By Lothar Jung
Date 2020-06-12 09:23
Edited 2020-06-12 09:26
Upvotes 1
Heute auf Discord:
Match: SF vs. lc0 nets 1m+1s gauntlet at equal cost hardware
LC0-version: v0.26.0-dev+git.dd46737 built Jun 1 2020
LC0 options: minibatch 40, backend multiplexing, backendOptions (backend=cudnn-fp16,gpu=0),(backend=cudnn-fp16,gpu=1), threads 3, MovesLeftMaxEffect 0.2, MovesLeftThreshold 0, MovesLeftSlope 0.007, MovesLeftConstantFactor 0, MovesLeftScaledFactor 1, MovesLeftQuadraticFactor 0
SF-version: SF-dev 2020.06.01
SF options: 1024mb cache, 32 threads
Hardware: CPU: Ryzen 3950x ($961 CAD) + GPU: 2x RTX 2060 ($997 CAD)
Time control: 1m+1s
Speed: SF ~50mnps LC0 ~20knps (Leela ratio 0.35)
Book: openings-6ply-1000.pgn
Tablebases: none
Adjudication: -draw movenumber=40 movecount=3 score=10, -resign movecount=3 score=600
Software: Cutechess GUI windows build care of zz4032 (pinned post)
Comments: expected outcome, will continue comparing winner against new entrants, using new book
# PLAYER : RATING ERROR POINTS PLAYED (%) CFS(%) W D L D(%)
1 lc0-3010mlh : 9.4 7.9 578.5 1078 53.7 70 164 829 85 76.9
2 lc0-3010 : 5.8 7.9 573.0 1078 53.2 74 154 838 86 77.7
3 lc0-63688 : 1.5 7.6 566.5 1078 52.6 100 154 825 99 76.5
4 sf200601 : -16.7 4.2 1516.0 3234 46.9 --- 270 2492 472 77.1
White advantage = 40.00 +/- 2.92
Draw rate (equal opponents) = 80.14 % +/- 0.75
gauntlet1.7z
2.33 MB
Match: lc0.25.63730 vs Stockfish 11 4CPU - 100 rapid games
LC0-version: v25.1 cuda
LC0-options: --backend=multiplexing --cpuct=2.147 --cpuct-factor=2.815 --cpuct-base=18368 --fpu-value=0.443 --policy-softmax-temp=1.607
Time control: 15min + 5s (CCRL 40/15)
Hardware: CPU i7-8700 4cores vs 1070Ti GPU
Book: Custom short lines played from both sides (!sheet4 for opening list)
Tablebase: 6 piece syzygy (DTZ + WDL) for both engines
Software: Arena
Speed: Leela ratio ~0.85; Lc0 npm~200K, SF npm~200M (based on 24x256 nets, T60 is slower)
Context: Same MLH params as prev; !sheet4 for test history
MLH2 params:
--moves-left-max-effect=0.15 --moves-left-threshold=0.03 --moves-left-slope=0.015 moves-left-quadratic-factor=1.0 --moves-left-scaled-factor=0.0 --moves-left-constant-factor=0.0
Ordo w/ comparison to previous T60 scores:
# PLAYER : RATING ERROR POINTS PLAYED (%) CFS(%) W D L D(%) (Against SF)
1 lc0.25.1.63677_MLH2 : 3529.6 30.0 54.5 100 55 98 15 79 6 79
2 lc0.25.63558 : 3525.7 29.6 54.0 100 54 97 14 80 6 80
3 lc0.25.63450 : 3511.2 29.7 52.0 100 52 83 12 80 8 80
+ 4 lc0.25.1.63730_MLH2 : 3500.6 32.0 50.5 100 51 59 12 77 11 77
5 lc0.25.1.63608_MLH : 3500.6 33.8 50.5 100 51 58 13 75 12 75
- 6 Stockfish_11_x64_bmi2 : 3497.0 ---- 52.5 100 53 --- 15 75 10 75 (against 62091)
7 lc0.24.384x30-t60-3010 : 3486.4 31.0 48.5 100 49 --- 9 79 12 79
lc0.25.1.63730_MLH2 - Stockfish_11_x64_bmi2 : 50.5/100 12-11-77 (===0=1=====1=======0==1==01===1==0===0========1==1=01===============10=000========1=========1===10==) 51% -> 3501 ordo score
[20:34]
Addendum - Avg game lengths:
Net MovesGame MovesWin MovesLoss MovesDraw
63608_MLH 68 52 75 71
63677_MLH2 82 54 34 88
+ 63730_MLH2 82 69 60 87
63558 80 76 56 83
63223 88 89 53 91
384x30-t60-3010 80 91 56 82
63450 88 92 63 89
63375 92 99 69 93
Zwar alles nicht vergleichbar, aber interessant.
Besonderes Hardwarekosten RTX/Ryzen 3950X
Lothar