For a match started like:
/home/vondele/chess/vondele/workspace/scratch/abcdefgh/testing/fastchess/fastchess \
-rounds 10 \
-games 2 \
-repeat \
-srand 42 \
-concurrency 280 \
--force-concurrency \
-openings file=/home/vondele/chess/vondele/workspace/data/UHO_Lichess_4852_v1.epd \
format=epd \
order=random \
-ratinginterval 280 \
-report penta=true \
-pgnout file=match.pgn \
-engine name=reference \
cmd=/home/vondele/chess/vondele/workspace/scratch/abcdefgh/testing/reference/Stockfish/src/stockfish \
-engine name=step_97fb32d68c61_nn-0861a1e95b35.nnue \
cmd=/home/vondele/chess/vondele/workspace/scratch/abcdefgh/testing/testing/Stockfish/src/stockfish \
option.EvalFile=/home/vondele/chess/vondele/workspace/scratch/97fb32d68c61/run/lightning_logs/version_0/checkpoints/nn-0861a1e95b35.nnue \
-each proto=uci \
option.Threads=1 \
option.Hash=16 \
tc=10+0.1
(most importantly -rounds 10 -games 2 -repeat and two engines)
The config file contains
"stats": {
"step_97fb32d68c61_nn-0861a1e95b35.nnue vs reference": {
"wins": 0,
"losses": 8,
"draws": 0,
"penta_WW": 0,
"penta_WD": 0,
"penta_WL": 0,
"penta_DD": 0,
"penta_LD": 0,
"penta_LL": 4
},
"reference vs step_97fb32d68c61_nn-0861a1e95b35.nnue": {
"wins": 11,
"losses": 0,
"draws": 1,
"penta_WW": 5,
"penta_WD": 1,
"penta_WL": 0,
"penta_DD": 0,
"penta_LD": 0,
"penta_LL": 0
}
}
which seems to suggest one engine played 12 games with white and the other only 8. That would be wrong given the options, and is also different from what I see in the pgn:
$ grep "\[White" match.pgn | sort | uniq -c
10 [White "reference"]
10 [White "step_97fb32d68c61_nn-0861a1e95b35.nnue"]
For a match started like:
(most importantly
-rounds 10 -games 2 -repeatand two engines)The config file contains
which seems to suggest one engine played 12 games with white and the other only 8. That would be wrong given the options, and is also different from what I see in the pgn: