deepcrayon
/
stockfish
1
0
Fork 0
Code Releases Activity

Increase see prune depth 

This seems good at short TC controls.

After 10000 games at 20+0.05
ELO: 9.56 +-6.8 (95%) LOS: 100.0%
Total: 10000 W: 1949 L: 1674 D: 6377

Testing at long TC and regression testing is still
ongoing. So this is a bit speculative commit and
could be reverted in the future.

Also re-testing at long TC the SEE pruning in PV nodes
seems less effective (perhaps even a regression, but
still ongoing) so disabled for now.

bench: 4968764
sf_3_base
Marco Costalba 2013-03-04 09:38:58 +01:00
parent db322e6a63
commit 10429dd616
1 changed files with 5 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
src/search.cpp
View File

@ -852,14 +852,14 @@ split_point_start: // At split points actual search starts from here
newDepth = depth - ONE_PLY + ext;
// Step 13. Futility pruning (is omitted in PV nodes)
if ( !captureOrPromotion
if ( !PvNode
&& !captureOrPromotion
&& !inCheck
&& !dangerous
&& move != ttMove)
{
// Move count based pruning
if ( !PvNode
&& depth < 16 * ONE_PLY
if ( depth < 16 * ONE_PLY
&& moveCount >= FutilityMoveCounts[depth]
&& (!threatMove || !refutes(pos, move, threatMove)))
{
@ -876,7 +876,7 @@ split_point_start: // At split points actual search starts from here
futilityValue = ss->staticEval + ss->evalMargin + futility_margin(predictedDepth, moveCount)
+ Gain[pos.piece_moved(move)][to_sq(move)];
if (!PvNode && futilityValue < beta)
if (futilityValue < beta)
{
if (SpNode)
splitPoint->mutex.lock();
@ -885,7 +885,7 @@ split_point_start: // At split points actual search starts from here
}
// Prune moves with negative SEE at low depths
if ( predictedDepth < 2 * ONE_PLY
if ( predictedDepth < 3 * ONE_PLY
&& pos.see_sign(move) < 0)
{
if (SpNode)