Разные матчи и турниры движков

[Владимир001]

PGN - http://kirill-kryukov.com/chess/disc...e.php?id=50176




PGN - http://kirill-kryukov.com/chess/disc...e.php?id=50175

[Владимир001]

PGN - http://kirill-kryukov.com/chess/disc...e.php?id=50177

Nu-chess 1.0 64-bit 4CPU 2990

[Владимир001]

PGN - http://kirill-kryukov.com/chess/disc...e.php?id=50178

CCRL 40/15 Rating List - Custom engine selection
1351949 games played by 3032 programs, run by 25 testers
Ponder off, General books (up to 12 moves), 3-4-5 piece EGTB
Time control: Equivalent to 40 moves in 15 minutes on an Intel i7-4770k.
Computed on February 26, 2022 with Bayeselo based on 1'351'949 games
Tested by CCRL team, 2005-2022, http://ccrl.chessdom.com/ccrl/4040/

Rank                Engine                  Elo  +    -  Score  AvOp  Games
1 Mantissa 3.3.0 64-bit                  3096  +23  -23  47.6%  +13.0  571
  Mantissa 3.0.0 64-bit                  2999  +28  -28  57.1%  -75.6  518

[Владимир001]

PGN - http://kirill-kryukov.com/chess/disc...e.php?id=50179

CCRL 40/15 Rating List - Custom engine selection
1351949 games played by 3032 programs, run by 25 testers
Ponder off, General books (up to 12 moves), 3-4-5 piece EGTB
Time control: Equivalent to 40 moves in 15 minutes on an Intel i7-4770k.
Computed on February 26, 2022 with Bayeselo based on 1'351'949 games
Tested by CCRL team, 2005-2022, http://ccrl.chessdom.com/ccrl/4040/

Rank                Engine                  Elo  +    -  Score  AvOp  Games
1 Protej 0.6.2 64-bit                    2845  +25  -25  52.0%  -13.3  530
  Protej 0.6.1 64-bit                    2717  +26  -25  59.3%  -63.9  515
  Protej 0.5.8c                          2258  +19  -19  48.5%  +11.1  1017
  Protej 0.5.7                            2056  +25  -25  45.6%  +31.8  575

[Владимир001]

Тесты от Skynet:

Tablet - Lenovo android 4.4.2 - 32 bit - Quad Core (4x Cortex A7) - 1GB RAM - 1,3 GHz
GUI=Cfa - Hash 64 mb - 4 cores per engine - TC: 2min+1sec - Book=8moves.pgn (100 pos) - Resign 600 cp
ELO rating has been set to 3498 (Houdini 6.03 with 4 cores and TC: more then 2 min). Engines armv7.

# PLAYER            :  RATING  ERROR  PLAYED   (%)    W    D    L  D(%)  CFS(%)
1 Fire 8.NN.MC.3    :    3616     18     400  58.4  125  217   58  54.3      64
2 Berserk 8.5.1     :    3610     16     400  57.3  127  204   69  51.0     100
3 Houdini 6.03      :    3498     17     400  34.4   59  157  184  39.3     ---

Games https://pixeldrain.com/u/rDoSR6B8

[Владимир001]

Match: T79 comparsion
LC0-version: PR1699
LC0 options: Network Id: 790107 vs 760231
mbs=208 as in that mbs the two net type has equal speed
Hardware: 1A100
Time control: 800npm
Book: book-ply8-unifen-Q-0.0-0.25.pgn
Tablebases: 5-man syzygy
Adjudication: TB only adjudication
Software: Cutechess-cli
ordo: ./ordo-win64.exe -Q -N 0 -D -a 0 -A "T79_107" -W -n4 -s500 -U "0,1,2,3,4,5,6,7,8,9,10" -N 1 -j h2h.txt -C matrix.csv -p t790107vst760231_800.pgn
Comment: testing new net lr run vs t76 that performed strongest early, same lr stage

   # PLAYER     :  RATING  ERROR  POINTS  PLAYED   (%)  CFS(%)     W     D     L  D(%)
   1 T79_107    :     0.0   ----  2741.5    5000  54.8     100  1469  2545   986  50.9
   2 T76_231    :   -34.1    6.5  2258.5    5000  45.2     ---   986  2545  1469  50.9

White advantage = 18.06 +/- 3.52
Draw rate (equal opponents) = 51.53 % +/- 0.75

https://pixeldrain.com/u/2S1bPy9v

[Владимир001]

Match: lc0.net.781534 vs lc0.net.781561
LC0-version: lc0-v0.28.2, --threads=1 --backend=cuda-fp16, (default settings)
Hardware: RTX 2060
Software: cutechess-cli
Time control: 2s+0.1s
Book: Sufi_12 to 21 (500 openings), sequential, color reversed
Tablebases: Syzygy 5-men
Adjudication: -draw movenumber=50 movecount=5 score=8 -resign movecount=5 score=1000
Comment: The next match (final) is lc0.net.781554 vs lc0.net.781534

# PLAYER         : RATING  ERROR  POINTS  PLAYED    W    L    D  D(%)  CFS(%)
1 lc0.net.781534 :    0.0   ----   505.0    1000  226  216  558    56      71
2 lc0.net.781561 :   -3.7   13.5   495.0    1000  216  226  558    56     ---

White advantage = 84.31 +/- 7.34
Draw rate (equal opponents) = 59.90 % +/- 1.67

[Владимир001]

Fischer-Kasparov match simulation

https://www.chess2u.com/t18017-fisch...ulation#113825

[Владимир001]

Match: lc0.net.781554 vs lc0.net.781534 (final)
LC0-version: lc0-v0.28.2, --threads=1 --backend=cuda-fp16, (default settings)
Hardware: RTX 2060
Software: cutechess-cli
Time control: 2s+0.1s
Book: Sufi_12 to 21 (500 openings), sequential, color reversed
Tablebases: Syzygy 5-men
Adjudication: -draw movenumber=50 movecount=5 score=8 -resign movecount=5 score=1000
Comment: lc0.net.781554 won the final against lc0.net.781534 and won my small championship.
The participants were lc0.net.781554, lc0.net.781534, lc0.net.781561 and lc0.net.781474.
lc0.net.781561 and lc0.net.781474 were already eliminated in the both semifinals (see test-results).

# PLAYER         : RATING  ERROR  POINTS  PLAYED    W    L    D  D(%)  CFS(%)
1 lc0.net.781554 :    0.0   ----   518.0    1000  235  199  566    57      98
2 lc0.net.781534 :  -13.4   13.3   482.0    1000  199  235  566    57     ---

White advantage = 85.91 +/- 7.33
Draw rate (equal opponents) = 61.13 % +/- 1.67

[Владимир001]

Comment: Double gauntlet: 2 Stockfish vs. 2 Lc0 nets
Lc0: 0.28.2 cuda-fp16 mult_stream=true MoveOverheadsMs=0
Stockfish: Threads=16 Hash=2048 Move Overhead=0 Sy.50MR=false
Hardware: Ryzen 7 2700X 8c16t | RTX 2060
Time controls: Stockfish G/2"+.02" Lc0 G/16"+1.6"
Book: UHO_XXL_+0.80_+1.09.pgn , random rev., new per game-pair
Tablebases: Syzygy-6
Adjudication: Draw: Move 35, count 2, 0.12; Resign: count 2, score 7.00, two-sided
Software: cutechess-cli timemargin=200, Ordo 1.2.6 -s 4000 -n 12
Note: Stockfish current and at mg net date vs.
pre-T78 mg-40b net and net now in CCC 17 Rapid Main.

 # PLAYER                        :  RATING  ERROR  POINTS  PLAYED   (%)  CFS(%)    W    D    L  D(%)
 1 Stockfish_21112912_@mg        :    49.2   18.7   549.0    1000    55      80  304  490  206    49
 2 Stockfish_22022508            :    39.9   19.2   538.0    1000    54      99  296  484  220    48
+3 Lc0-0.28.2_781561_C17rm       :    15.3   21.8   465.5    1000    47      92  225  481  294    48
 4 Lc0-0.28.2-mg-40b-swa-2000000 :     0.0   ----   447.5    1000    45     ---  201  493  306    49
 
 Head-to-head (%):
        Sfmg    Sf   LcC17  Lc-mg
Sfmg                  53.4   56.4
Sf                    53.5   54.1
LcC17   46.6   46.5
Lc-mg   43.6   45.9

(Ratings approximate - 1k games each. Stockfish decline from LTC tune?)

https://pixeldrain.com/u/9CJWxDU5

[Владимир001]

ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative

http://talkchess.com/forum3/viewtopi...art=40#p921960

http://talkchess.com/forum3/viewtopi...art=40#p921975

[Владимир001]

Match: lc0.net.69626 (609626) vs lc0.net.611062
LC0-version: lc0-v0.28.2, --threads=1 --backend=cuda-fp16, (default settings)
Hardware: RTX 2060
Software: cutechess-cli
Time control: 2sec./game + 0.2sec./move
Book: Sufi_17-18-19-20-21 (250 openings), sequential, color reversed
Tablebases: Syzygy 5-men
Adjudication: -draw movenumber=50 movecount=5 score=8 -resign movecount=5 score=1000
Comment:
(TCEC_Sufi-21: Stockfish 14_202107131735 vs LCZero 0.28-dev+_69626 56-44).
Stockfish has improved by about 30 ELO since SUFI_21.
Would lc0.net.611062 be strong enough for the S22 TCEC Premier Division?

# PLAYER         : RATING  ERROR  POINTS  PLAYED    W    L    D  D(%)  CFS(%)
1 lc0.net.611062 :   16.9   19.2   261.0     500  114   92  294    59      96
2 lc0.net.69626  :    0.0   ----   239.0     500   92  114  294    59     ---

White advantage = 105.65 +/- 9.46
Draw rate (equal opponents) = 67.12 % +/- 2.52

[Владимир001]

92ND AMATEUR SERIES (Division 2)

Intel i7 Quad
ChessGUI
256mb hash each where possible
3-4-5 piece tablebases
Ponder off
SALC-12moves.cgb book (limited to 8 moves depth)
40 moves in 16 minutes repeating (adapted for the CCRL)
4 cycles 44 rounds
All engines 1CPU (64-bit where available)

Цитата:

The top engine will gain automatic promotion and the bottom engine will face automatic relegation. Any other changes of division will be at my discretion and may not necessarily involve playoff matches.

Final Standings

29.0 - Tucano 10.00 64-bit
27.0 - Booot 6.5 64-bit
25.0 - Lc0 0.28.2 w752187 64-bit
24.0 - Minic 3.18 64-bit
23.5 - RofChade 2.3 64-bit
23.0 - Rebel 14.1 64-bit
22.0 - Zahak 9.0 64-bit
21.5 - Wasp 5.20 64-bit
18.0 - Clover 3.0 64-bit
18.0 - Weiss 2.0 64-bit
17.0 - Combusken 2.0.0 64-bit
16.0 - Marvin 5.2.0 64-bit


The complete tournament pgn (zipped) can be downloaded here:
http://kirill-kryukov.com/chess/disc...e.php?id=50186

[Владимир001]

92ND AMATEUR SERIES (Division 3)

Intel i7 Quad
ChessGUI
256mb hash each
3-4-5 piece tablebases
Ponder off
SALC-12moves.cgb book (limited to 8 moves depth)
40 moves in 16 minutes repeating (adapted for the CCRL)
4 cycles 44 rounds
All engines 64-bit 1CPU (where available)

Цитата:

The top engine will gain automatic promotion and the bottom engine will gain automatic relegation. Any other changes of division will be at my discretion and may not necessarily involve playoff matches.

Final Standings

26.0 - DanaSah 9.00 64-bit
26.0 - Dark Toga 1.1 64-bit
24.5 - Halogen 10 64-bit
24.0 - Roc 1.8 64-bit
23.0 - LG Evolution NN 3.15 64-bit
23.0 - Velvet 3.2.0 64-bit
22.5 - Stash 32.0 64-bit
21.5 - Black Marlin 4.0 64-bit
21.0 - Winter 0.9 64-bit
19.0 - Demolito 2021-10-04 64-bit
18.5 - Counter 4.0 64-bit
15.0 - Mantissa 3.3.0 64-bit


The complete tournament pgn (zipped) can be downloaded here:
http://kirill-kryukov.com/chess/disc...e.php?id=50187

[Владимир001]

Stockfish 130222 wins 1 League CEDR edition 03, 2022.02.18 - 2022.03.03

Stockfish 130222 wins 1 League CEDR edition 03, 2022.02.18 - 2022.03.03
Relegation to league 2: Ethereal 12.75 and RubiChess 2021.
15'+10" , Dell G15 SE, i7-11800H RAM 16,0 GB ,GUI Arena
240 tournament games , All games CEDR 17.349 games , Table: Scid vs PC

[Владимир001]

Hardware: AMD Ryzen 9 5900X CPU @ 3.70GHz, 12 cores, 64GB, gpu RTX 3080, 10GB
Gauntlet: SF-dev vs. LC0 512x40b nets
Network IDs: 780001, 780048, 780856, 780900, 781032, 781128, 781474, 781589
LC0-version: v0.28.2
LC0-options: default
SF-version: SF-dev 22011322
SF-options: 8 Threads, 1024MB Hash
Software: Cutechess-CLI
Time control: LC0: 1,500 nodes/move, SF-dev: 500,000 nodes/move
Book: Chad's 8-ply book, book-ply8-unifen-Q-0.0-0.25.pgn and book-ply8-unifen-Q-0.25-0.40.pgn (10K openings in each book), order=sequential
Tablebases: 6-men
Adjudication: -draw movenumber=50 movecount=5 score=8 -resign movecount=5 score=1000
Ordo: -a 0 -A "Stockfish.dev" -W -D -U "0,1,2,3,4,5,6,7,8,9,10" -s 2000 -n 1
Comment: The aim of the test is to order T78 nets. The order is considered sure with 99% CFS (a bit more than 3.5 Elo). CFS matrix and detailed results in 7z file.

   # PLAYER            :  RATING  ERROR    POINTS  PLAYED   (%)  CFS(%)       W       D      L  D(%)
   1 Stockfish.dev     :     0.0
   2 lc0.net.781474    :   -20.2    1.7   28353.5   60000    47      85   10237   36233  13530    60
+  3 lc0.net.781589    :   -21.5    1.8   28244.5   60000    47      55    9993   36503  13504    61
   4 lc0.net.781128    :   -21.7    1.8   28232.0   60000    47      99   10177   36110  13713    60
   5 lc0.net.781032    :   -24.5    1.7   28000.5   60000    47     100    9938   36125  13937    60
   6 lc0.net.780856    :   -30.3    1.8   27531.0   60000    46      57    9793   35476  14731    59
   7 lc0.net.780900    :   -30.5    1.7   27513.5   60000    46      93    9957   35113  14930    59
-  8 lc0.net.780048    :   -32.4    1.8   27360.5   60000    46     100    9737   35247  15016    59
   9 lc0.net.780001    :   -36.3    1.8   27049.5   60000    45     ---    9856   34387  15757    57

White advantage = 77.02 +/- 0.32
Draw rate (equal opponents) = 63.83 % +/- 0.08

PGN: https://pixeldrain.com/u/wp4p6gRZ

[Владимир001]

Comment: Time control relative strength scaling experiment.
Lc0: 0.28.2 cuda-fp16 mult_stream=true MoveOverheadsMs=0
Stockfish: Threads=16 Hash=2048 Move Overhead=0 Sy.50MR=false
Hardware: Ryzen 7 2700X 8c16t | RTX 2060
Time controls listed, Stockfish always 1/10th of Lc0
Book: UHO_XXL_+0.80_+1.09.pgn, 1st 100 per gauntlet, same per round (both Lc0's)
Tablebases: Syzygy-6
Adjudication: Draw: Move 35, count 2, 0.12; Resign: count 2, score 7.00, two-sided
Software: cutechess-cli policy=round timemargin=200, Ordo 1.2.6 -a 0 -s 4000 -n 12
Gauntlets: Time controls G/secs+incr/secs
Stockfish current version vs. Lc0 CCC 16 Rapid Final net and CCC 17 Rapid Main net.

#                            :  RATING  ERROR  POINTS  PLAYED   (%)  CFS(%)    W    D    L  D(%)
   Sf 1+.01 , Lc0 10+.1
1 Lc0-0.28.2-610536_C16Rf    :    55.7   27.1   122.5     200    61     100   78   89   33    44
2 Lc0-0.28.2_781561_C17Rm    :   -17.5   26.4   105.0     200    52      87   57   96   47    48
3 Stockfish_22030117         :   -38.1   17.0   172.5     400    43     ---   80  185  135    46

   Sf 2+.02 , Lc0 20+.2
1 Lc0-0.28.2-610536_C16Rf    :    50.9   26.2   120.0     200    60     100   75   90   35    45
2 Lc0-0.28.2_781561_C17Rm    :   -17.0   25.5   104.0     200    52      83   52  104   44    52
3 Stockfish_22030117         :   -33.8   16.5   176.0     400    44     ---   79  194  127    48

   Sf 4+.04 , Lc0 40+.2
1 Lc0-0.28.2-610536_C16Rf    :     9.2   23.8   103.5     200    52      71   52  103   45    52
2 Lc0-0.28.2_781561_C17Rm    :    -3.5   23.6   100.5     200    50      55   45  111   44    56
3 Stockfish_22030117         :    -5.6   15.0   196.0     400    49     ---   89  214   97    54

   Sf 8+.08 , Lc0 80+.8
1 Stockfish_22030117         :    14.3   15.5   209.5     400    52      89  109  201   90    50
2 Lc0-0.28.2_781561_C17Rm    :    -6.0   24.3    95.5     200    48      54   44  103   53    52
3 Lc0-0.28.2-610536_C16Rf    :    -8.3   24.5    95.0     200    48     ---   46   98   56    49

https://pixeldrain.com/u/mxToR2ry

[Владимир001]

Fischer-Kasparov match simulation

https://www.chess2u.com/t18017-fisch...ulation#113884

[Владимир001]

Match: lc0.net.781561 vs lc0.net.781739
LC0-version: lc0-v0.28.2, --threads=1 --backend=cuda-fp16, (default settings)
Hardware: RTX 2060
Software: cutechess-cli
Time control: 2sec/game + 0.1sec/move
Book: Sufi_12 to 21 (500 openings), sequential, color reversed
Tablebases: Syzygy 5-men
Adjudication: -draw movenumber=50 movecount=5 score=8 -resign movecount=5 score=1000
Comment:

# PLAYER         : RATING  ERROR  POINTS  PLAYED    W    L    D  D(%)  CFS(%)
1 lc0.net.781561 :    0.0   ----   508.5    1000  236  219  545    55      82
2 lc0.net.781739 :   -6.3   13.6   491.5    1000  219  236  545    55     ---

White advantage = 83.22 +/- 7.49
Draw rate (equal opponents) = 58.26 % +/- 1.68

[Владимир001]

Stockfish 170222 wins CEDR - Fritz 15 Test New Engines, 2022.02.24 - 2022.03.01

15'+10", Lenovo Z710 i7-4710MG 8GB RAM, GUI Fritz 15
182 tournament games, All games CEDR 17.349 games Table: Scid vs PC

[Владимир001]

PGN - http://kirill-kryukov.com/chess/disc...e.php?id=50195


Anka 0.6.3 64-bit 2487

[Владимир001]

PGN - http://kirill-kryukov.com/chess/disc...e.php?id=50196

Развернуть для просмотра

CCRL 40/15 Rating List - Custom engine selection
1351949 games played by 3032 programs, run by 25 testers
Ponder off, General books (up to 12 moves), 3-4-5 piece EGTB
Time control: Equivalent to 40 moves in 15 minutes on an Intel i7-4770k.
Computed on February 26, 2022 with Bayeselo based on 1'351'949 games
Tested by CCRL team, 2005-2022, http://ccrl.chessdom.com/ccrl/4040/

Rank                Engine                  Elo  +    -  Score  AvOp  Games
1 Cheese 2.2 64-bit                      2833  +18  -18  48.0%  +12.8  1056
  Cheese 3.0.0 64-bit                    2813  +22  -22  48.7%  +8.4  662
  Cheese 2.1 64-bit                      2780  +19  -19  48.3%  +10.7  928
  Cheese 2.0 64-bit                      2763  +13  -13  47.4%  +18.0  2017
  Cheese 1.9 64-bit                      2704  +15  -15  48.9%  +8.5  1382
  Cheese 1.8 64-bit                      2683  +20  -20  51.2%  -8.4  858
  Cheese 1.7 64-bit                      2678  +21  -22  51.9%  -14.2  699
  Cheese 1.6.1 64-bit                    2545  +22  -22  46.9%  +19.7  653
  Cheese 1.5 64-bit                      2519  +27  -27  51.0%  -7.3  458
  Cheese 1.4 64-bit                      2471  +20  -20  51.8%  -14.6  851
  Cheese 1.4 32-bit                      2441  +32  -32  50.6%  -4.7  324
  Cheese 1.3b 32-bit                      2314  +34  -34  48.8%  +7.0  303
  Cheese 1.3                              2310  +17  -17  48.5%  +10.5  1223
  Cheese 1.1b                            2309  +33  -33  51.0%  -6.2  312
  Cheese 1.2                              2262  +34  -34  51.0%  -6.3  300

[свернуть]

[Владимир001]

PGN - http://kirill-kryukov.com/chess/disc...e.php?id=50197

Развернуть для просмотра

CCRL 40/15 Rating List - Custom engine selection
1351949 games played by 3032 programs, run by 25 testers
Ponder off, General books (up to 12 moves), 3-4-5 piece EGTB
Time control: Equivalent to 40 moves in 15 minutes on an Intel i7-4770k.
Computed on February 26, 2022 with Bayeselo based on 1'351'949 games
Tested by CCRL team, 2005-2022, http://ccrl.chessdom.com/ccrl/4040/

Rank                Engine                  Elo  +    -  Score  AvOp  Games
1 Prophet 4.1 64-bit                      2381  +29  -28  58.6%  -61.1  432
  Prophet 2.0 64-bit                      2178  +27  -27  49.3%  +4.6  484
  Prophet 2.0 32-bit                      2139  +19  -19  49.3%  +4.3  1025
  Prophet3 20180811 64-bit                2130  +18  -18  52.3%  -17.5  1262
  Prophet3 20170909 64-bit                2101  +29  -29  48.4%  +17.3  425

[свернуть]

[Владимир001]

PGN - http://kirill-kryukov.com/chess/disc...e.php?id=50198


Tantabus 1.0.2 64-bit 2367

[Владимир001]

PGN - http://kirill-kryukov.com/chess/disc...e.php?id=50199

Развернуть для просмотра

CCRL 40/15 Rating List - Custom engine selection
1351949 games played by 3032 programs, run by 25 testers
Ponder off, General books (up to 12 moves), 3-4-5 piece EGTB
Time control: Equivalent to 40 moves in 15 minutes on an Intel i7-4770k.
Computed on February 26, 2022 with Bayeselo based on 1'351'949 games
Tested by CCRL team, 2005-2022, http://ccrl.chessdom.com/ccrl/4040/

Rank                Engine                  Elo  +    -  Score  AvOp  Games
1 Nalwald 15 64-bit                      2899  +28  -29  44.6%  +35.0  400
  Nalwald 14.0 64-bit                    2826  +21  -21  48.7%  +7.2  746
  Nalwald 1.9 64-bit                      2609  +29  -29  51.6%  -7.6  416

[свернуть]

[Владимир001]

Match: 780001 vs 781734
LC0 version: v0.28
LC0 options: --smart-pruning-factor=1 --minibatch-size=64 --cpuct=1.92 --cpuct-factor=3.973 --cpuct-base=45669 --fpu-value=0.32 --policy-softmax-temp=1.16
Hardware: EPYC 74F3 + A100
Time control: 1k npm
Book: book-ply8-unifen-Q-0.25-0.40.pgn
Tablebases: 6-men tablebase
Adjudication: -resign movecount=2 score=500 twosided=true -draw movenumber=50 movecount=20 score=10
Software: cutechess-cli
Comment: Used updates params from tune

   # PLAYER        :  RATING  ERROR  POINTS  PLAYED   (%)  CFS(%)     W     D     L  D(%)
   1 Lc0.781734    :    19.8    4.1  7340.5   14000    52     100  3827  7027  3146    50
   2 Lc0.780001    :     0.0   ----  6659.5   14000    48     ---  3146  7027  3827    50

White advantage = 136.06 +/- 2.05
Draw rate (equal opponents) = 60.21 % +/- 0.51

https://pixeldrain.com/u/7PjnwSQ9

[Владимир001]

Match: 780001 vs 781739
LC0 version: v0.28
LC0 options: --smart-pruning-factor=1 --minibatch-size=64 --cpuct=1.92 --cpuct-factor=3.973 --cpuct-base=45669 --fpu-value=0.32 --policy-softmax-temp=1.16
Hardware: EPYC 74F3 + A100
Time control: 1k npm
Book: book-ply8-unifen-Q-0.25-0.40.pgn
Tablebases: 6-men tablebase
Adjudication: -resign movecount=2 score=500 twosided=true -draw movenumber=50 movecount=20 score=10
Software: cutechess-cli
Comment:

   # PLAYER        :  RATING  ERROR  POINTS  PLAYED   (%)  CFS(%)     W     D     L  D(%)
   1 Lc0.781739    :    20.6    4.1  7353.5   14000    53     100  3853  7001  3146    50
   2 Lc0.780001    :     0.0   ----  6646.5   14000    47     ---  3146  7001  3853    50

White advantage = 138.36 +/- 2.05
Draw rate (equal opponents) = 60.38 % +/- 0.52

https://pixeldrain.com/u/PezPvRws

[Владимир001]

Fisherov 0.98i wins CEDR - Fritz 17 New Engines Test, 2022.02.27 - 2022.03.03

[Владимир001]

Match: lc0.net.781561 vs lc0.net.781778
LC0-version: lc0-v0.28.2, --threads=1 --backend=cuda-fp16, (default settings)
Hardware: RTX 2060
Software: cutechess-cli
Time control: 2sec/game + 0.1sec/move
Book: Sufi_12 to 21 (500 openings), sequential, color reversed
Tablebases: Syzygy 5-men
Adjudication: -draw movenumber=50 movecount=5 score=8 -resign movecount=5 score=1000
Comment:

# PLAYER         : RATING  ERROR  POINTS  PLAYED    W    L     D  D(%)  CFS(%)
+ 1 lc0.net.781778 :   16.3   15.1   522.0    1000  259  215   526    53      98
- 2 lc0.net.781561 :    0.0   ----   986.5    2000  451  478  1071    54      81
  3 lc0.net.781739 :   -6.3   14.1   491.5    1000  219  236   545    55     ---

White advantage = 84.59 +/- 5.30
Draw rate (equal opponents) = 57.34 % +/- 1.18

[Владимир001]

tests in glb style (sf with 500k nodes, leela with 1500 nodes). Currently (far from finished) it reads:

 # PLAYER        :  RATING  ERROR  POINTS  PLAYED   (%)  CFS(%)     W     D     L  D(%)
   1 stockfish     :     0.0   ----  1705.0    3161    54     100  1028  1354   779    43
   2 Lc0.781561    :   -35.1    9.3  1456.0    3161    46     ---   779  1354  1028    43

   # PLAYER        :  RATING  ERROR  POINTS  PLAYED   (%)  CFS(%)    W     D    L  D(%)
   1 stockfish     :     0.0   ----  1548.0    2948    53     100  897  1302  749    44
   2 Lc0.781739    :   -21.8    9.6  1400.0    2948    47     ---  749  1302  897    44

https://pixeldrain.com/u/71C87DDG