1
0
Fork 0
stockfish/src/search.cpp

1660 lines
57 KiB
C++
Raw Normal View History

2008-08-31 23:59:13 -06:00
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
Copyright (C) 2015-2016 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
2008-08-31 23:59:13 -06:00
Stockfish is free software: you can redistribute it and/or modify
2008-08-31 23:59:13 -06:00
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
2008-08-31 23:59:13 -06:00
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
2008-08-31 23:59:13 -06:00
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <algorithm>
2008-08-31 23:59:13 -06:00
#include <cassert>
#include <cmath>
#include <cstring> // For std::memset
#include <iostream>
2008-08-31 23:59:13 -06:00
#include <sstream>
#include "evaluate.h"
#include "misc.h"
#include "movegen.h"
2008-08-31 23:59:13 -06:00
#include "movepick.h"
#include "position.h"
2008-08-31 23:59:13 -06:00
#include "search.h"
#include "timeman.h"
2008-08-31 23:59:13 -06:00
#include "thread.h"
#include "tt.h"
#include "uci.h"
2015-01-18 00:05:05 -07:00
#include "syzygy/tbprobe.h"
namespace Search {
SignalsType Signals;
LimitsType Limits;
}
2015-01-18 00:05:05 -07:00
namespace Tablebases {
int Cardinality;
uint64_t Hits;
bool RootInTB;
bool UseRule50;
Depth ProbeDepth;
Value Score;
}
namespace TB = Tablebases;
using std::string;
using Eval::evaluate;
using namespace Search;
2008-08-31 23:59:13 -06:00
namespace {
// Different node types, used as a template parameter
enum NodeType { NonPV, PV };
// Razoring and futility margin based on depth
const int razor_margin[4] = { 483, 570, 603, 554 };
Value futility_margin(Depth d) { return Value(150 * d / ONE_PLY); }
2008-08-31 23:59:13 -06:00
// Futility and reductions lookup tables, initialized at startup
int FutilityMoveCounts[2][16]; // [improving][depth]
int Reductions[2][2][64][64]; // [pv][improving][depth][moveNumber]
template <bool PvNode> Depth reduction(bool i, Depth d, int mn) {
return Reductions[PvNode][i][std::min(d / ONE_PLY, 63)][std::min(mn, 63)] * ONE_PLY;
}
// Skill structure is used to implement strength limit
struct Skill {
Skill(int l) : level(l) {}
bool enabled() const { return level < 20; }
bool time_to_pick(Depth depth) const { return depth / ONE_PLY == 1 + level; }
Move best_move(size_t multiPV) { return best ? best : pick_best(multiPV); }
Move pick_best(size_t multiPV);
int level;
Move best = MOVE_NONE;
};
// EasyMoveManager structure is used to detect an 'easy move'. When the PV is
// stable across multiple search iterations, we can quickly return the best move.
struct EasyMoveManager {
void clear() {
stableCnt = 0;
expectedPosKey = 0;
pv[0] = pv[1] = pv[2] = MOVE_NONE;
}
Move get(Key key) const {
return expectedPosKey == key ? pv[2] : MOVE_NONE;
}
void update(Position& pos, const std::vector<Move>& newPv) {
assert(newPv.size() >= 3);
// Keep track of how many times in a row the 3rd ply remains stable
stableCnt = (newPv[2] == pv[2]) ? stableCnt + 1 : 0;
if (!std::equal(newPv.begin(), newPv.begin() + 3, pv))
{
std::copy(newPv.begin(), newPv.begin() + 3, pv);
StateInfo st[2];
pos.do_move(newPv[0], st[0], pos.gives_check(newPv[0]));
pos.do_move(newPv[1], st[1], pos.gives_check(newPv[1]));
expectedPosKey = pos.key();
pos.undo_move(newPv[1]);
pos.undo_move(newPv[0]);
}
}
int stableCnt;
Key expectedPosKey;
Move pv[3];
};
// Set of rows with half bits set to 1 and half to 0. It is used to allocate
// the search depths across the threads.
typedef std::vector<int> Row;
const Row HalfDensity[] = {
{0, 1},
{1, 0},
{0, 0, 1, 1},
{0, 1, 1, 0},
{1, 1, 0, 0},
{1, 0, 0, 1},
{0, 0, 0, 1, 1, 1},
{0, 0, 1, 1, 1, 0},
{0, 1, 1, 1, 0, 0},
{1, 1, 1, 0, 0, 0},
{1, 1, 0, 0, 0, 1},
{1, 0, 0, 0, 1, 1},
{0, 0, 0, 0, 1, 1, 1, 1},
{0, 0, 0, 1, 1, 1, 1, 0},
{0, 0, 1, 1, 1, 1, 0 ,0},
{0, 1, 1, 1, 1, 0, 0 ,0},
{1, 1, 1, 1, 0, 0, 0 ,0},
{1, 1, 1, 0, 0, 0, 0 ,1},
{1, 1, 0, 0, 0, 0, 1 ,1},
{1, 0, 0, 0, 0, 1, 1 ,1},
};
const size_t HalfDensitySize = std::extent<decltype(HalfDensity)>::value;
EasyMoveManager EasyMove;
Value DrawValue[COLOR_NB];
template <NodeType NT>
Value search(Position& pos, Stack* ss, Value alpha, Value beta, Depth depth, bool cutNode);
template <NodeType NT, bool InCheck>
Value qsearch(Position& pos, Stack* ss, Value alpha, Value beta, Depth depth);
Value value_to_tt(Value v, int ply);
Value value_from_tt(Value v, int ply);
void update_pv(Move* pv, Move move, Move* childPv);
void update_cm_stats(Stack* ss, Piece pc, Square s, Value bonus);
void update_stats(const Position& pos, Stack* ss, Move move, Move* quiets, int quietsCnt, Value bonus);
void check_time();
2008-08-31 23:59:13 -06:00
} // namespace
2008-08-31 23:59:13 -06:00
/// Search::init() is called during startup to initialize various lookup tables
2008-08-31 23:59:13 -06:00
void Search::init() {
for (int imp = 0; imp <= 1; ++imp)
for (int d = 1; d < 64; ++d)
for (int mc = 1; mc < 64; ++mc)
{
double r = log(d) * log(mc) / 2;
if (r < 0.80)
continue;
Reductions[NonPV][imp][d][mc] = int(std::round(r));
Reductions[PV][imp][d][mc] = std::max(Reductions[NonPV][imp][d][mc] - 1, 0);
// Increase reduction for non-PV nodes when eval is not improving
if (!imp && Reductions[NonPV][imp][d][mc] >= 2)
Reductions[NonPV][imp][d][mc]++;
}
for (int d = 0; d < 16; ++d)
{
FutilityMoveCounts[0][d] = int(2.4 + 0.773 * pow(d + 0.00, 1.8));
FutilityMoveCounts[1][d] = int(2.9 + 1.045 * pow(d + 0.49, 1.8));
}
}
/// Search::clear() resets search state to zero, to obtain reproducible results
void Search::clear() {
TT.clear();
for (Thread* th : Threads)
{
th->history.clear();
th->counterMoves.clear();
th->fromTo.clear();
Use per-thread counterMoveHistory Drops a scalability bottleneck due to memory contention of a single shared table across threads. The effect starts to be sensible with a high number of threads. Specifically we have a small regression with 7 threads both at 60 and 180 seconds TC: 10000 @ 60+0.6 th 7 ELO: -2.46 +-3.2 (95%) LOS: 6.5% Total: 9896 W: 1037 L: 1107 D: 7752 5000 @ 180+0.6 th 7 ELO: -1.95 +-4.1 (95%) LOS: 17.7% Total: 5000 W: 444 L: 472 D: 4084 We have a regression because counterMoveHistory table is quite big and it takes time for a single thread to fill it. Sharing the table yields to a higher fill rate and better quality of moves and up to 7 threads the benefits of sharing more then compensate the loss in speed due to contention. Interestingly even with a 3X longer TC, so with more time for the single thread to catch up, the improvment is quite limited and below noise level. It seems we really need much longer TC to saturate the table. When we move to high threads number it's another story: 5000 @ 60+0.6 th 22 ELO: 3.49 +-4.3 (95%) LOS: 94.6% Total: 4880 W: 490 L: 441 D: 3949 2000 @ 60+0.6 th 32 ELO: 8.34 +-6.9 (95%) LOS: 99.1% Total: 2000 W: 229 L: 181 D: 1590 As expected the speed-up more than compensates the filling rate, and we expect that with tournament TC, where single thread is able to saturate the table, the difference will be even stronger. For instance for TCEC 9 super-final time control will be 180 minutes + 15 seconds and this scalability improvement seems definitely the way to go. So, summarizing: GOOD: Measured big improvement in high core scenario Suitable for TCEC 9 superfinal (big hardware, very long TC) Consistent and natural patch that extends to counterMoveHistory what we already do for remaining history tables, that are all per-thread Non functional change for the common case of a single core Very simple (just 6 lines modified, no added ones) BAD: Small regression (within 2-3 ELO) with few threads and short TC bench: 5341477
2016-08-29 01:11:20 -06:00
th->counterMoveHistory.clear();
}
Threads.main()->previousScore = VALUE_INFINITE;
}
/// Search::perft() is our utility to verify move generation. All the leaf nodes
/// up to the given depth are generated and counted, and the sum is returned.
template<bool Root>
uint64_t Search::perft(Position& pos, Depth depth) {
StateInfo st;
uint64_t cnt, nodes = 0;
const bool leaf = (depth == 2 * ONE_PLY);
for (const auto& m : MoveList<LEGAL>(pos))
{
if (Root && depth <= ONE_PLY)
cnt = 1, nodes++;
else
{
pos.do_move(m, st, pos.gives_check(m));
cnt = leaf ? MoveList<LEGAL>(pos).size() : perft<false>(pos, depth - ONE_PLY);
nodes += cnt;
pos.undo_move(m);
}
if (Root)
sync_cout << UCI::move(m, pos.is_chess960()) << ": " << cnt << sync_endl;
}
return nodes;
}
template uint64_t Search::perft<true>(Position&, Depth);
/// MainThread::search() is called by the main thread when the program receives
/// the UCI 'go' command. It searches from the root position and outputs the "bestmove".
2008-08-31 23:59:13 -06:00
void MainThread::search() {
2008-08-31 23:59:13 -06:00
Color us = rootPos.side_to_move();
Time.init(Limits, us, rootPos.game_ply());
int contempt = Options["Contempt"] * PawnValueEg / 100; // From centipawns
Add support for playing in 'nodes as time' mode When running more games in parallel, or simply when running a game with a background process, due to how OS scheduling works, there is no guarantee that the CPU resources allocated evenly between the two players. This introduces noise in the result that leads to unreliable result and in the worst cases can even invalidate the result. For instance in SF test framework we avoid running from clouds virtual machines because are a known source of very unstable CPU speed. To overcome this issue, without requiring changes to the GUI, the idea is to use searched nodes instead of time, and to convert time to available nodes upfront, at the beginning of the game. When nodestime UCI option is set at a given nodes per milliseconds (npmsec), at the beginning of the game (and only once), the engine reads the available time to think, sent by the GUI with 'go wtime x' UCI command. Then it translates time in available nodes (nodes = npmsec * x), then feeds available nodes instead of time to the time management logic and starts the search. During the search the engine checks the searched nodes against the available ones in such a way that all the time management logic still fully applies, and the game mimics a real one played on real time. When the search finishes, before returning best move, the total available nodes are updated, subtracting the real searched nodes. After the first move, the time information sent by the GUI is ignored, and the engine fully relies on the updated total available nodes to feed time management. To avoid time losses, the speed of the engine (npms) must be set to a value lower than real speed so that if the real TC is for instance 30 secs, and npms is half of the real speed, the game will last on average 15 secs, so much less than the TC limit, providing for a safety 'time buffer'. There are 2 main limitations with this mode. 1. Engine speed should be the same for both players, and this limits the approach to mainly parameter tuning patches. 2. Because npms is fixed while, in real engines, the speed increases toward endgame, this introduces an artifact that is equivalent to an altered time management. Namely it is like the time management gives less available time than what should be in standard case. May be the second limitation could be mitigated in a future with a smarter 'dynamic npms' approach. Tests shows that the standard deviation of the results with 'nodestime' is lower than in standard TC, as is expected because now all the introduced noise due the random speed variability of the engines during the game is fully removed. Original NIT idea by Michael Hoffman that shows how to play in NIT mode without requiring changes to the GUI. This implementation goes a bit further, the key difference is that we read TC from GUI only once upfront instead of re-reading after every move as in Michael's implementation. No functional change.
2015-03-22 14:15:44 -06:00
DrawValue[ us] = VALUE_DRAW - Value(contempt);
DrawValue[~us] = VALUE_DRAW + Value(contempt);
if (rootMoves.empty())
{
rootMoves.push_back(RootMove(MOVE_NONE));
sync_cout << "info depth 0 score "
<< UCI::value(rootPos.checkers() ? -VALUE_MATE : VALUE_DRAW)
<< sync_endl;
}
else
{
for (Thread* th : Threads)
if (th != this)
th->start_searching();
Thread::search(); // Let's start searching!
}
Add support for playing in 'nodes as time' mode When running more games in parallel, or simply when running a game with a background process, due to how OS scheduling works, there is no guarantee that the CPU resources allocated evenly between the two players. This introduces noise in the result that leads to unreliable result and in the worst cases can even invalidate the result. For instance in SF test framework we avoid running from clouds virtual machines because are a known source of very unstable CPU speed. To overcome this issue, without requiring changes to the GUI, the idea is to use searched nodes instead of time, and to convert time to available nodes upfront, at the beginning of the game. When nodestime UCI option is set at a given nodes per milliseconds (npmsec), at the beginning of the game (and only once), the engine reads the available time to think, sent by the GUI with 'go wtime x' UCI command. Then it translates time in available nodes (nodes = npmsec * x), then feeds available nodes instead of time to the time management logic and starts the search. During the search the engine checks the searched nodes against the available ones in such a way that all the time management logic still fully applies, and the game mimics a real one played on real time. When the search finishes, before returning best move, the total available nodes are updated, subtracting the real searched nodes. After the first move, the time information sent by the GUI is ignored, and the engine fully relies on the updated total available nodes to feed time management. To avoid time losses, the speed of the engine (npms) must be set to a value lower than real speed so that if the real TC is for instance 30 secs, and npms is half of the real speed, the game will last on average 15 secs, so much less than the TC limit, providing for a safety 'time buffer'. There are 2 main limitations with this mode. 1. Engine speed should be the same for both players, and this limits the approach to mainly parameter tuning patches. 2. Because npms is fixed while, in real engines, the speed increases toward endgame, this introduces an artifact that is equivalent to an altered time management. Namely it is like the time management gives less available time than what should be in standard case. May be the second limitation could be mitigated in a future with a smarter 'dynamic npms' approach. Tests shows that the standard deviation of the results with 'nodestime' is lower than in standard TC, as is expected because now all the introduced noise due the random speed variability of the engines during the game is fully removed. Original NIT idea by Michael Hoffman that shows how to play in NIT mode without requiring changes to the GUI. This implementation goes a bit further, the key difference is that we read TC from GUI only once upfront instead of re-reading after every move as in Michael's implementation. No functional change.
2015-03-22 14:15:44 -06:00
// When playing in 'nodes as time' mode, subtract the searched nodes from
// the available ones before exiting.
Add support for playing in 'nodes as time' mode When running more games in parallel, or simply when running a game with a background process, due to how OS scheduling works, there is no guarantee that the CPU resources allocated evenly between the two players. This introduces noise in the result that leads to unreliable result and in the worst cases can even invalidate the result. For instance in SF test framework we avoid running from clouds virtual machines because are a known source of very unstable CPU speed. To overcome this issue, without requiring changes to the GUI, the idea is to use searched nodes instead of time, and to convert time to available nodes upfront, at the beginning of the game. When nodestime UCI option is set at a given nodes per milliseconds (npmsec), at the beginning of the game (and only once), the engine reads the available time to think, sent by the GUI with 'go wtime x' UCI command. Then it translates time in available nodes (nodes = npmsec * x), then feeds available nodes instead of time to the time management logic and starts the search. During the search the engine checks the searched nodes against the available ones in such a way that all the time management logic still fully applies, and the game mimics a real one played on real time. When the search finishes, before returning best move, the total available nodes are updated, subtracting the real searched nodes. After the first move, the time information sent by the GUI is ignored, and the engine fully relies on the updated total available nodes to feed time management. To avoid time losses, the speed of the engine (npms) must be set to a value lower than real speed so that if the real TC is for instance 30 secs, and npms is half of the real speed, the game will last on average 15 secs, so much less than the TC limit, providing for a safety 'time buffer'. There are 2 main limitations with this mode. 1. Engine speed should be the same for both players, and this limits the approach to mainly parameter tuning patches. 2. Because npms is fixed while, in real engines, the speed increases toward endgame, this introduces an artifact that is equivalent to an altered time management. Namely it is like the time management gives less available time than what should be in standard case. May be the second limitation could be mitigated in a future with a smarter 'dynamic npms' approach. Tests shows that the standard deviation of the results with 'nodestime' is lower than in standard TC, as is expected because now all the introduced noise due the random speed variability of the engines during the game is fully removed. Original NIT idea by Michael Hoffman that shows how to play in NIT mode without requiring changes to the GUI. This implementation goes a bit further, the key difference is that we read TC from GUI only once upfront instead of re-reading after every move as in Michael's implementation. No functional change.
2015-03-22 14:15:44 -06:00
if (Limits.npmsec)
Time.availableNodes += Limits.inc[us] - Threads.nodes_searched();
Add support for playing in 'nodes as time' mode When running more games in parallel, or simply when running a game with a background process, due to how OS scheduling works, there is no guarantee that the CPU resources allocated evenly between the two players. This introduces noise in the result that leads to unreliable result and in the worst cases can even invalidate the result. For instance in SF test framework we avoid running from clouds virtual machines because are a known source of very unstable CPU speed. To overcome this issue, without requiring changes to the GUI, the idea is to use searched nodes instead of time, and to convert time to available nodes upfront, at the beginning of the game. When nodestime UCI option is set at a given nodes per milliseconds (npmsec), at the beginning of the game (and only once), the engine reads the available time to think, sent by the GUI with 'go wtime x' UCI command. Then it translates time in available nodes (nodes = npmsec * x), then feeds available nodes instead of time to the time management logic and starts the search. During the search the engine checks the searched nodes against the available ones in such a way that all the time management logic still fully applies, and the game mimics a real one played on real time. When the search finishes, before returning best move, the total available nodes are updated, subtracting the real searched nodes. After the first move, the time information sent by the GUI is ignored, and the engine fully relies on the updated total available nodes to feed time management. To avoid time losses, the speed of the engine (npms) must be set to a value lower than real speed so that if the real TC is for instance 30 secs, and npms is half of the real speed, the game will last on average 15 secs, so much less than the TC limit, providing for a safety 'time buffer'. There are 2 main limitations with this mode. 1. Engine speed should be the same for both players, and this limits the approach to mainly parameter tuning patches. 2. Because npms is fixed while, in real engines, the speed increases toward endgame, this introduces an artifact that is equivalent to an altered time management. Namely it is like the time management gives less available time than what should be in standard case. May be the second limitation could be mitigated in a future with a smarter 'dynamic npms' approach. Tests shows that the standard deviation of the results with 'nodestime' is lower than in standard TC, as is expected because now all the introduced noise due the random speed variability of the engines during the game is fully removed. Original NIT idea by Michael Hoffman that shows how to play in NIT mode without requiring changes to the GUI. This implementation goes a bit further, the key difference is that we read TC from GUI only once upfront instead of re-reading after every move as in Michael's implementation. No functional change.
2015-03-22 14:15:44 -06:00
// When we reach the maximum depth, we can arrive here without a raise of
// Signals.stop. However, if we are pondering or in an infinite search,
// the UCI protocol states that we shouldn't print the best move before the
// GUI sends a "stop" or "ponderhit" command. We therefore simply wait here
// until the GUI sends one of those commands (which also raises Signals.stop).
if (!Signals.stop && (Limits.ponder || Limits.infinite))
{
Signals.stopOnPonderhit = true;
wait(Signals.stop);
}
// Stop the threads if not already stopped
Signals.stop = true;
// Wait until all threads have finished
for (Thread* th : Threads)
if (th != this)
th->wait_for_search_finished();
// Check if there are threads with a better score than main thread
Thread* bestThread = this;
if ( !this->easyMovePlayed
&& Options["MultiPV"] == 1
&& !Limits.depth
&& !Skill(Options["Skill Level"]).enabled()
&& rootMoves[0].pv[0] != MOVE_NONE)
{
for (Thread* th : Threads)
if ( th->completedDepth > bestThread->completedDepth
&& th->rootMoves[0].score > bestThread->rootMoves[0].score)
bestThread = th;
}
previousScore = bestThread->rootMoves[0].score;
// Send new PV when needed
if (bestThread != this)
sync_cout << UCI::pv(bestThread->rootPos, bestThread->completedDepth, -VALUE_INFINITE, VALUE_INFINITE) << sync_endl;
sync_cout << "bestmove " << UCI::move(bestThread->rootMoves[0].pv[0], rootPos.is_chess960());
if (bestThread->rootMoves[0].pv.size() > 1 || bestThread->rootMoves[0].extract_ponder_from_tt(rootPos))
std::cout << " ponder " << UCI::move(bestThread->rootMoves[0].pv[1], rootPos.is_chess960());
std::cout << sync_endl;
2008-08-31 23:59:13 -06:00
}
// Thread::search() is the main iterative deepening loop. It calls search()
// repeatedly with increasing depth until the allocated thinking time has been
// consumed, the user stops the search, or the maximum search depth is reached.
2008-08-31 23:59:13 -06:00
void Thread::search() {
2008-08-31 23:59:13 -06:00
Stack stack[MAX_PLY+7], *ss = stack+5; // To allow referencing (ss-5) and (ss+2)
Value bestValue, alpha, beta, delta;
Move easyMove = MOVE_NONE;
MainThread* mainThread = (this == Threads.main() ? Threads.main() : nullptr);
std::memset(ss-5, 0, 8 * sizeof(Stack));
bestValue = delta = alpha = -VALUE_INFINITE;
beta = VALUE_INFINITE;
completedDepth = DEPTH_ZERO;
if (mainThread)
{
easyMove = EasyMove.get(rootPos.key());
EasyMove.clear();
mainThread->easyMovePlayed = mainThread->failedLow = false;
mainThread->bestMoveChanges = 0;
TT.new_search();
}
size_t multiPV = Options["MultiPV"];
Skill skill(Options["Skill Level"]);
// When playing with strength handicap enable MultiPV search that we will
// use behind the scenes to retrieve a set of possible moves.
if (skill.enabled())
multiPV = std::max(multiPV, (size_t)4);
multiPV = std::min(multiPV, rootMoves.size());
// Iterative deepening loop until requested to stop or the target depth is reached
while ( (rootDepth += ONE_PLY) < DEPTH_MAX
&& !Signals.stop
&& (!Limits.depth || Threads.main()->rootDepth / ONE_PLY <= Limits.depth))
{
// Set up the new depths for the helper threads skipping on average every
// 2nd ply (using a half-density matrix).
if (!mainThread)
{
const Row& row = HalfDensity[(idx - 1) % HalfDensitySize];
if (row[(rootDepth / ONE_PLY + rootPos.game_ply()) % row.size()])
continue;
}
// Age out PV variability metric
if (mainThread)
mainThread->bestMoveChanges *= 0.505, mainThread->failedLow = false;
// Save the last iteration's scores before first PV line is searched and
// all the move scores except the (new) PV are set to -VALUE_INFINITE.
for (RootMove& rm : rootMoves)
rm.previousScore = rm.score;
// MultiPV loop. We perform a full root search for each PV line
for (PVIdx = 0; PVIdx < multiPV && !Signals.stop; ++PVIdx)
{
// Reset aspiration window starting size
if (rootDepth >= 5 * ONE_PLY)
{
delta = Value(18);
alpha = std::max(rootMoves[PVIdx].previousScore - delta,-VALUE_INFINITE);
beta = std::min(rootMoves[PVIdx].previousScore + delta, VALUE_INFINITE);
}
// Start with a small aspiration window and, in the case of a fail
// high/low, re-search with a bigger window until we're not failing
// high/low anymore.
while (true)
{
bestValue = ::search<PV>(rootPos, ss, alpha, beta, rootDepth, false);
// Bring the best move to the front. It is critical that sorting
// is done with a stable algorithm because all the values but the
// first and eventually the new best one are set to -VALUE_INFINITE
// and we want to keep the same order for all the moves except the
// new PV that goes to the front. Note that in case of MultiPV
// search the already searched PV lines are preserved.
std::stable_sort(rootMoves.begin() + PVIdx, rootMoves.end());
// If search has been stopped, break immediately. Sorting and
// writing PV back to TT is safe because RootMoves is still
// valid, although it refers to the previous iteration.
if (Signals.stop)
break;
// When failing high/low give some update (without cluttering
// the UI) before a re-search.
if ( mainThread
&& multiPV == 1
&& (bestValue <= alpha || bestValue >= beta)
&& Time.elapsed() > 3000)
sync_cout << UCI::pv(rootPos, rootDepth, alpha, beta) << sync_endl;
// In case of failing low/high increase aspiration window and
// re-search, otherwise exit the loop.
if (bestValue <= alpha)
{
beta = (alpha + beta) / 2;
alpha = std::max(bestValue - delta, -VALUE_INFINITE);
if (mainThread)
{
mainThread->failedLow = true;
Signals.stopOnPonderhit = false;
}
}
else if (bestValue >= beta)
{
alpha = (alpha + beta) / 2;
beta = std::min(bestValue + delta, VALUE_INFINITE);
}
else
break;
delta += delta / 4 + 5;
assert(alpha >= -VALUE_INFINITE && beta <= VALUE_INFINITE);
}
// Sort the PV lines searched so far and update the GUI
std::stable_sort(rootMoves.begin(), rootMoves.begin() + PVIdx + 1);
if (!mainThread)
continue;
if (Signals.stop)
sync_cout << "info nodes " << Threads.nodes_searched()
<< " time " << Time.elapsed() << sync_endl;
else if (PVIdx + 1 == multiPV || Time.elapsed() > 3000)
sync_cout << UCI::pv(rootPos, rootDepth, alpha, beta) << sync_endl;
}
if (!Signals.stop)
completedDepth = rootDepth;
if (!mainThread)
continue;
// If skill level is enabled and time is up, pick a sub-optimal best move
if (skill.enabled() && skill.time_to_pick(rootDepth))
skill.pick_best(multiPV);
// Have we found a "mate in x"?
if ( Limits.mate
&& bestValue >= VALUE_MATE_IN_MAX_PLY
&& VALUE_MATE - bestValue <= 2 * Limits.mate)
Signals.stop = true;
// Do we have time for the next iteration? Can we stop searching now?
if (Limits.use_time_management())
{
if (!Signals.stop && !Signals.stopOnPonderhit)
{
// Stop the search if only one legal move is available, or if all
// of the available time has been used, or if we matched an easyMove
// from the previous search and just did a fast verification.
const int F[] = { mainThread->failedLow,
bestValue - mainThread->previousScore };
int improvingFactor = std::max(229, std::min(715, 357 + 119 * F[0] - 6 * F[1]));
double unstablePvFactor = 1 + mainThread->bestMoveChanges;
bool doEasyMove = rootMoves[0].pv[0] == easyMove
&& mainThread->bestMoveChanges < 0.03
&& Time.elapsed() > Time.optimum() * 5 / 42;
if ( rootMoves.size() == 1
|| Time.elapsed() > Time.optimum() * unstablePvFactor * improvingFactor / 628
|| (mainThread->easyMovePlayed = doEasyMove, doEasyMove))
{
// If we are allowed to ponder do not stop the search now but
// keep pondering until the GUI sends "ponderhit" or "stop".
if (Limits.ponder)
Signals.stopOnPonderhit = true;
else
Signals.stop = true;
}
}
if (rootMoves[0].pv.size() >= 3)
EasyMove.update(rootPos, rootMoves[0].pv);
else
EasyMove.clear();
}
2008-08-31 23:59:13 -06:00
}
if (!mainThread)
return;
// Clear any candidate easy move that wasn't stable for the last search
// iterations; the second condition prevents consecutive fast moves.
if (EasyMove.stableCnt < 6 || mainThread->easyMovePlayed)
EasyMove.clear();
// If skill level is enabled, swap best PV line with the sub-optimal one
if (skill.enabled())
std::swap(rootMoves[0], *std::find(rootMoves.begin(),
rootMoves.end(), skill.best_move(multiPV)));
}
namespace {
2008-08-31 23:59:13 -06:00
// search<>() is the main search function for both PV and non-PV nodes
2008-08-31 23:59:13 -06:00
template <NodeType NT>
Value search(Position& pos, Stack* ss, Value alpha, Value beta, Depth depth, bool cutNode) {
const bool PvNode = NT == PV;
const bool rootNode = PvNode && (ss-1)->ply == 0;
assert(-VALUE_INFINITE <= alpha && alpha < beta && beta <= VALUE_INFINITE);
assert(PvNode || (alpha == beta - 1));
assert(DEPTH_ZERO < depth && depth < DEPTH_MAX);
assert(!(PvNode && cutNode));
assert(depth / ONE_PLY * ONE_PLY == depth);
2008-08-31 23:59:13 -06:00
Move pv[MAX_PLY+1], quietsSearched[64];
StateInfo st;
TTEntry* tte;
Key posKey;
Move ttMove, move, excludedMove, bestMove;
Depth extension, newDepth;
Value bestValue, value, ttValue, eval, nullValue;
bool ttHit, inCheck, givesCheck, singularExtensionNode, improving;
bool captureOrPromotion, doFullDepthSearch, moveCountPruning;
Piece moved_piece;
int moveCount, quietCount;
// Step 1. Initialize node
Thread* thisThread = pos.this_thread();
inCheck = pos.checkers();
moveCount = quietCount = ss->moveCount = 0;
bestValue = -VALUE_INFINITE;
ss->ply = (ss-1)->ply + 1;
// Check for the available remaining time
if (thisThread->resetCalls.load(std::memory_order_relaxed))
{
thisThread->resetCalls = false;
thisThread->callsCnt = 0;
}
if (++thisThread->callsCnt > 4096)
{
for (Thread* th : Threads)
th->resetCalls = true;
check_time();
}
// Used to send selDepth info to GUI
if (PvNode && thisThread->maxPly < ss->ply)
thisThread->maxPly = ss->ply;
if (!rootNode)
{
// Step 2. Check for aborted search and immediate draw
if (Signals.stop.load(std::memory_order_relaxed) || pos.is_draw() || ss->ply >= MAX_PLY)
return ss->ply >= MAX_PLY && !inCheck ? evaluate(pos)
: DrawValue[pos.side_to_move()];
// Step 3. Mate distance pruning. Even if we mate at the next move our score
// would be at best mate_in(ss->ply+1), but if alpha is already bigger because
// a shorter mate was found upward in the tree then there is no need to search
// because we will never beat the current alpha. Same logic but with reversed
// signs applies also in the opposite condition of being mated instead of giving
// mate. In this case return a fail-high score.
alpha = std::max(mated_in(ss->ply), alpha);
beta = std::min(mate_in(ss->ply+1), beta);
if (alpha >= beta)
return alpha;
}
assert(0 <= ss->ply && ss->ply < MAX_PLY);
ss->currentMove = (ss+1)->excludedMove = bestMove = MOVE_NONE;
ss->counterMoves = nullptr;
(ss+1)->skipEarlyPruning = false;
(ss+2)->killers[0] = (ss+2)->killers[1] = MOVE_NONE;
// Step 4. Transposition table lookup. We don't want the score of a partial
// search to overwrite a previous full search TT value, so we use a different
// position key in case of an excluded move.
excludedMove = ss->excludedMove;
posKey = pos.key() ^ Key(excludedMove);
tte = TT.probe(posKey, ttHit);
ttValue = ttHit ? value_from_tt(tte->value(), ss->ply) : VALUE_NONE;
ttMove = rootNode ? thisThread->rootMoves[thisThread->PVIdx].pv[0]
: ttHit ? tte->move() : MOVE_NONE;
// At non-PV nodes we check for an early TT cutoff
if ( !PvNode
&& ttHit
&& tte->depth() >= depth
&& ttValue != VALUE_NONE // Possible in case of TT access race
&& (ttValue >= beta ? (tte->bound() & BOUND_LOWER)
: (tte->bound() & BOUND_UPPER)))
{
ss->currentMove = ttMove; // Can be MOVE_NONE
// If ttMove is quiet, update killers, history, counter move on TT hit
if (ttValue >= beta && ttMove)
{
int d = depth / ONE_PLY;
if (!pos.capture_or_promotion(ttMove))
{
Value bonus = Value(d * d + 2 * d - 2);
update_stats(pos, ss, ttMove, nullptr, 0, bonus);
}
// Extra penalty for a quiet TT move in previous ply when it gets refuted
if ((ss-1)->moveCount == 1 && !pos.captured_piece())
{
Value penalty = Value(d * d + 4 * d + 1);
Square prevSq = to_sq((ss-1)->currentMove);
update_cm_stats(ss-1, pos.piece_on(prevSq), prevSq, -penalty);
}
}
return ttValue;
2008-08-31 23:59:13 -06:00
}
2015-01-18 00:05:05 -07:00
// Step 4a. Tablebase probe
if (!rootNode && TB::Cardinality)
2015-01-18 00:05:05 -07:00
{
int piecesCnt = pos.count<ALL_PIECES>(WHITE) + pos.count<ALL_PIECES>(BLACK);
if ( piecesCnt <= TB::Cardinality
&& (piecesCnt < TB::Cardinality || depth >= TB::ProbeDepth)
&& pos.rule50_count() == 0
&& !pos.can_castle(ANY_CASTLING))
2015-01-18 00:05:05 -07:00
{
int found, v = Tablebases::probe_wdl(pos, &found);
if (found)
{
TB::Hits++;
int drawScore = TB::UseRule50 ? 1 : 0;
value = v < -drawScore ? -VALUE_MATE + MAX_PLY + ss->ply
: v > drawScore ? VALUE_MATE - MAX_PLY - ss->ply
: VALUE_DRAW + 2 * v * drawScore;
tte->save(posKey, value_to_tt(value, ss->ply), BOUND_EXACT,
std::min(DEPTH_MAX - ONE_PLY, depth + 6 * ONE_PLY),
MOVE_NONE, VALUE_NONE, TT.generation());
return value;
}
}
}
// Step 5. Evaluate the position statically
if (inCheck)
{
ss->staticEval = eval = VALUE_NONE;
goto moves_loop;
}
else if (ttHit)
{
// Never assume anything on values stored in TT
if ((ss->staticEval = eval = tte->eval()) == VALUE_NONE)
eval = ss->staticEval = evaluate(pos);
// Can ttValue be used as a better position evaluation?
if (ttValue != VALUE_NONE)
if (tte->bound() & (ttValue > eval ? BOUND_LOWER : BOUND_UPPER))
eval = ttValue;
}
else
{
eval = ss->staticEval =
(ss-1)->currentMove != MOVE_NULL ? evaluate(pos)
: -(ss-1)->staticEval + 2 * Eval::Tempo;
tte->save(posKey, VALUE_NONE, BOUND_NONE, DEPTH_NONE, MOVE_NONE,
ss->staticEval, TT.generation());
2011-10-31 01:28:59 -06:00
}
2008-08-31 23:59:13 -06:00
if (ss->skipEarlyPruning)
goto moves_loop;
// Step 6. Razoring (skipped when in check)
if ( !PvNode
&& depth < 4 * ONE_PLY
&& ttMove == MOVE_NONE
&& eval + razor_margin[depth / ONE_PLY] <= alpha)
{
if ( depth <= ONE_PLY
&& eval + razor_margin[3] <= alpha)
return qsearch<NonPV, false>(pos, ss, alpha, beta, DEPTH_ZERO);
Value ralpha = alpha - razor_margin[depth / ONE_PLY];
Better document razoring Use ralpha instead of rbeta * rbeta is confusing people. It took THREE attempts to code razoring at PV nodes correctly in a recent test, because of the rbeta trick. Unnecessary tricks should be avoided. * The more correct and self-documenting way of doing this, is to say that we use a zero window around alpha-margin, not beta-margin. The fact that, because we only do it at PV nodes, alpha happens to be beta-1 and that the current stuff with rbeta works, may be correct, but is confusing. Remove the misleading and partially erroneous comment about returning v + margin: * comments should explain what the code does, not what it could have done. * this comment is partially wrong in saying that v+margin is "logical", and that it is "surprising" that is doesn't work. From a theoretical perspective, at least 3 ways of doing this are equally defendable: 1/ fail hard: return alpha: The most conservative. We bet that the search will fail low, but we don't know by how much and don't want to take risks. 2/ aggressive fail soft: return v (what the current code does). This corresponds to normal fail soft, with the added assumption that we don't care about the reduction effect (see below point 3/) 3/ conservative fail soft: return v + margin. If the reduced search (qsearch) gives us a score <= v, we bet that the non reduced search will give us a score <= v + margin. * Saying that 2/ is "logical" implies that 1/ and 3/ are not, which is arguably wrong. Besides, experimental results tell us that 2/ beats 3/, and that's not something we can argue against: experimental results are the only trusted metric. * Also, with the benefit of hindsight, I don't think the fact that 2/ is better than 3/ is surprising at all. The point is that it is YOUR turn to move, and you are assuming that by NOT playing (and letting the opponent capture your hanging pieces in QS) you cannot generally GAIN razor_margin(depth). No functional change.
2014-02-02 18:41:32 -07:00
Value v = qsearch<NonPV, false>(pos, ss, ralpha, ralpha+1, DEPTH_ZERO);
if (v <= ralpha)
return v;
}
// Step 7. Futility pruning: child node (skipped when in check)
if ( !rootNode
&& depth < 7 * ONE_PLY
&& eval - futility_margin(depth) >= beta
&& eval < VALUE_KNOWN_WIN // Do not return unproven wins
&& pos.non_pawn_material(pos.side_to_move()))
return eval - futility_margin(depth);
// Step 8. Null move search with verification search (is omitted in PV nodes)
if ( !PvNode
&& eval >= beta
&& (ss->staticEval >= beta - 35 * (depth / ONE_PLY - 6) || depth >= 13 * ONE_PLY)
&& pos.non_pawn_material(pos.side_to_move()))
{
ss->currentMove = MOVE_NULL;
ss->counterMoves = nullptr;
assert(eval - beta >= 0);
// Null move dynamic reduction based on depth and value
Depth R = ((823 + 67 * depth / ONE_PLY) / 256 + std::min((eval - beta) / PawnValueMg, 3)) * ONE_PLY;
pos.do_null_move(st);
(ss+1)->skipEarlyPruning = true;
nullValue = depth-R < ONE_PLY ? -qsearch<NonPV, false>(pos, ss+1, -beta, -beta+1, DEPTH_ZERO)
: - search<NonPV>(pos, ss+1, -beta, -beta+1, depth-R, !cutNode);
(ss+1)->skipEarlyPruning = false;
pos.undo_null_move();
if (nullValue >= beta)
{
// Do not return unproven mate scores
if (nullValue >= VALUE_MATE_IN_MAX_PLY)
nullValue = beta;
if (depth < 12 * ONE_PLY && abs(beta) < VALUE_KNOWN_WIN)
return nullValue;
// Do verification search at high depths
ss->skipEarlyPruning = true;
Value v = depth-R < ONE_PLY ? qsearch<NonPV, false>(pos, ss, beta-1, beta, DEPTH_ZERO)
: search<NonPV>(pos, ss, beta-1, beta, depth-R, false);
ss->skipEarlyPruning = false;
if (v >= beta)
return nullValue;
}
}
2008-08-31 23:59:13 -06:00
// Step 9. ProbCut (skipped when in check)
// If we have a good enough capture and a reduced search returns a value
// much above beta, we can (almost) safely prune the previous move.
if ( !PvNode
&& depth >= 5 * ONE_PLY
&& abs(beta) < VALUE_MATE_IN_MAX_PLY)
{
Value rbeta = std::min(beta + 200, VALUE_INFINITE);
Depth rdepth = depth - 4 * ONE_PLY;
assert(rdepth >= ONE_PLY);
assert((ss-1)->currentMove != MOVE_NONE);
assert((ss-1)->currentMove != MOVE_NULL);
MovePicker mp(pos, ttMove, rbeta - ss->staticEval);
while ((move = mp.next_move()) != MOVE_NONE)
if (pos.legal(move))
{
ss->currentMove = move;
Use per-thread counterMoveHistory Drops a scalability bottleneck due to memory contention of a single shared table across threads. The effect starts to be sensible with a high number of threads. Specifically we have a small regression with 7 threads both at 60 and 180 seconds TC: 10000 @ 60+0.6 th 7 ELO: -2.46 +-3.2 (95%) LOS: 6.5% Total: 9896 W: 1037 L: 1107 D: 7752 5000 @ 180+0.6 th 7 ELO: -1.95 +-4.1 (95%) LOS: 17.7% Total: 5000 W: 444 L: 472 D: 4084 We have a regression because counterMoveHistory table is quite big and it takes time for a single thread to fill it. Sharing the table yields to a higher fill rate and better quality of moves and up to 7 threads the benefits of sharing more then compensate the loss in speed due to contention. Interestingly even with a 3X longer TC, so with more time for the single thread to catch up, the improvment is quite limited and below noise level. It seems we really need much longer TC to saturate the table. When we move to high threads number it's another story: 5000 @ 60+0.6 th 22 ELO: 3.49 +-4.3 (95%) LOS: 94.6% Total: 4880 W: 490 L: 441 D: 3949 2000 @ 60+0.6 th 32 ELO: 8.34 +-6.9 (95%) LOS: 99.1% Total: 2000 W: 229 L: 181 D: 1590 As expected the speed-up more than compensates the filling rate, and we expect that with tournament TC, where single thread is able to saturate the table, the difference will be even stronger. For instance for TCEC 9 super-final time control will be 180 minutes + 15 seconds and this scalability improvement seems definitely the way to go. So, summarizing: GOOD: Measured big improvement in high core scenario Suitable for TCEC 9 superfinal (big hardware, very long TC) Consistent and natural patch that extends to counterMoveHistory what we already do for remaining history tables, that are all per-thread Non functional change for the common case of a single core Very simple (just 6 lines modified, no added ones) BAD: Small regression (within 2-3 ELO) with few threads and short TC bench: 5341477
2016-08-29 01:11:20 -06:00
ss->counterMoves = &thisThread->counterMoveHistory[pos.moved_piece(move)][to_sq(move)];
pos.do_move(move, st, pos.gives_check(move));
value = -search<NonPV>(pos, ss+1, -rbeta, -rbeta+1, rdepth, !cutNode);
pos.undo_move(move);
if (value >= rbeta)
return value;
}
}
// Step 10. Internal iterative deepening (skipped when in check)
if ( depth >= 6 * ONE_PLY
&& !ttMove
&& (PvNode || ss->staticEval + 256 >= beta))
{
Depth d = (3 * depth / (4 * ONE_PLY) - 2) * ONE_PLY;
ss->skipEarlyPruning = true;
search<NT>(pos, ss, alpha, beta, d, cutNode);
ss->skipEarlyPruning = false;
tte = TT.probe(posKey, ttHit);
ttMove = ttHit ? tte->move() : MOVE_NONE;
2008-08-31 23:59:13 -06:00
}
moves_loop: // When in check search starts from here
const CounterMoveStats* cmh = (ss-1)->counterMoves;
const CounterMoveStats* fmh = (ss-2)->counterMoves;
const CounterMoveStats* fmh2 = (ss-4)->counterMoves;
MovePicker mp(pos, ttMove, depth, ss);
value = bestValue; // Workaround a bogus 'uninitialized' warning under gcc
improving = ss->staticEval >= (ss-2)->staticEval
/* || ss->staticEval == VALUE_NONE Already implicit in the previous condition */
||(ss-2)->staticEval == VALUE_NONE;
singularExtensionNode = !rootNode
&& depth >= 8 * ONE_PLY
&& ttMove != MOVE_NONE
&& ttValue != VALUE_NONE
&& !excludedMove // Recursive singular search is not allowed
&& (tte->bound() & BOUND_LOWER)
&& tte->depth() >= depth - 3 * ONE_PLY;
// Step 11. Loop through moves
// Loop through all pseudo-legal moves until no moves remain or a beta cutoff occurs
while ((move = mp.next_move()) != MOVE_NONE)
{
assert(is_ok(move));
if (move == excludedMove)
continue;
// At root obey the "searchmoves" option and skip moves not listed in Root
// Move List. As a consequence any illegal move is also skipped. In MultiPV
// mode we also skip PV moves which have been already searched.
if (rootNode && !std::count(thisThread->rootMoves.begin() + thisThread->PVIdx,
thisThread->rootMoves.end(), move))
continue;
ss->moveCount = ++moveCount;
if (rootNode && thisThread == Threads.main() && Time.elapsed() > 3000)
sync_cout << "info depth " << depth / ONE_PLY
<< " currmove " << UCI::move(move, pos.is_chess960())
<< " currmovenumber " << moveCount + thisThread->PVIdx << sync_endl;
if (PvNode)
(ss+1)->pv = nullptr;
extension = DEPTH_ZERO;
captureOrPromotion = pos.capture_or_promotion(move);
moved_piece = pos.moved_piece(move);
givesCheck = type_of(move) == NORMAL && !pos.discovered_check_candidates()
? pos.check_squares(type_of(pos.piece_on(from_sq(move)))) & to_sq(move)
: pos.gives_check(move);
moveCountPruning = depth < 16 * ONE_PLY
&& moveCount >= FutilityMoveCounts[improving][depth / ONE_PLY];
// Step 12. Extend checks
if ( givesCheck
&& !moveCountPruning
&& pos.see_sign(move) >= VALUE_ZERO)
extension = ONE_PLY;
2008-08-31 23:59:13 -06:00
// Singular extension search. If all moves but one fail low on a search of
// (alpha-s, beta-s), and just one fails high on (alpha, beta), then that move
// is singular and should be extended. To verify this we do a reduced search
// on all the other moves but the ttMove and if the result is lower than
// ttValue minus a margin then we extend the ttMove.
if ( singularExtensionNode
&& move == ttMove
&& !extension
&& pos.legal(move))
{
Value rBeta = std::max(ttValue - 2 * depth / ONE_PLY, -VALUE_MATE);
Depth d = (depth / (2 * ONE_PLY)) * ONE_PLY;
ss->excludedMove = move;
ss->skipEarlyPruning = true;
value = search<NonPV>(pos, ss, rBeta - 1, rBeta, d, cutNode);
ss->skipEarlyPruning = false;
ss->excludedMove = MOVE_NONE;
if (value < rBeta)
extension = ONE_PLY;
}
// Update the current move (this must be done after singular extension search)
newDepth = depth - ONE_PLY + extension;
// Step 13. Pruning at shallow depth
if ( !rootNode
&& !inCheck
&& bestValue > VALUE_MATED_IN_MAX_PLY)
{
if ( !captureOrPromotion
&& !givesCheck
&& !pos.advanced_pawn_push(move))
{
// Move count based pruning
if (moveCountPruning)
continue;
2008-08-31 23:59:13 -06:00
// Reduced depth of the next LMR search
int lmrDepth = std::max(newDepth - reduction<PvNode>(improving, depth, moveCount), DEPTH_ZERO) / ONE_PLY;
// Countermoves based pruning
if ( lmrDepth < 3
&& (!cmh || (*cmh )[moved_piece][to_sq(move)] < VALUE_ZERO)
&& (!fmh || (*fmh )[moved_piece][to_sq(move)] < VALUE_ZERO)
&& (!fmh2 || (*fmh2)[moved_piece][to_sq(move)] < VALUE_ZERO || (cmh && fmh)))
continue;
// Futility pruning: parent node
if ( lmrDepth < 7
&& ss->staticEval + 256 + 200 * lmrDepth <= alpha)
continue;
// Prune moves with negative SEE
if ( lmrDepth < 8
&& pos.see_sign(move) < Value(-35 * lmrDepth * lmrDepth))
continue;
}
else if ( depth < 7 * ONE_PLY
&& pos.see_sign(move) < Value(-35 * depth / ONE_PLY * depth / ONE_PLY))
continue;
2008-08-31 23:59:13 -06:00
}
// Speculative prefetch as early as possible
prefetch(TT.first_entry(pos.key_after(move)));
// Check for legality just before making the move
if (!rootNode && !pos.legal(move))
{
ss->moveCount = --moveCount;
continue;
}
ss->currentMove = move;
Use per-thread counterMoveHistory Drops a scalability bottleneck due to memory contention of a single shared table across threads. The effect starts to be sensible with a high number of threads. Specifically we have a small regression with 7 threads both at 60 and 180 seconds TC: 10000 @ 60+0.6 th 7 ELO: -2.46 +-3.2 (95%) LOS: 6.5% Total: 9896 W: 1037 L: 1107 D: 7752 5000 @ 180+0.6 th 7 ELO: -1.95 +-4.1 (95%) LOS: 17.7% Total: 5000 W: 444 L: 472 D: 4084 We have a regression because counterMoveHistory table is quite big and it takes time for a single thread to fill it. Sharing the table yields to a higher fill rate and better quality of moves and up to 7 threads the benefits of sharing more then compensate the loss in speed due to contention. Interestingly even with a 3X longer TC, so with more time for the single thread to catch up, the improvment is quite limited and below noise level. It seems we really need much longer TC to saturate the table. When we move to high threads number it's another story: 5000 @ 60+0.6 th 22 ELO: 3.49 +-4.3 (95%) LOS: 94.6% Total: 4880 W: 490 L: 441 D: 3949 2000 @ 60+0.6 th 32 ELO: 8.34 +-6.9 (95%) LOS: 99.1% Total: 2000 W: 229 L: 181 D: 1590 As expected the speed-up more than compensates the filling rate, and we expect that with tournament TC, where single thread is able to saturate the table, the difference will be even stronger. For instance for TCEC 9 super-final time control will be 180 minutes + 15 seconds and this scalability improvement seems definitely the way to go. So, summarizing: GOOD: Measured big improvement in high core scenario Suitable for TCEC 9 superfinal (big hardware, very long TC) Consistent and natural patch that extends to counterMoveHistory what we already do for remaining history tables, that are all per-thread Non functional change for the common case of a single core Very simple (just 6 lines modified, no added ones) BAD: Small regression (within 2-3 ELO) with few threads and short TC bench: 5341477
2016-08-29 01:11:20 -06:00
ss->counterMoves = &thisThread->counterMoveHistory[moved_piece][to_sq(move)];
// Step 14. Make the move
pos.do_move(move, st, givesCheck);
// Step 15. Reduced depth search (LMR). If the move fails high it will be
// re-searched at full depth.
if ( depth >= 3 * ONE_PLY
&& moveCount > 1
&& (!captureOrPromotion || moveCountPruning))
{
Depth r = reduction<PvNode>(improving, depth, moveCount);
if (captureOrPromotion)
r -= r ? ONE_PLY : DEPTH_ZERO;
else
{
// Increase reduction for cut nodes
if (cutNode)
r += 2 * ONE_PLY;
// Decrease reduction for moves that escape a capture. Filter out
// castling moves, because they are coded as "king captures rook" and
// hence break make_move(). Also use see() instead of see_sign(),
// because the destination square is empty.
else if ( type_of(move) == NORMAL
&& type_of(pos.piece_on(to_sq(move))) != PAWN
&& pos.see(make_move(to_sq(move), from_sq(move))) < VALUE_ZERO)
r -= 2 * ONE_PLY;
// Decrease/increase reduction for moves with a good/bad history
Value val = thisThread->history[moved_piece][to_sq(move)]
+ (cmh ? (*cmh )[moved_piece][to_sq(move)] : VALUE_ZERO)
+ (fmh ? (*fmh )[moved_piece][to_sq(move)] : VALUE_ZERO)
+ (fmh2 ? (*fmh2)[moved_piece][to_sq(move)] : VALUE_ZERO)
+ thisThread->fromTo.get(~pos.side_to_move(), move);
int rHist = (val - 8000) / 20000;
r = std::max(DEPTH_ZERO, (r / ONE_PLY - rHist) * ONE_PLY);
}
Depth d = std::max(newDepth - r, ONE_PLY);
value = -search<NonPV>(pos, ss+1, -(alpha+1), -alpha, d, true);
doFullDepthSearch = (value > alpha && d != newDepth);
}
else
doFullDepthSearch = !PvNode || moveCount > 1;
// Step 16. Full depth search when LMR is skipped or fails high
if (doFullDepthSearch)
value = newDepth < ONE_PLY ?
givesCheck ? -qsearch<NonPV, true>(pos, ss+1, -(alpha+1), -alpha, DEPTH_ZERO)
: -qsearch<NonPV, false>(pos, ss+1, -(alpha+1), -alpha, DEPTH_ZERO)
: - search<NonPV>(pos, ss+1, -(alpha+1), -alpha, newDepth, !cutNode);
// For PV nodes only, do a full PV search on the first move or after a fail
// high (in the latter case search only if value < beta), otherwise let the
// parent node fail low with value <= alpha and try another move.
if (PvNode && (moveCount == 1 || (value > alpha && (rootNode || value < beta))))
{
(ss+1)->pv = pv;
(ss+1)->pv[0] = MOVE_NONE;
value = newDepth < ONE_PLY ?
givesCheck ? -qsearch<PV, true>(pos, ss+1, -beta, -alpha, DEPTH_ZERO)
: -qsearch<PV, false>(pos, ss+1, -beta, -alpha, DEPTH_ZERO)
: - search<PV>(pos, ss+1, -beta, -alpha, newDepth, false);
}
// Step 17. Undo move
pos.undo_move(move);
2008-08-31 23:59:13 -06:00
assert(value > -VALUE_INFINITE && value < VALUE_INFINITE);
// Step 18. Check for a new best move
// Finished searching the move. If a stop occurred, the return value of
// the search cannot be trusted, and we return immediately without
// updating best move, PV and TT.
if (Signals.stop.load(std::memory_order_relaxed))
return VALUE_ZERO;
if (rootNode)
{
RootMove& rm = *std::find(thisThread->rootMoves.begin(),
thisThread->rootMoves.end(), move);
// PV move or new best move ?
if (moveCount == 1 || value > alpha)
{
rm.score = value;
rm.pv.resize(1);
assert((ss+1)->pv);
for (Move* m = (ss+1)->pv; *m != MOVE_NONE; ++m)
rm.pv.push_back(*m);
// We record how often the best move has been changed in each
// iteration. This information is used for time management: When
// the best move changes frequently, we allocate some more time.
if (moveCount > 1 && thisThread == Threads.main())
++static_cast<MainThread*>(thisThread)->bestMoveChanges;
}
else
// All other moves but the PV are set to the lowest value: this is
// not a problem when sorting because the sort is stable and the
// move position in the list is preserved - just the PV is pushed up.
rm.score = -VALUE_INFINITE;
}
if (value > bestValue)
{
bestValue = value;
if (value > alpha)
{
// If there is an easy move for this position, clear it if unstable
if ( PvNode
&& thisThread == Threads.main()
&& EasyMove.get(pos.key())
&& (move != EasyMove.get(pos.key()) || moveCount > 1))
EasyMove.clear();
bestMove = move;
if (PvNode && !rootNode) // Update pv even in fail-high case
update_pv(ss->pv, move, (ss+1)->pv);
if (PvNode && value < beta) // Update alpha! Always alpha < beta
alpha = value;
else
{
assert(value >= beta); // Fail high
break;
}
}
}
if (!captureOrPromotion && move != bestMove && quietCount < 64)
quietsSearched[quietCount++] = move;
2008-08-31 23:59:13 -06:00
}
// The following condition would detect a stop only after move loop has been
// completed. But in this case bestValue is valid because we have fully
// searched our subtree, and we can anyhow save the result in TT.
/*
if (Signals.stop)
return VALUE_DRAW;
*/
// Step 20. Check for mate and stalemate
// All legal moves have been searched and if there are no legal moves, it
// must be a mate or a stalemate. If we are in a singular extension search then
// return a fail low score.
if (!moveCount)
bestValue = excludedMove ? alpha
: inCheck ? mated_in(ss->ply) : DrawValue[pos.side_to_move()];
else if (bestMove)
{
int d = depth / ONE_PLY;
2008-08-31 23:59:13 -06:00
// Quiet best move: update killers, history and countermoves
if (!pos.capture_or_promotion(bestMove))
{
Value bonus = Value(d * d + 2 * d - 2);
update_stats(pos, ss, bestMove, quietsSearched, quietCount, bonus);
}
// Extra penalty for a quiet TT move in previous ply when it gets refuted
if ((ss-1)->moveCount == 1 && !pos.captured_piece())
{
Value penalty = Value(d * d + 4 * d + 1);
Square prevSq = to_sq((ss-1)->currentMove);
update_cm_stats(ss-1, pos.piece_on(prevSq), prevSq, -penalty);
}
}
// Bonus for prior countermove that caused the fail low
else if ( depth >= 3 * ONE_PLY
&& !pos.captured_piece()
&& is_ok((ss-1)->currentMove))
{
int d = depth / ONE_PLY;
Value bonus = Value(d * d + 2 * d - 2);
Square prevSq = to_sq((ss-1)->currentMove);
update_cm_stats(ss-1, pos.piece_on(prevSq), prevSq, bonus);
}
tte->save(posKey, value_to_tt(bestValue, ss->ply),
bestValue >= beta ? BOUND_LOWER :
PvNode && bestMove ? BOUND_EXACT : BOUND_UPPER,
depth, bestMove, ss->staticEval, TT.generation());
assert(bestValue > -VALUE_INFINITE && bestValue < VALUE_INFINITE);
2008-08-31 23:59:13 -06:00
return bestValue;
}
2008-08-31 23:59:13 -06:00
// qsearch() is the quiescence search function, which is called by the main
// search function when the remaining depth is zero (or, to be more precise,
// less than ONE_PLY).
2008-08-31 23:59:13 -06:00
template <NodeType NT, bool InCheck>
Value qsearch(Position& pos, Stack* ss, Value alpha, Value beta, Depth depth) {
2008-08-31 23:59:13 -06:00
const bool PvNode = NT == PV;
assert(InCheck == !!pos.checkers());
assert(alpha >= -VALUE_INFINITE && alpha < beta && beta <= VALUE_INFINITE);
assert(PvNode || (alpha == beta - 1));
assert(depth <= DEPTH_ZERO);
assert(depth / ONE_PLY * ONE_PLY == depth);
2008-08-31 23:59:13 -06:00
Move pv[MAX_PLY+1];
StateInfo st;
TTEntry* tte;
Key posKey;
Move ttMove, move, bestMove;
Value bestValue, value, ttValue, futilityValue, futilityBase, oldAlpha;
bool ttHit, givesCheck, evasionPrunable;
Depth ttDepth;
if (PvNode)
{
oldAlpha = alpha; // To flag BOUND_EXACT when eval above alpha and no available moves
(ss+1)->pv = pv;
ss->pv[0] = MOVE_NONE;
}
ss->currentMove = bestMove = MOVE_NONE;
ss->ply = (ss-1)->ply + 1;
2008-08-31 23:59:13 -06:00
// Check for an instant draw or if the maximum ply has been reached
if (pos.is_draw() || ss->ply >= MAX_PLY)
return ss->ply >= MAX_PLY && !InCheck ? evaluate(pos)
: DrawValue[pos.side_to_move()];
assert(0 <= ss->ply && ss->ply < MAX_PLY);
// Decide whether or not to include checks: this fixes also the type of
// TT entry depth that we are going to use. Note that in qsearch we use
// only two types of depth in TT: DEPTH_QS_CHECKS or DEPTH_QS_NO_CHECKS.
ttDepth = InCheck || depth >= DEPTH_QS_CHECKS ? DEPTH_QS_CHECKS
: DEPTH_QS_NO_CHECKS;
// Transposition table lookup
posKey = pos.key();
tte = TT.probe(posKey, ttHit);
ttMove = ttHit ? tte->move() : MOVE_NONE;
ttValue = ttHit ? value_from_tt(tte->value(), ss->ply) : VALUE_NONE;
if ( !PvNode
&& ttHit
&& tte->depth() >= ttDepth
&& ttValue != VALUE_NONE // Only in case of TT access race
&& (ttValue >= beta ? (tte->bound() & BOUND_LOWER)
: (tte->bound() & BOUND_UPPER)))
{
ss->currentMove = ttMove; // Can be MOVE_NONE
return ttValue;
}
// Evaluate the position statically
if (InCheck)
{
ss->staticEval = VALUE_NONE;
bestValue = futilityBase = -VALUE_INFINITE;
}
else
{
if (ttHit)
{
// Never assume anything on values stored in TT
if ((ss->staticEval = bestValue = tte->eval()) == VALUE_NONE)
ss->staticEval = bestValue = evaluate(pos);
// Can ttValue be used as a better position evaluation?
if (ttValue != VALUE_NONE)
if (tte->bound() & (ttValue > bestValue ? BOUND_LOWER : BOUND_UPPER))
bestValue = ttValue;
}
else
ss->staticEval = bestValue =
(ss-1)->currentMove != MOVE_NULL ? evaluate(pos)
: -(ss-1)->staticEval + 2 * Eval::Tempo;
// Stand pat. Return immediately if static value is at least beta
if (bestValue >= beta)
{
if (!ttHit)
tte->save(pos.key(), value_to_tt(bestValue, ss->ply), BOUND_LOWER,
DEPTH_NONE, MOVE_NONE, ss->staticEval, TT.generation());
return bestValue;
}
if (PvNode && bestValue > alpha)
alpha = bestValue;
futilityBase = bestValue + 128;
}
2008-08-31 23:59:13 -06:00
// Initialize a MovePicker object for the current position, and prepare
// to search the moves. Because the depth is <= 0 here, only captures,
// queen promotions and checks (only if depth >= DEPTH_QS_CHECKS) will
// be generated.
MovePicker mp(pos, ttMove, depth, to_sq((ss-1)->currentMove));
2008-08-31 23:59:13 -06:00
// Loop through the moves until no moves remain or a beta cutoff occurs
while ((move = mp.next_move()) != MOVE_NONE)
{
assert(is_ok(move));
givesCheck = type_of(move) == NORMAL && !pos.discovered_check_candidates()
? pos.check_squares(type_of(pos.piece_on(from_sq(move)))) & to_sq(move)
: pos.gives_check(move);
2008-08-31 23:59:13 -06:00
// Futility pruning
if ( !InCheck
&& !givesCheck
&& futilityBase > -VALUE_KNOWN_WIN
&& !pos.advanced_pawn_push(move))
{
assert(type_of(move) != ENPASSANT); // Due to !pos.advanced_pawn_push
futilityValue = futilityBase + PieceValue[EG][pos.piece_on(to_sq(move))];
if (futilityValue <= alpha)
{
bestValue = std::max(bestValue, futilityValue);
continue;
}
if (futilityBase <= alpha && pos.see(move) <= VALUE_ZERO)
{
bestValue = std::max(bestValue, futilityBase);
continue;
}
2008-08-31 23:59:13 -06:00
}
// Detect non-capture evasions that are candidates to be pruned
evasionPrunable = InCheck
&& bestValue > VALUE_MATED_IN_MAX_PLY
&& !pos.capture(move);
// Don't search moves with negative SEE values
if ( (!InCheck || evasionPrunable)
&& type_of(move) != PROMOTION
&& pos.see_sign(move) < VALUE_ZERO)
continue;
2008-08-31 23:59:13 -06:00
// Speculative prefetch as early as possible
prefetch(TT.first_entry(pos.key_after(move)));
// Check for legality just before making the move
if (!pos.legal(move))
continue;
ss->currentMove = move;
// Make and search the move
pos.do_move(move, st, givesCheck);
value = givesCheck ? -qsearch<NT, true>(pos, ss+1, -beta, -alpha, depth - ONE_PLY)
: -qsearch<NT, false>(pos, ss+1, -beta, -alpha, depth - ONE_PLY);
pos.undo_move(move);
2008-08-31 23:59:13 -06:00
assert(value > -VALUE_INFINITE && value < VALUE_INFINITE);
// Check for a new best move
if (value > bestValue)
{
bestValue = value;
if (value > alpha)
{
if (PvNode) // Update pv even in fail-high case
update_pv(ss->pv, move, (ss+1)->pv);
if (PvNode && value < beta) // Update alpha here!
{
alpha = value;
bestMove = move;
}
else // Fail high
{
tte->save(posKey, value_to_tt(value, ss->ply), BOUND_LOWER,
ttDepth, move, ss->staticEval, TT.generation());
return value;
}
}
}
2008-08-31 23:59:13 -06:00
}
// All legal moves have been searched. A special case: If we're in check
// and no legal moves were found, it is checkmate.
if (InCheck && bestValue == -VALUE_INFINITE)
return mated_in(ss->ply); // Plies to mate from the root
2008-08-31 23:59:13 -06:00
tte->save(posKey, value_to_tt(bestValue, ss->ply),
PvNode && bestValue > oldAlpha ? BOUND_EXACT : BOUND_UPPER,
ttDepth, bestMove, ss->staticEval, TT.generation());
assert(bestValue > -VALUE_INFINITE && bestValue < VALUE_INFINITE);
2008-08-31 23:59:13 -06:00
return bestValue;
}
// value_to_tt() adjusts a mate score from "plies to mate from the root" to
// "plies to mate from the current position". Non-mate scores are unchanged.
// The function is called before storing a value in the transposition table.
Value value_to_tt(Value v, int ply) {
assert(v != VALUE_NONE);
return v >= VALUE_MATE_IN_MAX_PLY ? v + ply
: v <= VALUE_MATED_IN_MAX_PLY ? v - ply : v;
}
// value_from_tt() is the inverse of value_to_tt(): It adjusts a mate score
// from the transposition table (which refers to the plies to mate/be mated
// from current position) to "plies to mate/be mated from the root".
Value value_from_tt(Value v, int ply) {
return v == VALUE_NONE ? VALUE_NONE
: v >= VALUE_MATE_IN_MAX_PLY ? v - ply
: v <= VALUE_MATED_IN_MAX_PLY ? v + ply : v;
}
// update_pv() adds current move and appends child pv[]
void update_pv(Move* pv, Move move, Move* childPv) {
for (*pv++ = move; childPv && *childPv != MOVE_NONE; )
*pv++ = *childPv++;
*pv = MOVE_NONE;
}
// update_cm_stats() updates countermove and follow-up move history
void update_cm_stats(Stack* ss, Piece pc, Square s, Value bonus) {
CounterMoveStats* cmh = (ss-1)->counterMoves;
CounterMoveStats* fmh1 = (ss-2)->counterMoves;
CounterMoveStats* fmh2 = (ss-4)->counterMoves;
if (cmh)
cmh->update(pc, s, bonus);
if (fmh1)
fmh1->update(pc, s, bonus);
if (fmh2)
fmh2->update(pc, s, bonus);
}
// update_stats() updates killers, history, countermove and countermove plus
// follow-up move history when a new quiet best move is found.
void update_stats(const Position& pos, Stack* ss, Move move,
Move* quiets, int quietsCnt, Value bonus) {
if (ss->killers[0] != move)
{
ss->killers[1] = ss->killers[0];
ss->killers[0] = move;
}
Color c = pos.side_to_move();
Thread* thisThread = pos.this_thread();
thisThread->fromTo.update(c, move, bonus);
thisThread->history.update(pos.moved_piece(move), to_sq(move), bonus);
update_cm_stats(ss, pos.moved_piece(move), to_sq(move), bonus);
if ((ss-1)->counterMoves)
{
Square prevSq = to_sq((ss-1)->currentMove);
thisThread->counterMoves.update(pos.piece_on(prevSq), prevSq, move);
}
// Decrease all the other played quiet moves
for (int i = 0; i < quietsCnt; ++i)
{
thisThread->fromTo.update(c, quiets[i], -bonus);
thisThread->history.update(pos.moved_piece(quiets[i]), to_sq(quiets[i]), -bonus);
update_cm_stats(ss, pos.moved_piece(quiets[i]), to_sq(quiets[i]), -bonus);
}
}
// When playing with strength handicap, choose best move among a set of RootMoves
// using a statistical rule dependent on 'level'. Idea by Heinz van Saanen.
Move Skill::pick_best(size_t multiPV) {
const RootMoves& rootMoves = Threads.main()->rootMoves;
static PRNG rng(now()); // PRNG sequence should be non-deterministic
// RootMoves are already sorted by score in descending order
Value topScore = rootMoves[0].score;
int delta = std::min(topScore - rootMoves[multiPV - 1].score, PawnValueMg);
int weakness = 120 - 2 * level;
int maxScore = -VALUE_INFINITE;
// Choose best move. For each move score we add two terms, both dependent on
// weakness. One is deterministic and bigger for weaker levels, and one is
// random. Then we choose the move with the resulting highest score.
for (size_t i = 0; i < multiPV; ++i)
{
// This is our magic formula
int push = ( weakness * int(topScore - rootMoves[i].score)
+ delta * (rng.rand<unsigned>() % weakness)) / 128;
if (rootMoves[i].score + push > maxScore)
{
maxScore = rootMoves[i].score + push;
best = rootMoves[i].pv[0];
}
}
return best;
}
// check_time() is used to print debug info and, more importantly, to detect
// when we are out of available time and thus stop the search.
void check_time() {
static TimePoint lastInfoTime = now();
int elapsed = Time.elapsed();
TimePoint tick = Limits.startTime + elapsed;
if (tick - lastInfoTime >= 1000)
{
lastInfoTime = tick;
dbg_print();
}
// An engine may not stop pondering until told so by the GUI
if (Limits.ponder)
return;
if ( (Limits.use_time_management() && elapsed > Time.maximum() - 10)
|| (Limits.movetime && elapsed >= Limits.movetime)
|| (Limits.nodes && Threads.nodes_searched() >= Limits.nodes))
Signals.stop = true;
}
} // namespace
/// UCI::pv() formats PV information according to the UCI protocol. UCI requires
/// that all (if any) unsearched PV lines are sent using a previous search score.
string UCI::pv(const Position& pos, Depth depth, Value alpha, Value beta) {
std::stringstream ss;
int elapsed = Time.elapsed() + 1;
const RootMoves& rootMoves = pos.this_thread()->rootMoves;
size_t PVIdx = pos.this_thread()->PVIdx;
size_t multiPV = std::min((size_t)Options["MultiPV"], rootMoves.size());
uint64_t nodes_searched = Threads.nodes_searched();
for (size_t i = 0; i < multiPV; ++i)
{
bool updated = (i <= PVIdx);
if (depth == ONE_PLY && !updated)
continue;
Depth d = updated ? depth : depth - ONE_PLY;
Value v = updated ? rootMoves[i].score : rootMoves[i].previousScore;
2015-01-18 00:05:05 -07:00
bool tb = TB::RootInTB && abs(v) < VALUE_MATE - MAX_PLY;
v = tb ? TB::Score : v;
if (ss.rdbuf()->in_avail()) // Not at first line
ss << "\n";
ss << "info"
<< " depth " << d / ONE_PLY
<< " seldepth " << pos.this_thread()->maxPly
<< " multipv " << i + 1
<< " score " << UCI::value(v);
if (!tb && i == PVIdx)
ss << (v >= beta ? " lowerbound" : v <= alpha ? " upperbound" : "");
ss << " nodes " << nodes_searched
<< " nps " << nodes_searched * 1000 / elapsed;
2015-01-31 05:04:15 -07:00
if (elapsed > 1000) // Earlier makes little sense
ss << " hashfull " << TT.hashfull();
2015-01-31 05:04:15 -07:00
ss << " tbhits " << TB::Hits
<< " time " << elapsed
<< " pv";
for (Move m : rootMoves[i].pv)
ss << " " << UCI::move(m, pos.is_chess960());
}
return ss.str();
}
2008-08-31 23:59:13 -06:00
/// RootMove::extract_ponder_from_tt() is called in case we have no ponder move
/// before exiting the search, for instance, in case we stop the search during a
/// fail high at root. We try hard to have a ponder move to return to the GUI,
/// otherwise in case of 'ponder on' we have nothing to think on.
bool RootMove::extract_ponder_from_tt(Position& pos) {
StateInfo st;
bool ttHit;
assert(pv.size() == 1);
if (!pv[0])
return false;
pos.do_move(pv[0], st, pos.gives_check(pv[0]));
TTEntry* tte = TT.probe(pos.key(), ttHit);
if (ttHit)
{
Move m = tte->move(); // Local copy to be SMP safe
if (MoveList<LEGAL>(pos).contains(m))
pv.push_back(m);
}
pos.undo_move(pv[0]);
return pv.size() > 1;
}
void Tablebases::filter_root_moves(Position& pos, Search::RootMoves& rootMoves) {
Hits = 0;
RootInTB = false;
UseRule50 = Options["Syzygy50MoveRule"];
ProbeDepth = Options["SyzygyProbeDepth"] * ONE_PLY;
Cardinality = Options["SyzygyProbeLimit"];
// Skip TB probing when no TB found: !TBLargest -> !TB::Cardinality
if (Cardinality > MaxCardinality)
{
Cardinality = MaxCardinality;
ProbeDepth = DEPTH_ZERO;
}
if (Cardinality < popcount(pos.pieces()) || pos.can_castle(ANY_CASTLING))
return;
// If the current root position is in the tablebases, then RootMoves
// contains only moves that preserve the draw or the win.
RootInTB = root_probe(pos, rootMoves, TB::Score);
if (RootInTB)
Cardinality = 0; // Do not probe tablebases during the search
else // If DTZ tables are missing, use WDL tables as a fallback
{
// Filter out moves that do not preserve the draw or the win.
RootInTB = root_probe_wdl(pos, rootMoves, TB::Score);
// Only probe during search if winning
if (RootInTB && TB::Score <= VALUE_DRAW)
Cardinality = 0;
}
if (RootInTB)
{
Hits = rootMoves.size();
if (!UseRule50)
TB::Score = TB::Score > VALUE_DRAW ? VALUE_MATE - MAX_PLY - 1
: TB::Score < VALUE_DRAW ? -VALUE_MATE + MAX_PLY + 1
: VALUE_DRAW;
}
}