Commit Graph
10 Commits
Author SHA1 Message Date
Stefan GeschwentnerandMarco Costalba 2ead74d1e2 Remove depth condition for ttPv (#2166)
Currently PV nodes with a depth <= 4 were ignored for ttPv. Now remove this constraint and use all PV nodes.

STC:
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 52209 W: 11755 L: 11694 D: 28760
http://tests.stockfishchess.org/tests/view/5cebc2d30ebc5925cf07b93a

LTC:
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 20874 W: 3689 L: 3568 D: 13617
http://tests.stockfishchess.org/tests/view/5cec01fc0ebc5925cf07c62d
2019-06-09 14:34:51 +02:00
Stefan GeschwentnerandStéphane Nicolet a4eda3056e Rank threats on pinned pawns
Add for pinned pawns half of the standard rank based threat bonus.

STC:
LLR: 2.97 (-2.94,2.94) [0.00,5.00]
Total: 44010 W: 9987 L: 9635 D: 24388
http://tests.stockfishchess.org/tests/view/5b58aa780ebc5902bdb88c7a

LTC:
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 29475 W: 5089 L: 4847 D: 19539
http://tests.stockfishchess.org/tests/view/5b58b56c0ebc5902bdb88f37

Closes https://github.com/official-stockfish/Stockfish/pull/1701

Bench: 4503866
2018-07-26 01:29:12 +02:00
Stefan GeschwentnerandStéphane Nicolet c4c2e08f0d Tweak stat bonus
Increase stat bonus by 1/32 and adjust the divisor of main and capture
history tables to 10692.

STC:
LLR: 2.96 (-2.94,2.94) [0.00,4.00]
Total: 28437 W: 6444 L: 6166 D: 15827
http://tests.stockfishchess.org/tests/view/5b579b4d0ebc5902bdb87139

LTC:
LLR: 2.96 (-2.94,2.94) [0.00,4.00]
Total: 111204 W: 19160 L: 18644 D: 73400
http://tests.stockfishchess.org/tests/view/5b57a7c60ebc5902bdb872d3

Closes https://github.com/official-stockfish/Stockfish/pull/1698

Bench: 4778882
2018-07-25 18:02:07 +02:00
Stefan GeschwentnerandStéphane Nicolet 50287a55d3 Tweak reductions formula: 0.88 * depth + 0.12
Replace the depth part in the reduction formula for higher depths
with a slower growing linear function. So for depth > 3 less reductions
are used.

What we can try next:
- move the break point to even higher depths
- tweak the slope for lower and higher depth
- even possibly use a further higher depth threshold for a another
  slower growing function

STC:
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 25317 W: 5763 L: 5505 D: 14049
http://tests.stockfishchess.org/tests/view/5b54f9f70ebc5902bdb840ed

LTC:
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 7451 W: 1320 L: 1167 D: 4964
http://tests.stockfishchess.org/tests/view/5b54feeb0ebc5902bdb84244

Closes https://github.com/official-stockfish/Stockfish/pull/1692

Bench: 4617359
2018-07-23 09:16:29 +02:00
Stefan GeschwentnerandStéphane Nicolet 12e79be910 Better check evasion move sorting
Use in addition the counter move history table for sorting quiet
check evasion moves in main and quiecence search. Also rename
"contHistory" to "continuationHistory" while there.

STC:
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 73284 W: 16433 L: 15938 D: 40913
http://tests.stockfishchess.org/tests/view/5b4f526e0ebc5902bdb7a401

LTC:
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 12135 W: 2171 L: 1997 D: 7967
http://tests.stockfishchess.org/tests/view/5b4fc0ef0ebc5902bdb7ae0e

Closes https://github.com/official-stockfish/Stockfish/pull/1685

Bench 4817583
2018-07-19 18:27:20 +02:00
Stefan GeschwentnerandStéphane Nicolet 76d4a80084 Tune null move pruning
Use a lower stat score threshold of 22500.

Failed yellow at STC after many games (~0.92 Elo):
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 73978 W: 14947 L: 14834 D: 44197
http://tests.stockfishchess.org/tests/view/5af2deec0ebc5968e6523f80

But scales good and passed LTC:
LLR: 2.96 (-2.94,2.94) [0.00,4.00]
Total: 23147 W: 3453 L: 3237 D: 16457
http://tests.stockfishchess.org/tests/view/5af3cf820ebc5968e6524022

Closes https://github.com/official-stockfish/Stockfish/pull/1591

Bench: 4777674
2018-05-10 12:28:18 +02:00
Stefan GeschwentnerandStéphane Nicolet 59d10374ca Blocked pawn storm
In pawn structures like white pawns f6,h6 against black pawns f7,g6,h7
the attack on the king is blocked by the own pawns. So decrease the
penalty for king safety.

See diagram and discussion in
https://github.com/official-stockfish/Stockfish/pull/1434

A sample position that this patch wants to avoid is the following
1rr2bk1/3q1p1p/2n1bPpP/pp1pP3/2pP4/P1P1B3/1PBQN1P1/1K3R1R w - - 0 1

White pawn storm on the king side was a disaster, it locked the king
side completely. Therefore, all the king tropism bonus that white have
on the king side are useless, and kingadjacent attacks too. Master
gives White a static +4.5 advantage, but White cannot win that game.
The patch is lowering this evaluation artefact.

STC:
LLR: 2.94 (-2.94,2.94) [0.00,5.00]
Total: 16467 W: 3750 L: 3537 D: 9180
http://tests.stockfishchess.org/tests/view/5a92102d0ebc590297cc87d0

LTC:
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 64242 W: 11130 L: 10745 D: 42367
http://tests.stockfishchess.org/tests/view/5a923dc80ebc590297cc8806

This version includes reformatting and speed optimization by Alain Savard.

Bench: 5643527
2018-02-28 13:03:35 +01:00
Stefan GeschwentnerandStéphane Nicolet cccbecb6f8 Stat score initialization: grandchildren
This is the sequel of the previous patch, we now let the parent node initialize
stat score to zero once for all grandchildren.

Initialize statScore to zero for the grandchildren of the current position.
So statScore is shared between all grandchildren and only the first grandchild
starts with statScore = 0. Later grandchildren start with the last calculated
statScore of the previous grandchild. This influences the reduction rules in
LMR which are based on the statScore of parent position.

Tests results against the previous patch:

STC:
LLR: 2.96 (-2.94,2.94) [0.00,4.00]
Total: 23676 W: 5417 L: 5157 D: 13102
http://tests.stockfishchess.org/tests/view/5a9423a90ebc590297cc8a46

LTC:
LLR: 2.96 (-2.94,2.94) [0.00,4.00]
Total: 35485 W: 6168 L: 5898 D: 23419
http://tests.stockfishchess.org/tests/view/5a9435550ebc590297cc8a54

Bench: 5643520
2018-02-27 18:42:59 +01:00
Stefan GeschwentnerandStéphane Nicolet 1463881153 Stat score initialization: children
Let the parent node initialize stat score to zero once for all siblings.

Initialize statScore to zero for the children of the current position.
So statScore is shared between sibling positions and only the first sibling
starts with statScore = 0. Later siblings start with the last calculated
statScore of the previous sibling. This influences the reduction rules in
in LMR which are based on the statScore of parent position.

STC:
LLR: 2.96 (-2.94,2.94) [0.00,4.00]
Total: 22683 W: 5202 L: 4946 D: 12535
http://tests.stockfishchess.org/tests/view/5a93315f0ebc590297cc894f

LTC:
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 48548 W: 8346 L: 8035 D: 32167
http://tests.stockfishchess.org/tests/view/5a933ba90ebc590297cc8962

Bench: 5833683
2018-02-27 18:18:56 +01:00
Stefan GeschwentnerandMarco Costalba a87a1005ad Attack threats
Give bonus for safe attack threats from bishops and rooks on opponent queen

STC:
LLR: 2.97 (-2.94,2.94) [0.00,5.00]
Total: 8629 W: 1599 L: 1438 D: 5592
http://tests.stockfishchess.org/tests/view/5a1ad4490ebc590ccbb8b30d

LTC:
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 7093 W: 987 L: 846 D: 5260
http://tests.stockfishchess.org/tests/view/5a1aec5d0ebc590ccbb8b317

Bench: 5051254
2017-12-03 10:41:48 +01:00