mirror of
https://github.com/HChaZZY/Stockfish.git
synced 2025-12-06 10:53:50 +08:00
Do less aggressive pruning for higher movecounts
Move part of heuristic that makes reduction less before pruning stage. Passed STC: https://tests.stockfishchess.org/tests/view/679fdf1b0774dfd78deb13b3 LLR: 2.93 (-2.94,2.94) <0.00,2.00> Total: 47136 W: 12484 L: 12146 D: 22506 Ptnml(0-2): 211, 5472, 11866, 5806, 213 Passed LTC: https://tests.stockfishchess.org/tests/view/679fe6790774dfd78deb1753 LLR: 2.94 (-2.94,2.94) <0.50,2.50> Total: 100536 W: 25837 L: 25383 D: 49316 Ptnml(0-2): 103, 10990, 27622, 11456, 97 closes https://github.com/official-stockfish/Stockfish/pull/5853 Bench: 3265587
This commit is contained in:
@@ -999,6 +999,8 @@ moves_loop: // When in check, search starts here
|
||||
|
||||
Depth r = reduction(improving, depth, moveCount, delta);
|
||||
|
||||
r -= 32 * moveCount;
|
||||
|
||||
// Increase reduction for ttPv nodes (*Scaler)
|
||||
// Smaller or even negative value is better for short time controls
|
||||
// Bigger value is better for long time controls
|
||||
@@ -1169,7 +1171,7 @@ moves_loop: // When in check, search starts here
|
||||
|
||||
// These reduction adjustments have no proven non-linear scaling
|
||||
|
||||
r += 307 - moveCount * 64;
|
||||
r += 307 - moveCount * 32;
|
||||
|
||||
r -= std::abs(correctionValue) / 34112;
|
||||
|
||||
|
||||
Reference in New Issue
Block a user