Obsidian 14.0 and three mover

Jouni · Post by **Jouni** » Thu Oct 31, 2024 8:17 pm

SF17 solves instantly in normal mode:
info depth 4 seldepth 6 multipv 1 score mate 3 nodes 49132 nps 8188666 hashfull 0 tbhits 0 time 6

peter · Post by **peter** » Fri Nov 01, 2024 6:42 pm

Jouni wrote: ↑Thu Oct 31, 2024 8:17 pm SF17 solves instantly in normal mode:
info depth 4 seldepth 6 multipv 1 score mate 3 nodes 49132 nps 8188666 hashfull 0 tbhits 0 time 6

So does Alexandria 7.1 (single thread):

Code: Select all

Alexandria-7.1.0
position fen 8/3P4/8/1p6/kp6/ppp5/brpp1p2/nnqrbK2 w - - 0 0
setoption name Hash value 1024
Set Hash to 1024 MB
TT init complete with 33554429 buckets and 100663287 entries
go infinite
info time: -1 start: 2013187231 stopOpt: 0 stopMax: 0 depth: 256 timeset: 0 nodeset: 0
info score cp -8897 depth 1 seldepth 1 multipv 1 nodes 6 nps 6000 hashfull 0 time 1 pv f1g2
...
info score cp 0 depth 30 seldepth 9 multipv 1 nodes 831 nps 22000 hashfull 0 time 38 pv d7d8q
info score cp 0 depth 31 seldepth 4 multipv 1 nodes 848 nps 22000 hashfull 0 time 39 pv d7d8q
...
info score mate 3 depth 32 seldepth 10 multipv 1 nodes 1144 nps 28000 time 41 pv d7d8r a4a5 d8d6 a5a4 d6a6
info score mate 3 depth 33 seldepth 5 multipv 1 nodes 1243 nps 28000 time 44 pv d7d8r a4a5 d8d6 a5a4 d6a6
info score mate 3 depth 255 seldepth 5 multipv 1 nodes 19139 nps 40000 time 474 pv d7d8r a4a5 d8d6 a5a4 d6a6

And Berserk 13 (also single thread):

Code: Select all

position fen 8/3P4/8/1p6/kp6/ppp5/brpp1p2/nnqrbK2 w - - 0 0
setoption name Hash value 1024
info string set Hash to value 1024 (1073741824 bytes) (100663296 entries)
go infinite
info string time -1 start 2013646031 alloc 0 max 2147483647 depth 200 timeset 0 searchmoves 0
info depth 1 seldepth 1 multipv 1 score cp -1202 wdl 0 0 1000 nodes 11 nps 11000 hashfull 0 tbhits 0 time 1 pv d7d8q
...
info depth 39 seldepth 2 multipv 1 score cp 0 wdl 87 826 87 nodes 5139 nps 54670 hashfull 0 tbhits 0 time 94 pv d7d8q
info depth 40 seldepth 6 multipv 1 score mate 3 wdl 1000 0 0 nodes 5973 nps 54798 hashfull 0 tbhits 0 time 109 pv d7d8r a4a5 d8d6 a5a4 d6a6
info depth 41 seldepth 6 multipv 1 score mate 3 wdl 1000 0 0 nodes 6056 nps 55559 hashfull 0 tbhits 0 time 109 pv d7d8r a4a5 d8d6 a5a4 d6a6
...
info depth 200 seldepth 6 multipv 1 score mate 3 wdl 1000 0 0 nodes 33040 nps 45952 hashfull 0 tbhits 0 time 719 pv d7d8r a4a5 d8d6 a5a4 d6a6

Uri Blass · Post by **Uri Blass** » Fri Nov 01, 2024 7:29 pm

Jouni wrote: ↑Thu Oct 31, 2024 8:57 am Another .

[d]8/3P4/8/1p6/kp6/ppp5/brpp1p2/nnqrbK2 w - - 0 0

Analysis by Obsidian 14.0:
...
1.d8Q
= (0.00) Depth: 121 00:00:00 5406kN
1.d8Q
= (0.00) Depth: 122 00:00:01 6574kN
1.d8Q
= (0.00) Depth: 123 00:00:01 6628kN

Obsidian seems to have a bug and cannot get deeper than depth 123.

It even cannot see the following mate in 2 that is easier with 6 cores and suggest Ra8+ with a draw score.
[d]3R4/8/8/kp6/1p6/ppp5/brpp1p2/nnqrbK2 w - - 1 2

peter · Post by **peter** » Fri Nov 01, 2024 8:05 pm

Uri Blass wrote: ↑Fri Nov 01, 2024 7:29 pm Obsidian seems to have a bug and cannot get deeper than depth 123.

It even cannot see the following mate in 2 that is easier with 6 cores and suggest Ra8+ with a draw score.
[d]3R4/8/8/kp6/1p6/ppp5/brpp1p2/nnqrbK2 w - - 1 2

Suspicious, indeed.
Even with MultiPV=6 no #2 found single threaded:

Code: Select all

Obsidian 14.0 by Gabriele Lombardo
position fen 3R4/8/8/kp6/1p6/ppp5/brpp1p2/nnqrbK2 w - - 1 2
setoption name Hash value 1024
setoption name MultiPV value 6
go infinite
...
info depth 123 multipv 1 score cp 0 nodes 1515693 nps 1531003 hashfull 0 tbhits 0 time 990 pv d8a8 a5b6 a8b8 b6c7 b8f8 c7d7 f8a8 d7e6 a8f8
info depth 123 multipv 2 score cp 0 nodes 1515693 nps 1531003 hashfull 0 tbhits 0 time 990 pv d8b8 a5a6 b8a8 a6b7 a8f8 b7c7 f8a8
info depth 123 multipv 3 score cp 0 nodes 1515693 nps 1527916 hashfull 0 tbhits 0 time 992 pv d8c8 a5b6 c8d8 b6c7 d8f8 c7d7 f8a8 d7e6 a8f8
info depth 123 multipv 4 score cp 0 nodes 1515693 nps 1526377 hashfull 0 tbhits 0 time 993 pv d8f8 a5b6 f8c8 b6a5
info depth 123 multipv 5 score cp 0 nodes 1515693 nps 1524842 hashfull 0 tbhits 0 time 994 pv d8h8 a5a6 h8a8 a6b7 a8f8 b7c7 f8a8
info depth 123 multipv 6 score cp 0 nodes 1515693 nps 1523309 hashfull 0 tbhits 0 time 995 pv d8e8 a5b6 e8c8 b6a5
bestmove d8a8

Not until MultiPV=8 (didn't try 7):

Code: Select all

setoption name MultiPV value 8
go infinite
...
info depth 62 multipv 1 score cp 0 nodes 203683 nps 288094 hashfull 0 tbhits 0 time 707 pv d8g8 a5a6 g8a8 a6b7 a8f8 b7c7 f8a8
info depth 62 multipv 2 score cp 0 nodes 203683 nps 287687 hashfull 0 tbhits 0 time 708 pv d8e8 a5b6 e8c8 b6a5
info depth 62 multipv 3 score cp 0 nodes 203683 nps 287282 hashfull 0 tbhits 0 time 709 pv d8c8 a5b6 c8d8 b6c7 d8f8 c7d7 f8a8 d7e6 a8f8
info depth 62 multipv 4 score cp 0 nodes 203683 nps 286877 hashfull 0 tbhits 0 time 710 pv d8b8 a5a6 b8a8 a6b7 a8f8 b7c7 f8a8
info depth 62 multipv 5 score cp 0 nodes 203683 nps 286473 hashfull 0 tbhits 0 time 711 pv d8h8 a5a6 h8a8 a6b7 a8f8 b7c7 f8a8
info depth 62 multipv 6 score cp 0 nodes 203683 nps 285670 hashfull 0 tbhits 0 time 713 pv d8a8 a5b6 a8d8 b6c7 d8f8 c7d7 f8a8 d7e6 a8f8
info depth 62 multipv 7 score cp 0 nodes 203683 nps 284473 hashfull 0 tbhits 0 time 716 pv d8f8 a5b6 f8c8 b6a5
info depth 62 multipv 8 score cp 0 nodes 203683 nps 284076 hashfull 0 tbhits 0 time 717 pv d8d4 a5b6 d4d8 b6c7 d8f8 c7d7 f8a8 d7e6 a8f8

info depth 63 multipv 1 score mate 2 nodes 210478 nps 290715 hashfull 0 tbhits 0 time 724 pv d8d6 a5a4 d6a6
info depth 63 multipv 2 score cp 0 nodes 210478 nps 290715 hashfull 0 tbhits 0 time 724 pv d8e8 a5b6 e8c8 b6a5
info depth 63 multipv 3 score cp 0 nodes 210478 nps 290314 hashfull 0 tbhits 0 time 725 pv d8c8 a5b6 c8d8 b6c7 d8f8 c7d7 f8a8 d7e6 a8f8
info depth 63 multipv 4 score cp 0 nodes 210478 nps 289914 hashfull 0 tbhits 0 time 726 pv d8b8 a5a6 b8a8 a6b7 a8f8 b7c7 f8a8
info depth 63 multipv 5 score cp 0 nodes 210478 nps 289515 hashfull 0 tbhits 0 time 727 pv d8h8 a5a6 h8a8 a6b7 a8f8 b7c7 f8a8
info depth 63 multipv 6 score cp 0 nodes 210478 nps 288326 hashfull 0 tbhits 0 time 730 pv d8a8 a5b6 a8d8 b6c7 d8f8 c7d7 f8a8 d7e6 a8f8
info depth 63 multipv 7 score cp 0 nodes 210478 nps 287931 hashfull 0 tbhits 0 time 731 pv d8f8 a5b6 f8c8 b6a5
info depth 63 multipv 8 score cp 0 nodes 210478 nps 287538 hashfull 0 tbhits 0 time 732 pv d8g8 a5a6 g8a8 a6b7 a8f8 b7c7 f8a8

gabe_obsidian · Post by **gabe_obsidian** » Fri Nov 01, 2024 10:06 pm

Uri Blass wrote: ↑Fri Nov 01, 2024 7:29 pm
Jouni wrote: ↑Thu Oct 31, 2024 8:57 am Another .

[d]8/3P4/8/1p6/kp6/ppp5/brpp1p2/nnqrbK2 w - - 0 0

Analysis by Obsidian 14.0:
...
1.d8Q
= (0.00) Depth: 121 00:00:00 5406kN
1.d8Q
= (0.00) Depth: 122 00:00:01 6574kN
1.d8Q
= (0.00) Depth: 123 00:00:01 6628kN

Obsidian seems to have a bug and cannot get deeper than depth 123.

It even cannot see the following mate in 2 that is easier with 6 cores and suggest Ra8+ with a draw score.
[d]3R4/8/8/kp6/1p6/ppp5/brpp1p2/nnqrbK2 w - - 1 2

There is no bug, 123 is simply the maximum depth.
Anyway, I always tuned Obsidian towards Elo, not "tactical ability" or whatever.
It can happen that out of thousands of puzzles, it doesn't solve some.

chessica · Post by **chessica** » Sat Nov 02, 2024 10:58 am

Jouni wrote: ↑Thu Oct 31, 2024 8:57 am Another .

[d]8/3P4/8/1p6/kp6/ppp5/brpp1p2/nnqrbK2 w - - 0 0

Analysis by Obsidian 14.0:
...
1.d8Q
= (0.00) Depth: 121 00:00:00 5406kN
1.d8Q
= (0.00) Depth: 122 00:00:01 6574kN
1.d8Q
= (0.00) Depth: 123 00:00:01 6628kN

Code: Select all

FEN: 8/3P4/8/1p6/kp6/ppp5/brpp1p2/nnqrbK2 w - - 0 1

[b][u]Obsidian 14.0:[/u][/b]

---------------------------------------------------------------------------
 35	00:01,893	 908k	480k	-M6	Kf1-g2 f2-f1D+ Kg2xf1 Le1-g3+ Kf1-g2 Td1-g1+ Kg2-h3 d2-d1D d7-d8D Dc1-h6+ Dd8-h4 Dh6xh4+
 35	00:01,893	 908k	480k	-M6	Kf1-e2 f2-f1D+ Ke2xf1 Le1-g3+ Kf1-g2 Td1-g1+ Kg2-h3 d2-d1D d7-d8D Dc1-h6+ Dd8-h4 Dh6xh4+
 35	00:01,893	 908k	480k	-32,77	d7-d8S Ka4-a5 Sd8-b7+ Ka5-b6 Sb7-d6 Kb6-c5 Sd6-e8 Kc5-c6 Se8-f6 Kc6-d6 Sf6-e8+ Kd6-c5 Se8-c7 Kc5-b6 Sc7-d5+ Kb6-a5 Sd5-f4
 35	00:01,893	 908k	480k	-24,48	d7-d8T Ka4-a5 Td8-b8 Ka5-a6 Tb8-c8 Ka6-b7 Tc8-f8 Kb7-c6 Tf8-c8+ Kc6-d6 Tc8-d8+ Kd6-e6 Td8-e8+ Ke6-f7 Te8-h8 Kf7-g6 Th8-f8
 35	00:01,893	 908k	480k	 0,00	d7-d8L
 35	00:01,893	 908k	480k	 0,00	d7-d8D
---------------------------------------------------------------------------
 36	00:02,303	 1.098k	477k	-M6	Kf1-g2 f2-f1D+ Kg2xf1 Le1-g3+ Kf1-g2 Td1-g1+ Kg2-h3 d2-d1D d7-d8D Dc1-h6+ Dd8-h4 Dh6xh4+
 36	00:02,303	 1.098k	477k	-M6	Kf1-e2 f2-f1D+ Ke2xf1 Le1-g3+ Kf1-g2 Td1-g1+ Kg2-h3 d2-d1D d7-d8D Dc1-h6+ Dd8-h4 Dh6xh4+
 36	00:02,303	 1.098k	477k	-29,11	d7-d8S Ka4-a5 Sd8-b7+ Ka5-b6 Sb7-d6 Kb6-c5 Sd6-e4+ Kc5-d5 Se4-f6+ Kd5-e5 Sf6-e8 Ke5-d4 Se8-c7 Kd4-c5 Sc7-e6+ Kc5-d6 Se6-f4 Kd6-e5 Sf4-g6+
 36	00:02,303	 1.098k	477k	 0,00	d7-d8D
 36	00:02,303	 1.098k	477k	 0,00	d7-d8L
 36	00:02,303	 1.098k	477k	+M3	d7-d8T Ka4-a5 Td8-d6 Ka5-a4 Td6-a6+
---------------------------------------------------------------------------

Obsidian 14.0 and three mover

Re: Obsidian 14.0 and three mover

Re: Obsidian 14.0 and three mover

Re: Obsidian 14.0 and three mover

Re: Obsidian 14.0 and three mover

Re: Obsidian 14.0 and three mover

Re: Obsidian 14.0 and three mover