Lc0 vs 1# pos BT-2630

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

Moderators: hgm, Rebel, chrisw

hammerklavier
Posts: 113
Joined: Mon Nov 28, 2011 10:17 am

Lc0 vs 1# pos BT-2630

Post by hammerklavier »

Hi Friends! ,
[d]rq2r1k1/5pp1/p7/4bNP1/1p2P2P/5Q2/PP4K1/5R1R w - - 0 1

This is strange. Leela cannot found Nxg7 ???

I use leela in CPU mode, maybe some friend can help me with the analysis in a power GPU.


Thanks to everyone!
zullil
Posts: 6442
Joined: Tue Jan 09, 2007 12:31 am
Location: PA USA
Full name: Louis Zulli

Re: Lc0 vs 1# pos BT-2630

Post by zullil »

hammerklavier wrote: Wed Aug 07, 2019 4:45 pm Hi Friends! ,
[d]rq2r1k1/5pp1/p7/4bNP1/1p2P2P/5Q2/PP4K1/5R1R w - - 0 1

This is strange. Leela cannot found Nxg7 ???

I use leela in CPU mode, maybe some friend can help me with the analysis in a power GPU.


Thanks to everyone!

Code: Select all

$ ./lc0
       _
|   _ | |
|_ |_ |_| v0.22.0 built Aug  5 2019
setoption name Backend value cudnn-fp16
setoption name NNCacheSize value 20000000
ucinewgame
Found pb network file: ./network42850
Creating backend [cudnn-fp16]...
GPU: GeForce RTX 2080 Ti
GPU memory: 10.7534 Gb
GPU clock frequency: 1635 MHz
GPU compute capability: 7.5
CUDA Runtime version: 10.1.0
Cudnn version: 7.6.2
Latest version of CUDA supported by the driver: 10.1.0
position fen rq2r1k1/5pp1/p7/4bNP1/1p2P2P/5Q2/PP4K1/5R1R w - - 0 1
go infinite
info depth 1 seldepth 2 time 24 nodes 3 score cp 179 hashfull 0 nps 125 tbhits 0 pv f5e3 a8a7
info depth 1 seldepth 2 time 28 nodes 5 score cp 220 hashfull 0 nps 178 tbhits 0 pv f5h6 g7h6
info depth 2 seldepth 3 time 32 nodes 12 score cp 131 hashfull 0 nps 375 tbhits 0 pv f5e3 b8b7 e3d5
info depth 2 seldepth 4 time 41 nodes 20 score cp 147 hashfull 0 nps 487 tbhits 0 pv f1f2 a8a7 h4h5 b8b6
info depth 3 seldepth 4 time 44 nodes 38 score cp 135 hashfull 0 nps 863 tbhits 0 pv f1f2 b8b6 h4h5 a8d8 h5h6
info depth 3 seldepth 5 time 48 nodes 52 score cp 132 hashfull 0 nps 1083 tbhits 0 pv f1f2 b8b6 h4h5 b6e6 f3g4
info depth 3 seldepth 5 time 53 nodes 64 score cp 146 hashfull 0 nps 1207 tbhits 0 pv h4h5 a8a7 g5g6 b8b6 f3g4
info depth 4 seldepth 6 time 57 nodes 109 score cp 136 hashfull 0 nps 1912 tbhits 0 pv f1f2 b8b6 h4h5 a8d8 f5e3 b6e6
info depth 4 seldepth 7 time 68 nodes 126 score cp 136 hashfull 0 nps 1852 tbhits 0 pv f1f2 b8b6 h4h5 a8d8 f5e3 b6e6
info depth 4 seldepth 8 time 74 nodes 129 score cp 136 hashfull 0 nps 1743 tbhits 0 pv f1f2 b8b6 h4h5 a8d8 f5e3 b6e6
info depth 4 seldepth 9 time 81 nodes 157 score cp 136 hashfull 0 nps 1938 tbhits 0 pv f1f2 b8b6 h4h5 a8d8 f5e3 b6e6
info depth 4 seldepth 10 time 85 nodes 164 score cp 136 hashfull 0 nps 1929 tbhits 0 pv f1f2 b8b6 h4h5 a8d8 f5e3 b6e6
info depth 5 seldepth 11 time 90 nodes 230 score cp 139 hashfull 0 nps 2555 tbhits 0 pv f1f2 b8b7 h4h5 a8d8 g5g6 e8e6 f3g4 e6f6
info depth 6 seldepth 12 time 123 nodes 463 score cp 137 hashfull 0 nps 3764 tbhits 0 pv f1f2 b8b6 h4h5 b6e6 g5g6 f7g6 h5g6
info depth 6 seldepth 12 time 127 nodes 506 score cp 239 hashfull 0 nps 3984 tbhits 0 pv f5g7 e5g7 f3f7 g8h7 f7h5 h7g8 g5g6 g7h6 h5h6 a8a7 h6f4 b8a8
info depth 6 seldepth 13 time 131 nodes 574 score cp 231 hashfull 0 nps 4381 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 a8a7 f1f4 b8e5 h1f1
info depth 7 seldepth 13 time 135 nodes 607 score cp 236 hashfull 0 nps 4496 tbhits 0 pv f5g7 e5g7 f3f7 g8h7 f7h5 h7g8 g5g6 g7h6 h5h6 a8a7 h6f4 b8a8 f1e1
info depth 7 seldepth 14 time 152 nodes 732 score cp 237 hashfull 0 nps 4815 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 a8a7 f1f4 b8d6 h1f1
info depth 7 seldepth 15 time 170 nodes 1008 score cp 257 hashfull 0 nps 5929 tbhits 0 pv f5g7 g8g7 f3f7 g7h8 f7h5 h8g8 f1f7 e5g7 g5g6 b8g3 g2g3 g7e5 h5e5
info depth 7 seldepth 16 time 178 nodes 1164 score cp 266 hashfull 0 nps 6539 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 a8a7 f1f4 a7g7 h1f1 b8d6
info depth 8 seldepth 16 time 187 nodes 1381 score cp 285 hashfull 0 nps 7385 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 a8a7 f1f4 a7g7 h1f1 b8d6
info depth 8 seldepth 17 time 198 nodes 1646 score cp 301 hashfull 0 nps 8313 tbhits 0 pv f5g7 g8g7 f3f7 g7h8 f7h5 h8g8 f1f7 e5g7 g5g6 b8g3 g2g3 g7e5 h5e5
info depth 8 seldepth 18 time 227 nodes 2416 score cp 316 hashfull 0 nps 10643 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8b7 h6h7 b7h7 g6h7 g8h7 g2f3 b4b3
info depth 8 seldepth 19 time 246 nodes 2924 score cp 305 hashfull 0 nps 11886 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8b7 f1f4 a8d8 h1e1 d8d2 g2g3 d2d3 f4f3
info depth 8 seldepth 20 time 267 nodes 3975 score cp 308 hashfull 0 nps 14887 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8b7 h6h7 b7h7 g6h7 g8h7 g2f3 b4b3 a2a3
info depth 9 seldepth 21 time 300 nodes 5905 score cp 320 hashfull 0 nps 19683 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8b7 f1f4 e8e4 h6h7 b7h7 g6h7 g8h7 f4e4 a8g8 g2f3
info depth 9 seldepth 22 time 361 nodes 9255 score cp 340 hashfull 0 nps 25637 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8b7 f1f4 e8e4 h6h7 b7h7 g6h7 g8h7 f4e4 a8g8 g2f2 g8g4
info depth 9 seldepth 23 time 420 nodes 11916 score cp 350 hashfull 0 nps 28371 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8c7 h6f4 c7c2 f1f2 c2e4 f4e4 e8e4 h4h5 g8g7 f2f7 g7h6 f7h7 h6g5 h1h3
info depth 9 seldepth 24 time 474 nodes 14325 score cp 356 hashfull 0 nps 30221 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8c7 h6f4 c7c2 f1f2 c2e4 f4e4 e8e4 h4h5 g8g7 f2f7 g7h6 f7h7 h6g5 h1h3
info depth 10 seldepth 24 time 568 nodes 18117 score cp 359 hashfull 0 nps 31896 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8c7 h6f4 c7c2 f1f2 c2e4 f4e4 e8e4 h4h5 g8g7 f2f7 g7h6 f7h7 h6g5 h1h3
info depth 10 seldepth 25 time 577 nodes 18480 score cp 358 hashfull 0 nps 32027 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8c7 h6f4 c7c2 f1f2 c2e4 f4e4 e8e4 h4h5 g8g7 f2f7 g7h6 f7h7 h6g5 h1h3
info depth 10 seldepth 26 time 807 nodes 29934 score cp 370 hashfull 0 nps 37092 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8c7 h6f4 c7c2 f1f2 c2e4 f4e4 e8e4 h4h5 g8g7 f2f7 g7h6 f7h7 h6g5 h1h3
info depth 10 seldepth 27 time 1432 nodes 65932 score cp 397 hashfull 1 nps 46041 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8b7 f1f4 e8e4 h6h7 b7h7 g6h7 g8h8 f4e4 a8g8 h7g8q h8g8 h4h5 g8h7 h5h6 a6a5
info depth 10 seldepth 28 time 1458 nodes 67428 score cp 396 hashfull 1 nps 46246 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8b7 f1f4 e8e4 h6h7 b7h7 g6h7 g8h8 f4e4 a8g8 h7g8q h8g8 h4h5 g8h7 h5h6 a6a5
info depth 11 seldepth 28 time 1627 nodes 77144 score cp 400 hashfull 1 nps 47414 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8b7 f1f4 a8d8 h1f1 d8d2 g2g3 d2d3 g3h2 d3d2 f1f2 d2f2
info depth 11 seldepth 29 time 2066 nodes 105489 score cp 414 hashfull 1 nps 51059 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8b7 f1f4 a8d8 h1f1 d8d2 g2g3 b7c7 f1f3 d2b2 h4h5 b2c2 h6h7
info depth 11 seldepth 30 time 2164 nodes 112105 score cp 420 hashfull 2 nps 51804 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5h6 b8b7 f1f4 a8d8 h1f1 d8d2 g2g3 b7c7 f1f3 d2b2 h4h5 b2c2 h6h7
info depth 11 seldepth 31 time 3483 nodes 192874 score cp 425 hashfull 3 nps 55375 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5d5 g8h8 f1f7 a8a7 d5d4 b8e5 d4e5 e8e5 f7a7 h6c1 h1c1 e5g5 g2f2 g5d5 e4e5 d5c5
info depth 12 seldepth 31 time 5581 nodes 332582 score cp 450 hashfull 5 nps 59591 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5d5 g8h8 f1f7 a8a7 d5d4 h8g8 f7a7 b8c8 d4d5 g8h8 a7h7
info depth 11 seldepth 31 time 5599 nodes 333692 score cp 450 hashfull 5 nps 59598 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5d5 g8h8 f1f7 a8a7 d5d4 h8g8 f7a7 b8c8 d4d5 g8h8 a7h7
info depth 12 seldepth 31 time 5751 nodes 343534 score cp 452 hashfull 5 nps 59734 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5d5 g8h8 f1f7 a8a7 d5d4 h8g8 f7a7 b8c8 d4d5 g8h8 a7h7
info depth 12 seldepth 32 time 6082 nodes 364436 score cp 453 hashfull 6 nps 59920 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5d5 g8h8 f1f7 a8a7 d5d4 h8g8 f7a7 b8c8 d4d5 g8h8 a7h7
stop
info depth 12 seldepth 32 time 8928 nodes 567331 score cp 467 hashfull 8 nps 63545 tbhits 0 pv f5g7 e5g7 f3f7 g8h8 f7h5 h8g8 g5g6 g7h6 h5d5 g8h8 f1f7 a8a7 d5d4 h8g8 f7a7 b8c8 d4d5 g8h8 a7h7
bestmove f5g7 ponder e5g7
Max
Posts: 247
Joined: Tue Apr 13, 2010 10:41 am

Re: Lc0 vs 1# pos BT-2630

Post by Max »

hammerklavier wrote: Wed Aug 07, 2019 4:45 pmThis is strange. Leela cannot found Nxg7 ???

I use leela in CPU mode, maybe some friend can help me with the analysis in a power GPU.


Thanks to everyone!
What network did you choose?

Bad-gyal-5 needs 4 seconds - also on cpu - and less than 1000 nodes.
BT-2630-1.png
Hope we're not just the biological boot loader for digital super intelligence. Unfortunately, that is increasingly probable - Elon Musk
hammerklavier
Posts: 113
Joined: Mon Nov 28, 2011 10:17 am

Re: Lc0 vs 1# pos BT-2630

Post by hammerklavier »

Thanks Friends! I used 11258-48x5-se, but now change to 42850 and found in 20 sec....

What network recommend in a old cpu (Pentium E2200 2,2Ghz)?

Thanks !

Regards from Buenos Aires Argentina!
Max
Posts: 247
Joined: Tue Apr 13, 2010 10:41 am

Re: Lc0 vs 1# pos BT-2630

Post by Max »

hammerklavier wrote: Wed Aug 07, 2019 5:48 pm Thanks Friends! I used 11258-48x5-se, but now change to 42850 and found in 20 sec....

What network recommend in a old cpu (Pentium E2200 2,2Ghz)?

Thanks !

Regards from Buenos Aires Argentina!
You know, that depends on what you want to do.

Anyway, I would go for a small net (128x10 or less). E.g bad-gyal-5 or something like 11258-96x8-se and follow the evolution of the new 56-net.

Maybe an used Nvidia GTX 10xx GPU could be an option for an old pc too? But I'm no expert here; living in the mac diaspora 8-)
Hope we're not just the biological boot loader for digital super intelligence. Unfortunately, that is increasingly probable - Elon Musk
Robert Flesher
Posts: 1280
Joined: Tue Aug 18, 2009 3:06 am

Re: Lc0 vs 1# pos BT-2630

Post by Robert Flesher »

New game
rq2r1k1/5pp1/p7/4bNP1/1p2P2P/5Q2/PP4K1/5R1R w - - 0 1

Analysis by Lc0 v0.21.1 0.1:

1.Nxg7 Bxg7 2.Qxf7+ Kh8 3.Qh5+ Kg8 4.g6 Bh6 5.Qxh6 Qb7 6.Rf4 Rad8 7.Rhf1 Rd2+ 8.Kg3 Qc7 9.R1f3 Rxb2 10.h5 Rc2 11.Qh7+ Qxh7 12.gxh7+
+- (19.77) Depth: 12/28 00:00:12 205kN
(Doe, 07.08.2019)


No problem on my hardware
Nay Lin Tun
Posts: 708
Joined: Mon Jan 16, 2012 6:34 am

Re: Lc0 vs 1# pos BT-2630

Post by Nay Lin Tun »

This is my Lc0 on my Samsung S10 with Little Demon 2 net on android Leela.
Leela see 1. Nxg7 within 15 seconds and stick with it.

Note. She found it around 4.5k nodes. It means, 1060 GTX will find in 2 seconds and 20xx RTX GPUs will find in 1 second.
mesilikas
Posts: 39
Joined: Mon Jan 01, 2018 12:54 pm

Re: Lc0 vs 1# pos BT-2630

Post by mesilikas »

Max wrote: Wed Aug 07, 2019 6:37 pm
hammerklavier wrote: Wed Aug 07, 2019 5:48 pm Thanks Friends! I used 11258-48x5-se, but now change to 42850 and found in 20 sec....

What network recommend in a old cpu (Pentium E2200 2,2Ghz)?

Thanks !

Regards from Buenos Aires Argentina!
You know, that depends on what you want to do.

Anyway, I would go for a small net (128x10 or less). E.g bad-gyal-5 or something like 11258-96x8-se and follow the evolution of the new 56-net.

Maybe an used Nvidia GTX 10xx GPU could be an option for an old pc too? But I'm no expert here; living in the mac diaspora 8-)
Max how can you group 56-net? When i seek for a network in lc0 official site i see all networks stirred. Can you explain?
Nay Lin Tun
Posts: 708
Joined: Mon Jan 16, 2012 6:34 am

Re: Lc0 vs 1# pos BT-2630

Post by Nay Lin Tun »

I suggest Little Demon 2, 10x128 net.
https://lc0.org/ld2

There could be other better smaller nets ( only minimal) , LD2 is the one officially using in TCEC and CCC .
Max
Posts: 247
Joined: Tue Apr 13, 2010 10:41 am

Re: Lc0 vs 1# pos BT-2630

Post by Max »

mesilikas wrote: Thu Aug 08, 2019 11:07 pm Max how can you group 56-net? When i seek for a network in lc0 official site i see all networks stirred. Can you explain?
Did you click on the 56-net link above? It should lead you direct to official run 2, where these networks are published.
Hope we're not just the biological boot loader for digital super intelligence. Unfortunately, that is increasingly probable - Elon Musk