This may be the wrong section to ask so I apologize in advance if it is but when does a slight modification to a chess engine such as Stockfish 2.2 become a really significant elo improvement. I recently made a slight modification and got these results.
Games Completed = 1500 of 1500 (Avg game length = 28.678 sec)
Settings = RR/128MB/14000ms in 150 moves/M 1000000cp for 1000 moves, D 200 moves/EPD:C:\-->your openings.epd(1)
Time = 19113 sec elapsed, 0 sec remaining
1. Experimental Engine 779.5/1500 413-354-733 (L: m=352 t=2 i=0 a=0) (D: r=575 i=127 f=24 s=6 a=1) (tpm=251.2 d=12.2 nps=1127688)
2. Stockfish-222-Factory 720.5/1500 354-413-733 (L: m=413 t=0 i=0 a=0) (D: r=575 i=127 f=24 s=6 a=1) (tpm=149.7 d=12.2 nps=1122634)
As you can see, my experimental engine was slightly better 413 wins vs 354 losses but because this is over 1500 games, does this still remain a significant gain or is it in the +5/-5 error margin?
Games Completed = 1500 of 1500 (Avg game length = 28.678 sec)
Settings = RR/128MB/14000ms in 150 moves/M 1000000cp for 1000 moves, D 200 moves/EPD:C:\-->your openings.epd(1)
Time = 19113 sec elapsed, 0 sec remaining
1. Experimental Engine 779.5/1500 413-354-733 (L: m=352 t=2 i=0 a=0) (D: r=575 i=127 f=24 s=6 a=1) (tpm=251.2 d=12.2 nps=1127688)
2. Stockfish-222-Factory 720.5/1500 354-413-733 (L: m=413 t=0 i=0 a=0) (D: r=575 i=127 f=24 s=6 a=1) (tpm=149.7 d=12.2 nps=1122634)
As you can see, my experimental engine was slightly better 413 wins vs 354 losses but because this is over 1500 games, does this still remain a significant gain or is it in the +5/-5 error margin?