeper more quickly.

Spike apparently failed to find it despite a lot of TB hits (5692.2 K = 5.7 million?).
Top
ouachita
Posts: 454
Joined: Tue Jan 15, 2013 4:33 pm
Location: Ritz-Carlton, NYC
Full name: Bobby Johnson

Re: Position test without and with SYzygy bases

Post by ouachita »

BBauer wrote:
my Stockfish-100314 no tb finds Rc6 in 2:30.

my Stockfish-090314 with 5-piece bases finds Rc6 in 00:34 sec.

my Stockfish-090314 with 6-piece bases finds Rc6 in ~1 second.
SIM, PhD, MBA, PE
Top
User avatar
michiguel
Posts: 6401
Joined: Thu Mar 09, 2006 8:30 pm
Location: Chicago, Illinois, USA

Re: Position test without and with SYzygy bases

Post by michiguel »

syzygy wrote:Nice position :-)

It seems SF can find Re6 in the end:

Code: Select all

info depth 57 seldepth 77 score cp 293 lowerbound nodes 3664624523 nps 15069782 tbhits 0 time 243177 multipv 1 pv e6c6 b5b4 c6c5 f5e5 c5c4 e5d6 f3e2 d6e5 e2d3 e5d6 d3d2 d6e5 c4c6 e5f5 d2d3 d5f4 d3d4 f4g2 c6b6 g2e1 d4e3 e1c2 e3d2 b4b3 b6b3 c2d4 b3b6 g6g5 b6b4 f5e5 d2d3 d4e6 d3e3 e6c7 b4e4 e5f5 e4c4 c7d5 e3d4 d5f6 b2b4 g5g4 d4e3 f6d7 c4c7 f5e6 c7c6 e6e5 c6g6 e5f5 g6g7 d7f6 b4b5 f6e8 g7e7 e8d6 e7c7 f5e5 c7c5 e5e6 b5b6 g4g3 e3f3 d6b7 c5c7 g3g2 f3g2 b7d6 b6b7

...

info depth 58 seldepth 80 score cp 700 lowerbound nodes 6688451066 nps 13864545 tbhits 0 time 482414 multipv 1 pv e6c6 f5e5 c6c5 e5d4 c5b5 d5e3 b5b4 d4c5 b4b8 e3f5 b8c8 c5d4 c8d8 d4c4 d8g8 f5e7 g8b8 e7d5 b8d8 d5b6 f3e4 b6a4 d8c8 c4b3 c8b8 b3c2 b2b4 g6g5 b4b5 c2d2 e4d4 a4c3 b5b6 c3e2 d4c4 d2e1 b8f8
Probing 5-piece tables (probedepth=1) it finds it at depth 58, so finding it at depth 57 without probing was probably lucky:

Code: Select all

info depth 58 seldepth 75 score cp 292 lowerbound nodes 607194364 nps 12975070 tbhits 4193825 time 46797 multipv 1 pv e6c6 f5e5 c6c5 b5b4 c5c4 e5f5 c4g4 f5f6 f3e4 d5b6 e4d3 b4b3 g4b4 b6d7 b4b5 g6g5 d3e3 d7f8 b5b3 f6f5 b3b5 f5g4 b2b4 f8d7 b5a5 d7b6 e3e4 b6c4 a5c5 c4d2 e4d3 d2f3 d3e3 f3h2 b4b5

...

info depth 59 seldepth 88 score cp 2274 lowerbound nodes 1300053531 nps 12210629 tbhits 8714471 time 106469 multipv 1 pv e6c6 f5e5 c6c5 b5b4 c5c4 e5f5 c4c6 g6g5 c6c5 f5e5 c5c4 e5f5 f3e2 d5f4 e2e3 f4d5 e3d3 d5f6 c4b4 g5g4 b4c4 g4g3 d3e2 g3g2 e2f2 f6g4 f2g1 g4e3 c4c5 f5e4 b2b4 e3d5 b4b5 d5e3 c5c8 e4d4 b5b6 d4d3 b6b7 d3e2 c8c3
So this is also a nice example of how 5-piece tables may help to search deeper more quickly.
I have been saying that for years, but nobody listened to me.

Miguel
Spike apparently failed to find it despite a lot of TB hits (5692.2 K = 5.7 million?).
Top
syzygy
Posts: 5563
Joined: Tue Feb 28, 2012 11:56 pm

Re: Position test without and with SYzygy bases

Post by syzygy »

michiguel wrote:I have been saying that for years, but nobody listened to me.
I have said so before, too.
Top
User avatar
hgm
Posts: 27794
Joined: Fri Mar 10, 2006 10:06 am
Location: Amsterdam
Full name: H G Muller

Re: Position test without and with SYzygy bases

Post by hgm »

Well, it wouldn't be possible to beat the 1 sec, but I would be surprised if on-the-fly retrograde analysis could not easily beat the 5 minutes for this position. Only ~100 Pawn structures are still reachable.
Top
ernest
Posts: 2041
Joined: Wed Mar 08, 2006 8:30 pm

Re: Position test without and with SYzygy bases

Post by ernest »

BBauer wrote:So on my small notebook I get:
Houdini 4 with 5-men syzygy finds Rc6 immediately! 8-)
Top

Return to “Computer Chess Club: General Topics”