CorChess is a clone
of Stockfish maintained for better performance on long time controls,
trying at least partially to fill the gap between regular tests and demands
of correspondence players.
The first idea was
to improve the following scalability
of Stockfish 8
and from CorChess 2 I follow the idea from the following picture,
obtained by mathematical model for search tree efficiency.
Please contact
me (ivan.ivec@gmail.com) about your experiences with CorChess,
or in the case that you want to be on the CorChess mailing list.
I use strong 2moves book.
| Regression tests between CorChess 2 and Stockfish 220821 (Ryzen 7 3700X - 1 core) | |||||
| Hash | Time control in seconds | #games | Result | Elo gain (± 2) | Draw rate |
| 1 MB | 0.5+0.005 | 80000 | 12699-34076-33225 | -95.15 | 41.5% |
| 8 MB | 5+0.05 | 25000 | 1293-3596-20111 | -32.1 | 80.4% |
| 64 MB | 50+0.5 | 8000 | 98-244-7658 | -6.34 | 95.7% |
| 512 MB | 500+5 | 2500 | 10-18-2472 | -1.11 | 98.9% |
| 900 MB | 5000+50 | 800 | 0-0-800 | 0.0 | 100% |
| Regression tests between CorChess 3 and Stockfish (Ryzen 7 3700X - 1 core) | |||||
| Hash | Time control in seconds | #games | Result | SF version | Elo gain / Draw rate |
| 900 MB | 1200+12 | 1080 | 3-3-1074 | 061221 | 0.0 / 99.4% |
| 128 MB | 100+1* | 2000 | 17-24-1959 | 170922 | -1.22 / 98% |
| Watch the longest games LIVE.
* Slight regression is expected and normal at 100+1 time control. This will be our checkpoint. |
|||||
Motto: Alone I can nothing, with God I can everything (St. Vincent Pallotti).
Conclusion: There is no strict proof, but there is a hope that CorChess is better on long time controls.