kate/BadFish
1
0
Fork 0
mirror of https://github.com/sockspls/badfish synced 2025-07-11 19:49:14 +00:00
Code Issues Releases Wiki Activity

Merge branch 'master' into simplify_eval

Browse source
This commit is contained in:
Gary Linscott 2013-02-11 10:26:25 -05:00
commit 30c2e3828a
14 changed files with 306 additions and 368 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

23
src/endgame.cpp
View file

@ -471,6 +471,29 @@ ScaleFactor Endgame<KBPsK>::operator()(const Position& pos) const {
return SCALE_FACTOR_DRAW;
}
}
// All pawns on same B or G file? Then potential draw
if ( (pawnFile == FILE_B || pawnFile == FILE_G)
&& !(pos.pieces(PAWN) & ~file_bb(pawnFile))
&& pos.non_pawn_material(weakerSide) == 0
&& pos.piece_count(weakerSide, PAWN) >= 1)
{
// Get weaker pawn closest to opponent's queening square
Bitboard wkPawns = pos.pieces(weakerSide, PAWN);
Square weakerPawnSq = strongerSide == WHITE ? msb(wkPawns) : lsb(wkPawns);
Square strongerKingSq = pos.king_square(strongerSide);
Square weakerKingSq = pos.king_square(weakerSide);
Square bishopSq = pos.piece_list(strongerSide, BISHOP)[0];
// Draw if weaker pawn is on rank 7, bishop can't attack the pawn, and
// weaker king can stop opposing opponent's king from penetrating.
if ( relative_rank(strongerSide, weakerPawnSq) == RANK_7
&& opposite_colors(bishopSq, weakerPawnSq)
&& square_distance(weakerPawnSq, weakerKingSq) <= square_distance(weakerPawnSq, strongerKingSq))
return SCALE_FACTOR_DRAW;
}
return SCALE_FACTOR_NONE;
}

72
src/history.h
View file

@ -1,72 +0,0 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2012 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
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,
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.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(HISTORY_H_INCLUDED)
#define HISTORY_H_INCLUDED
#include <algorithm>
#include <cstring>
#include "types.h"
/// The History class stores statistics about how often different moves
/// have been successful or unsuccessful during the current search. These
/// statistics are used for reduction and move ordering decisions. History
/// entries are stored according only to moving piece and destination square,
/// in particular two moves with different origin but same destination and
/// same piece will be considered identical.
class History {
public:
void clear();
Value value(Piece p, Square to) const;
void add(Piece p, Square to, Value bonus);
Value gain(Piece p, Square to) const;
void update_gain(Piece p, Square to, Value g);
static const Value MaxValue = Value(2000);
private:
Value history[PIECE_NB][SQUARE_NB];
Value maxGains[PIECE_NB][SQUARE_NB];
};
inline void History::clear() {
memset(history, 0, 16 * 64 * sizeof(Value));
memset(maxGains, 0, 16 * 64 * sizeof(Value));
}
inline Value History::value(Piece p, Square to) const {
return history[p][to];
}
inline void History::add(Piece p, Square to, Value bonus) {
if (abs(history[p][to] + bonus) < MaxValue) history[p][to] += bonus;
}
inline Value History::gain(Piece p, Square to) const {
return maxGains[p][to];
}
inline void History::update_gain(Piece p, Square to, Value g) {
maxGains[p][to] = std::max(g, maxGains[p][to] - 1);
}
#endif // !defined(HISTORY_H_INCLUDED)

83
src/movepick.cpp
View file

@ -18,10 +18,8 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <algorithm>
#include <cassert>
#include "movegen.h"
#include "movepick.h"
#include "thread.h"
@ -37,6 +35,20 @@ namespace {
STOP
};
// Our insertion sort, guaranteed to be stable, as is needed
void insertion_sort(MoveStack* begin, MoveStack* end)
{
MoveStack tmp, *p, *q;
for (p = begin + 1; p < end; ++p)
{
tmp = *p;
for (q = p; q != begin && *(q-1) < tmp; --q)
*q = *(q-1);
*q = tmp;
}
}
// Unary predicate used by std::partition to split positive scores from remaining
// ones so to sort separately the two sets, and with the second sort delayed.
inline bool has_positive_score(const MoveStack& ms) { return ms.score > 0; }
@ -59,7 +71,7 @@ namespace {
/// move ordering is at the current node.
MovePicker::MovePicker(const Position& p, Move ttm, Depth d, const History& h,
Search::Stack* s, Value beta) : pos(p), H(h), depth(d) {
Search::Stack* s, Value beta) : pos(p), Hist(h), depth(d) {
assert(d > DEPTH_ZERO);
@ -92,7 +104,7 @@ MovePicker::MovePicker(const Position& p, Move ttm, Depth d, const History& h,
}
MovePicker::MovePicker(const Position& p, Move ttm, Depth d, const History& h,
Square sq) : pos(p), H(h), cur(moves), end(moves) {
Square sq) : pos(p), Hist(h), cur(moves), end(moves) {
assert(d <= DEPTH_ZERO);
@ -124,7 +136,7 @@ MovePicker::MovePicker(const Position& p, Move ttm, Depth d, const History& h,
}
MovePicker::MovePicker(const Position& p, Move ttm, const History& h, PieceType pt)
: pos(p), H(h), cur(moves), end(moves) {
: pos(p), Hist(h), cur(moves), end(moves) {
assert(!pos.checkers());
@ -141,12 +153,10 @@ MovePicker::MovePicker(const Position& p, Move ttm, const History& h, PieceType
}
/// MovePicker::score_captures(), MovePicker::score_noncaptures() and
/// MovePicker::score_evasions() assign a numerical move ordering score
/// to each move in a move list. The moves with highest scores will be
/// picked first by next_move().
void MovePicker::score_captures() {
/// score() assign a numerical move ordering score to each move in a move list.
/// The moves with highest scores will be picked first.
template<>
void MovePicker::score<CAPTURES>() {
// Winning and equal captures in the main search are ordered by MVV/LVA.
// Suprisingly, this appears to perform slightly better than SEE based
// move ordering. The reason is probably that in a position with a winning
@ -169,47 +179,50 @@ void MovePicker::score_captures() {
- type_of(pos.piece_moved(m));
if (type_of(m) == PROMOTION)
it->score += PieceValue[MG][promotion_type(m)];
it->score += PieceValue[MG][promotion_type(m)] - PieceValue[MG][PAWN];
else if (type_of(m) == ENPASSANT)
it->score += PieceValue[MG][PAWN];
}
}
void MovePicker::score_noncaptures() {
template<>
void MovePicker::score<QUIETS>() {
Move m;
for (MoveStack* it = moves; it != end; ++it)
{
m = it->move;
it->score = H.value(pos.piece_moved(m), to_sq(m));
it->score = Hist[pos.piece_moved(m)][to_sq(m)];
}
}
void MovePicker::score_evasions() {
template<>
void MovePicker::score<EVASIONS>() {
// Try good captures ordered by MVV/LVA, then non-captures if destination square
// is not under attack, ordered by history value, then bad-captures and quiet
// moves with a negative SEE. This last group is ordered by the SEE score.
Move m;
int seeScore;
if (end < moves + 2)
return;
for (MoveStack* it = moves; it != end; ++it)
{
m = it->move;
if ((seeScore = pos.see_sign(m)) < 0)
it->score = seeScore - History::MaxValue; // Be sure we are at the bottom
it->score = seeScore - History::Max; // At the bottom
else if (pos.is_capture(m))
it->score = PieceValue[MG][pos.piece_on(to_sq(m))]
- type_of(pos.piece_moved(m)) + History::MaxValue;
- type_of(pos.piece_moved(m)) + History::Max;
else
it->score = H.value(pos.piece_moved(m), to_sq(m));
it->score = Hist[pos.piece_moved(m)][to_sq(m)];
}
}
/// MovePicker::generate_next() generates, scores and sorts the next bunch of moves,
/// when there are no more moves to try for the current phase.
/// generate_next() generates, scores and sorts the next bunch of moves, when
/// there are no more moves to try for the current phase.
void MovePicker::generate_next() {
@ -219,7 +232,7 @@ void MovePicker::generate_next() {
case CAPTURES_S1: case CAPTURES_S3: case CAPTURES_S4: case CAPTURES_S5: case CAPTURES_S6:
end = generate<CAPTURES>(pos, moves);
score_captures();
score<CAPTURES>();
return;
case KILLERS_S1:
@ -229,16 +242,16 @@ void MovePicker::generate_next() {
case QUIETS_1_S1:
endQuiets = end = generate<QUIETS>(pos, moves);
score_noncaptures();
score<QUIETS>();
end = std::partition(cur, end, has_positive_score);
sort<MoveStack>(cur, end);
insertion_sort(cur, end);
return;
case QUIETS_2_S1:
cur = end;
end = endQuiets;
if (depth >= 3 * ONE_PLY)
sort<MoveStack>(cur, end);
insertion_sort(cur, end);
return;
case BAD_CAPTURES_S1:
@ -249,7 +262,8 @@ void MovePicker::generate_next() {
case EVASIONS_S2:
end = generate<EVASIONS>(pos, moves);
score_evasions();
if (end > moves + 1)
score<EVASIONS>();
return;
case QUIET_CHECKS_S3:
@ -268,11 +282,10 @@ void MovePicker::generate_next() {
}
/// MovePicker::next_move() is the most important method of the MovePicker class.
/// It returns a new pseudo legal move every time it is called, until there
/// are no more moves left. It picks the move with the biggest score from a list
/// of generated moves taking care not to return the tt move if has already been
/// searched previously.
/// next_move() is the most important method of the MovePicker class. It returns
/// a new pseudo legal move every time is called, until there are no more moves
/// left. It picks the move with the biggest score from a list of generated moves
/// taking care not returning the ttMove if has already been searched previously.
template<>
Move MovePicker::next_move<false>() {
@ -359,6 +372,6 @@ Move MovePicker::next_move<false>() {
/// Version of next_move() to use at split point nodes where the move is grabbed
/// from the split point's shared MovePicker object. This function is not thread
/// safe so should be lock protected by the caller.
/// safe so must be lock protected by the caller.
template<>
Move MovePicker::next_move<true>() { return ss->sp->mp->next_move<false>(); }
Move MovePicker::next_move<true>() { return ss->sp->movePicker->next_move<false>(); }

44
src/movepick.h
View file

@ -20,12 +20,48 @@
#if !defined MOVEPICK_H_INCLUDED
#define MOVEPICK_H_INCLUDED
#include "history.h"
#include <algorithm> // For std::max
#include <cstring> // For memset
#include "movegen.h"
#include "position.h"
#include "search.h"
#include "types.h"
/// The Stats struct stores moves statistics. According to the template parameter
/// the class can store both History and Gains type statistics. History records
/// how often different moves have been successful or unsuccessful during the
/// current search and is used for reduction and move ordering decisions. Gains
/// records the move's best evaluation gain from one ply to the next and is used
/// for pruning decisions. Entries are stored according only to moving piece and
/// destination square, in particular two moves with different origin but same
/// destination and same piece will be considered identical.
template<bool Gain>
struct Stats {
static const Value Max = Value(2000);
const Value* operator[](Piece p) const { return &table[p][0]; }
void clear() { memset(table, 0, sizeof(table)); }
void update(Piece p, Square to, Value v) {
if (Gain)
table[p][to] = std::max(v, table[p][to] - 1);
else if (abs(table[p][to] + v) < Max)
table[p][to] += v;
}
private:
Value table[PIECE_NB][SQUARE_NB];
};
typedef Stats<false> History;
typedef Stats<true> Gains;
/// MovePicker class is used to pick one pseudo legal move at a time from the
/// current position. The most important method is next_move(), which returns a
/// new pseudo legal move each time it is called, until there are no moves left,
@ -44,13 +80,11 @@ public:
template<bool SpNode> Move next_move();
private:
void score_captures();
void score_noncaptures();
void score_evasions();
template<GenType> void score();
void generate_next();
const Position& pos;
const History& H;
const History& Hist;
Search::Stack* ss;
Depth depth;
Move ttMove;

31
src/notation.cpp
View file

@ -110,8 +110,7 @@ const string move_to_san(Position& pos, Move m) {
assert(MoveList<LEGAL>(pos).contains(m));
Bitboard attackers;
bool ambiguousMove, ambiguousFile, ambiguousRank;
Bitboard others, b;
string san;
Color us = pos.side_to_move();
Square from = from_sq(m);
@ -127,31 +126,23 @@ const string move_to_san(Position& pos, Move m) {
{
san = PieceToChar[WHITE][pt]; // Upper case
// Disambiguation if we have more then one piece with destination 'to'
// note that for pawns is not needed because starting file is explicit.
ambiguousMove = ambiguousFile = ambiguousRank = false;
// Disambiguation if we have more then one piece of type 'pt' that can
// reach 'to' with a legal move.
others = b = (pos.attacks_from(pc, to) & pos.pieces(us, pt)) ^ from;
attackers = (pos.attacks_from(pc, to) & pos.pieces(us, pt)) ^ from;
while (attackers)
while (b)
{
Square sq = pop_lsb(&attackers);
// If the move is illegal, the piece is not included in the sub-set
if (!pos.pl_move_is_legal(make_move(sq, to), pos.pinned_pieces()))
continue;
ambiguousFile |= file_of(sq) == file_of(from);
ambiguousRank |= rank_of(sq) == rank_of(from);
ambiguousMove = true;
Move move = make_move(pop_lsb(&b), to);
if (!pos.pl_move_is_legal(move, pos.pinned_pieces()))
others ^= from_sq(move);
}
if (ambiguousMove)
if (others)
{
if (!ambiguousFile)
if (!(others & file_bb(from)))
san += file_to_char(file_of(from));
else if (!ambiguousRank)
else if (!(others & rank_bb(from)))
san += rank_to_char(rank_of(from));
else

2
src/position.cpp
View file

@ -802,7 +802,7 @@ void Position::do_move(Move m, StateInfo& newSt, const CheckInfo& ci, bool moveI
// Update piece list, move the last piece at index[capsq] position and
// shrink the list.
//
// WARNING: This is a not revresible operation. When we will reinsert the
// WARNING: This is a not reversible operation. When we will reinsert the
// captured piece in undo_move() we will put it at the end of the list and
// not in its original place, it means index[] and pieceList[] are not
// guaranteed to be invariant to a do_move() + undo_move() sequence.

111
src/search.cpp
View file

@ -26,7 +26,6 @@
#include "book.h"
#include "evaluate.h"
#include "history.h"
#include "movegen.h"
#include "movepick.h"
#include "notation.h"
@ -87,7 +86,8 @@ namespace {
TimeManager TimeMgr;
int BestMoveChanges;
Value DrawValue[COLOR_NB];
History H;
History Hist;
Gains Gain;
template <NodeType NT>
Value search(Position& pos, Stack* ss, Value alpha, Value beta, Depth depth);
@ -98,9 +98,9 @@ namespace {
void id_loop(Position& pos);
Value value_to_tt(Value v, int ply);
Value value_from_tt(Value v, int ply);
bool check_is_dangerous(Position& pos, Move move, Value futilityBase, Value beta);
bool allows_move(const Position& pos, Move first, Move second);
bool prevents_move(const Position& pos, Move first, Move second);
bool check_is_dangerous(const Position& pos, Move move, Value futilityBase, Value beta);
bool allows(const Position& pos, Move first, Move second);
bool refutes(const Position& pos, Move first, Move second);
string uci_pv(const Position& pos, int depth, Value alpha, Value beta);
struct Skill {
@ -229,22 +229,22 @@ void Search::think() {
// Reset the threads, still sleeping: will be wake up at split time
for (size_t i = 0; i < Threads.size(); i++)
Threads[i].maxPly = 0;
Threads[i]->maxPly = 0;
Threads.sleepWhileIdle = Options["Use Sleeping Threads"];
// Set best timer interval to avoid lagging under time pressure. Timer is
// used to check for remaining available thinking time.
Threads.timer_thread()->msec =
Threads.timer->msec =
Limits.use_time_management() ? std::min(100, std::max(TimeMgr.available_time() / 16, TimerResolution)) :
Limits.nodes ? 2 * TimerResolution
: 100;
Threads.timer_thread()->notify_one(); // Wake up the recurring timer
Threads.timer->notify_one(); // Wake up the recurring timer
id_loop(RootPos); // Let's start searching !
Threads.timer_thread()->msec = 0; // Stop the timer
Threads.timer->msec = 0; // Stop the timer
Threads.sleepWhileIdle = true; // Send idle threads to sleep
if (Options["Use Search Log"])
@ -300,7 +300,8 @@ namespace {
bestValue = delta = -VALUE_INFINITE;
ss->currentMove = MOVE_NULL; // Hack to skip update gains
TT.new_search();
H.clear();
Hist.clear();
Gain.clear();
PVSize = Options["MultiPV"];
Skill skill(Options["Skill Level"]);
@ -353,7 +354,7 @@ namespace {
// we want to keep the same order for all the moves but the new
// PV that goes to the front. Note that in case of MultiPV search
// the already searched PV lines are preserved.
sort<RootMove>(RootMoves.begin() + PVIdx, RootMoves.end());
std::stable_sort(RootMoves.begin() + PVIdx, RootMoves.end());
// Write PV back to transposition table in case the relevant
// entries have been overwritten during the search.
@ -398,7 +399,8 @@ namespace {
}
// Sort the PV lines searched so far and update the GUI
sort<RootMove>(RootMoves.begin(), RootMoves.begin() + PVIdx + 1);
std::stable_sort(RootMoves.begin(), RootMoves.begin() + PVIdx + 1);
if (PVIdx + 1 == PVSize || Time::now() - SearchTime > 3000)
sync_cout << uci_pv(pos, depth, alpha, beta) << sync_endl;
}
@ -592,8 +594,8 @@ namespace {
else if (tte)
{
// Never assume anything on values stored in TT
if ( (ss->staticEval = eval = tte->static_value()) == VALUE_NONE
||(ss->evalMargin = tte->static_value_margin()) == VALUE_NONE)
if ( (ss->staticEval = eval = tte->eval_value()) == VALUE_NONE
||(ss->evalMargin = tte->eval_margin()) == VALUE_NONE)
eval = ss->staticEval = evaluate(pos, ss->evalMargin);
// Can ttValue be used as a better position evaluation?
@ -618,7 +620,7 @@ namespace {
&& type_of(move) == NORMAL)
{
Square to = to_sq(move);
H.update_gain(pos.piece_on(to), to, -(ss-1)->staticEval - ss->staticEval);
Gain.update(pos.piece_on(to), to, -(ss-1)->staticEval - ss->staticEval);
}
// Step 6. Razoring (is omitted in PV nodes)
@ -705,7 +707,7 @@ namespace {
if ( depth < 5 * ONE_PLY
&& (ss-1)->reduction
&& threatMove != MOVE_NONE
&& allows_move(pos, (ss-1)->currentMove, threatMove))
&& allows(pos, (ss-1)->currentMove, threatMove))
return beta - 1;
}
}
@ -728,7 +730,7 @@ namespace {
assert((ss-1)->currentMove != MOVE_NONE);
assert((ss-1)->currentMove != MOVE_NULL);
MovePicker mp(pos, ttMove, H, pos.captured_piece_type());
MovePicker mp(pos, ttMove, Hist, pos.captured_piece_type());
CheckInfo ci(pos);
while ((move = mp.next_move<false>()) != MOVE_NONE)
@ -760,7 +762,7 @@ namespace {
split_point_start: // At split points actual search starts from here
MovePicker mp(pos, ttMove, depth, H, ss, PvNode ? -VALUE_INFINITE : beta);
MovePicker mp(pos, ttMove, depth, Hist, ss, PvNode ? -VALUE_INFINITE : beta);
CheckInfo ci(pos);
value = bestValue; // Workaround a bogus 'uninitialized' warning under gcc
singularExtensionNode = !RootNode
@ -866,7 +868,7 @@ split_point_start: // At split points actual search starts from here
// Move count based pruning
if ( depth < 16 * ONE_PLY
&& moveCount >= FutilityMoveCounts[depth]
&& (!threatMove || !prevents_move(pos, move, threatMove)))
&& (!threatMove || !refutes(pos, move, threatMove)))
{
if (SpNode)
sp->mutex.lock();
@ -879,7 +881,7 @@ split_point_start: // At split points actual search starts from here
// but fixing this made program slightly weaker.
Depth predictedDepth = newDepth - reduction<PvNode>(depth, moveCount);
futilityValue = ss->staticEval + ss->evalMargin + futility_margin(predictedDepth, moveCount)
+ H.gain(pos.piece_moved(move), to_sq(move));
+ Gain[pos.piece_moved(move)][to_sq(move)];
if (futilityValue < beta)
{
@ -921,8 +923,9 @@ split_point_start: // At split points actual search starts from here
&& !pvMove
&& !captureOrPromotion
&& !dangerous
&& ss->killers[0] != move
&& ss->killers[1] != move)
&& move != ttMove
&& move != ss->killers[0]
&& move != ss->killers[1])
{
ss->reduction = reduction<PvNode>(depth, moveCount);
Depth d = std::max(newDepth - ss->reduction, ONE_PLY);
@ -1022,13 +1025,13 @@ split_point_start: // At split points actual search starts from here
// Step 19. Check for splitting the search
if ( !SpNode
&& depth >= Threads.minimumSplitDepth
&& Threads.slave_available(thisThread)
&& Threads.available_slave(thisThread)
&& thisThread->splitPointsSize < MAX_SPLITPOINTS_PER_THREAD)
{
assert(bestValue < beta);
bestValue = Threads.split<FakeSplit>(pos, ss, alpha, beta, bestValue, &bestMove,
depth, threatMove, moveCount, mp, NT);
thisThread->split<FakeSplit>(pos, ss, alpha, beta, &bestValue, &bestMove,
depth, threatMove, moveCount, &mp, NT);
if (bestValue >= beta)
break;
}
@ -1071,13 +1074,13 @@ split_point_start: // At split points actual search starts from here
// Increase history value of the cut-off move
Value bonus = Value(int(depth) * int(depth));
H.add(pos.piece_moved(bestMove), to_sq(bestMove), bonus);
Hist.update(pos.piece_moved(bestMove), to_sq(bestMove), bonus);
// Decrease history of all the other played non-capture moves
for (int i = 0; i < playedMoveCount - 1; i++)
{
Move m = movesSearched[i];
H.add(pos.piece_moved(m), to_sq(m), -bonus);
Hist.update(pos.piece_moved(m), to_sq(m), -bonus);
}
}
}
@ -1161,8 +1164,8 @@ split_point_start: // At split points actual search starts from here
if (tte)
{
// Never assume anything on values stored in TT
if ( (ss->staticEval = bestValue = tte->static_value()) == VALUE_NONE
||(ss->evalMargin = tte->static_value_margin()) == VALUE_NONE)
if ( (ss->staticEval = bestValue = tte->eval_value()) == VALUE_NONE
||(ss->evalMargin = tte->eval_margin()) == VALUE_NONE)
ss->staticEval = bestValue = evaluate(pos, ss->evalMargin);
}
else
@ -1189,7 +1192,7 @@ split_point_start: // At split points actual search starts from here
// 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, H, to_sq((ss-1)->currentMove));
MovePicker mp(pos, ttMove, depth, Hist, to_sq((ss-1)->currentMove));
CheckInfo ci(pos);
// Loop through the moves until no moves remain or a beta cutoff occurs
@ -1332,7 +1335,7 @@ split_point_start: // At split points actual search starts from here
// check_is_dangerous() tests if a checking move can be pruned in qsearch()
bool check_is_dangerous(Position& pos, Move move, Value futilityBase, Value beta)
bool check_is_dangerous(const Position& pos, Move move, Value futilityBase, Value beta)
{
Piece pc = pos.piece_moved(move);
Square from = from_sq(move);
@ -1366,12 +1369,12 @@ split_point_start: // At split points actual search starts from here
}
// allows_move() tests whether the move at previous ply (first) somehow makes a
// second move possible, for instance if the moving piece is the same in both
// moves. Normally the second move is the threat move (the best move returned
// allows() tests whether the 'first' move at previous ply somehow makes the
// 'second' move possible, for instance if the moving piece is the same in
// both moves. Normally the second move is the threat (the best move returned
// from a null search that fails low).
bool allows_move(const Position& pos, Move first, Move second) {
bool allows(const Position& pos, Move first, Move second) {
assert(is_ok(first));
assert(is_ok(second));
@ -1407,12 +1410,11 @@ split_point_start: // At split points actual search starts from here
}
// prevents_move() tests whether a move (first) is able to defend against an
// opponent's move (second). In this case will not be pruned. Normally the
// second move is the threat move (the best move returned from a null search
// that fails low).
// refutes() tests whether a 'first' move is able to defend against a 'second'
// opponent's move. In this case will not be pruned. Normally the second move
// is the threat (the best move returned from a null search that fails low).
bool prevents_move(const Position& pos, Move first, Move second) {
bool refutes(const Position& pos, Move first, Move second) {
assert(is_ok(first));
assert(is_ok(second));
@ -1511,8 +1513,8 @@ split_point_start: // At split points actual search starts from here
int selDepth = 0;
for (size_t i = 0; i < Threads.size(); i++)
if (Threads[i].maxPly > selDepth)
selDepth = Threads[i].maxPly;
if (Threads[i]->maxPly > selDepth)
selDepth = Threads[i]->maxPly;
for (size_t i = 0; i < uciPVSize; i++)
{
@ -1614,7 +1616,7 @@ void Thread::idle_loop() {
// at the thread creation. So it means we are the split point's master.
const SplitPoint* this_sp = splitPointsSize ? activeSplitPoint : NULL;
assert(!this_sp || (this_sp->master == this && searching));
assert(!this_sp || (this_sp->masterThread == this && searching));
// If this thread is the master of a split point and all slaves have finished
// their work at this split point, return from the idle loop.
@ -1670,9 +1672,9 @@ void Thread::idle_loop() {
sp->mutex.lock();
assert(sp->slavesPositions[idx] == NULL);
assert(activePosition == NULL);
sp->slavesPositions[idx] = &pos;
activePosition = &pos;
switch (sp->nodeType) {
case Root:
@ -1691,18 +1693,18 @@ void Thread::idle_loop() {
assert(searching);
searching = false;
sp->slavesPositions[idx] = NULL;
activePosition = NULL;
sp->slavesMask &= ~(1ULL << idx);
sp->nodes += pos.nodes_searched();
// Wake up master thread so to allow it to return from the idle loop
// in case we are the last slave of the split point.
if ( Threads.sleepWhileIdle
&& this != sp->master
&& this != sp->masterThread
&& !sp->slavesMask)
{
assert(!sp->master->searching);
sp->master->notify_one();
assert(!sp->masterThread->searching);
sp->masterThread->notify_one();
}
// After releasing the lock we cannot access anymore any SplitPoint
@ -1740,11 +1742,11 @@ void check_time() {
nodes = RootPos.nodes_searched();
// Loop across all split points and sum accumulated SplitPoint nodes plus
// all the currently active slaves positions.
// all the currently active positions nodes.
for (size_t i = 0; i < Threads.size(); i++)
for (int j = 0; j < Threads[i].splitPointsSize; j++)
for (int j = 0; j < Threads[i]->splitPointsSize; j++)
{
SplitPoint& sp = Threads[i].splitPoints[j];
SplitPoint& sp = Threads[i]->splitPoints[j];
sp.mutex.lock();
@ -1752,8 +1754,9 @@ void check_time() {
Bitboard sm = sp.slavesMask;
while (sm)
{
Position* pos = sp.slavesPositions[pop_lsb(&sm)];
nodes += pos ? pos->nodes_searched() : 0;
Position* pos = Threads[pop_lsb(&sm)]->activePosition;
if (pos)
nodes += pos->nodes_searched();
}
sp.mutex.unlock();

3
src/search.h
View file

@ -56,12 +56,11 @@ struct Stack {
/// all non-pv moves.
struct RootMove {
RootMove(){} // Needed by sort()
RootMove(Move m) : score(-VALUE_INFINITE), prevScore(-VALUE_INFINITE) {
pv.push_back(m); pv.push_back(MOVE_NONE);
}
bool operator<(const RootMove& m) const { return score < m.score; }
bool operator<(const RootMove& m) const { return score > m.score; } // Ascending sort
bool operator==(const Move& m) const { return pv[0] == m; }
void extract_pv_from_tt(Position& pos);

115
src/thread.cpp
View file

@ -17,6 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <algorithm> // For std::count
#include <cassert>
#include <iostream>
@ -42,11 +43,12 @@ namespace { extern "C" {
// Thread c'tor starts a newly-created thread of execution that will call
// the the virtual function idle_loop(), going immediately to sleep.
Thread::Thread() : splitPoints() {
Thread::Thread() /* : splitPoints() */ { // Value-initialization bug in MSVC
searching = exit = false;
maxPly = splitPointsSize = 0;
activeSplitPoint = NULL;
activePosition = NULL;
idx = Threads.size();
if (!thread_create(handle, start_routine, this))
@ -146,7 +148,7 @@ void Thread::wait_for(volatile const bool& b) {
bool Thread::cutoff_occurred() const {
for (SplitPoint* sp = activeSplitPoint; sp; sp = sp->parent)
for (SplitPoint* sp = activeSplitPoint; sp; sp = sp->parentSplitPoint)
if (sp->cutoff)
return true;
@ -185,7 +187,7 @@ void ThreadPool::init() {
sleepWhileIdle = true;
timer = new TimerThread();
threads.push_back(new MainThread());
push_back(new MainThread());
read_uci_options();
}
@ -196,8 +198,8 @@ void ThreadPool::exit() {
delete timer; // As first because check_time() accesses threads data
for (size_t i = 0; i < threads.size(); i++)
delete threads[i];
for (iterator it = begin(); it != end(); ++it)
delete *it;
}
@ -214,13 +216,13 @@ void ThreadPool::read_uci_options() {
assert(requested > 0);
while (threads.size() < requested)
threads.push_back(new Thread());
while (size() < requested)
push_back(new Thread());
while (threads.size() > requested)
while (size() > requested)
{
delete threads.back();
threads.pop_back();
delete back();
pop_back();
}
}
@ -228,13 +230,13 @@ void ThreadPool::read_uci_options() {
// slave_available() tries to find an idle thread which is available as a slave
// for the thread 'master'.
bool ThreadPool::slave_available(Thread* master) const {
Thread* ThreadPool::available_slave(Thread* master) const {
for (size_t i = 0; i < threads.size(); i++)
if (threads[i]->is_available_to(master))
return true;
for (const_iterator it = begin(); it != end(); ++it)
if ((*it)->is_available_to(master))
return *it;
return false;
return NULL;
}
@ -248,34 +250,31 @@ bool ThreadPool::slave_available(Thread* master) const {
// search() then split() returns.
template <bool Fake>
Value ThreadPool::split(Position& pos, Stack* ss, Value alpha, Value beta,
Value bestValue, Move* bestMove, Depth depth, Move threatMove,
int moveCount, MovePicker& mp, int nodeType) {
void Thread::split(Position& pos, Stack* ss, Value alpha, Value beta, Value* bestValue,
Move* bestMove, Depth depth, Move threatMove, int moveCount,
MovePicker* movePicker, int nodeType) {
assert(pos.pos_is_ok());
assert(bestValue <= alpha && alpha < beta && beta <= VALUE_INFINITE);
assert(bestValue > -VALUE_INFINITE);
assert(*bestValue <= alpha && alpha < beta && beta <= VALUE_INFINITE);
assert(*bestValue > -VALUE_INFINITE);
assert(depth >= Threads.minimumSplitDepth);
Thread* master = pos.this_thread();
assert(master->searching);
assert(master->splitPointsSize < MAX_SPLITPOINTS_PER_THREAD);
assert(searching);
assert(splitPointsSize < MAX_SPLITPOINTS_PER_THREAD);
// Pick the next available split point from the split point stack
SplitPoint& sp = master->splitPoints[master->splitPointsSize];
SplitPoint& sp = splitPoints[splitPointsSize];
sp.master = master;
sp.parent = master->activeSplitPoint;
sp.slavesMask = 1ULL << master->idx;
sp.masterThread = this;
sp.parentSplitPoint = activeSplitPoint;
sp.slavesMask = 1ULL << idx;
sp.depth = depth;
sp.bestValue = *bestValue;
sp.bestMove = *bestMove;
sp.threatMove = threatMove;
sp.alpha = alpha;
sp.beta = beta;
sp.nodeType = nodeType;
sp.bestValue = bestValue;
sp.mp = &mp;
sp.movePicker = movePicker;
sp.moveCount = moveCount;
sp.pos = &pos;
sp.nodes = 0;
@ -285,25 +284,27 @@ Value ThreadPool::split(Position& pos, Stack* ss, Value alpha, Value beta,
// Try to allocate available threads and ask them to start searching setting
// 'searching' flag. This must be done under lock protection to avoid concurrent
// allocation of the same slave by another master.
mutex.lock();
Threads.mutex.lock();
sp.mutex.lock();
master->splitPointsSize++;
master->activeSplitPoint = &sp;
splitPointsSize++;
activeSplitPoint = &sp;
activePosition = NULL;
size_t slavesCnt = 1; // Master is always included
size_t slavesCnt = 1; // This thread is always included
Thread* slave;
for (size_t i = 0; i < threads.size() && !Fake; ++i)
if (threads[i]->is_available_to(master) && ++slavesCnt <= maxThreadsPerSplitPoint)
{
sp.slavesMask |= 1ULL << threads[i]->idx;
threads[i]->activeSplitPoint = &sp;
threads[i]->searching = true; // Slave leaves idle_loop()
threads[i]->notify_one(); // Could be sleeping
}
while ( (slave = Threads.available_slave(this)) != NULL
&& ++slavesCnt <= Threads.maxThreadsPerSplitPoint && !Fake)
{
sp.slavesMask |= 1ULL << slave->idx;
slave->activeSplitPoint = &sp;
slave->searching = true; // Slave leaves idle_loop()
slave->notify_one(); // Could be sleeping
}
sp.mutex.unlock();
mutex.unlock();
Threads.mutex.unlock();
// Everything is set up. The master thread enters the idle loop, from which
// it will instantly launch a search, because its 'searching' flag is set.
@ -311,34 +312,35 @@ Value ThreadPool::split(Position& pos, Stack* ss, Value alpha, Value beta,
// their work at this split point.
if (slavesCnt > 1 || Fake)
{
master->Thread::idle_loop(); // Force a call to base class idle_loop()
Thread::idle_loop(); // Force a call to base class idle_loop()
// In helpful master concept a master can help only a sub-tree of its split
// point, and because here is all finished is not possible master is booked.
assert(!master->searching);
assert(!searching);
assert(!activePosition);
}
// We have returned from the idle loop, which means that all threads are
// finished. Note that setting 'searching' and decreasing splitPointsSize is
// done under lock protection to avoid a race with Thread::is_available_to().
mutex.lock();
Threads.mutex.lock();
sp.mutex.lock();
master->searching = true;
master->splitPointsSize--;
master->activeSplitPoint = sp.parent;
searching = true;
splitPointsSize--;
activeSplitPoint = sp.parentSplitPoint;
activePosition = &pos;
pos.set_nodes_searched(pos.nodes_searched() + sp.nodes);
*bestMove = sp.bestMove;
*bestValue = sp.bestValue;
sp.mutex.unlock();
mutex.unlock();
return sp.bestValue;
Threads.mutex.unlock();
}
// Explicit template instantiations
template Value ThreadPool::split<false>(Position&, Stack*, Value, Value, Value, Move*, Depth, Move, int, MovePicker&, int);
template Value ThreadPool::split<true>(Position&, Stack*, Value, Value, Value, Move*, Depth, Move, int, MovePicker&, int);
template void Thread::split<false>(Position&, Stack*, Value, Value, Value*, Move*, Depth, Move, int, MovePicker*, int);
template void Thread::split< true>(Position&, Stack*, Value, Value, Value*, Move*, Depth, Move, int, MovePicker*, int);
// wait_for_think_finished() waits for main thread to go to sleep then returns
@ -370,7 +372,8 @@ void ThreadPool::start_thinking(const Position& pos, const LimitsType& limits,
RootMoves.clear();
for (MoveList<LEGAL> ml(pos); !ml.end(); ++ml)
if (searchMoves.empty() || count(searchMoves.begin(), searchMoves.end(), ml.move()))
if ( searchMoves.empty()
|| std::count(searchMoves.begin(), searchMoves.end(), ml.move()))
RootMoves.push_back(RootMove(ml.move()));
main_thread()->thinking = true;

34
src/thread.h
View file

@ -63,19 +63,18 @@ struct SplitPoint {
// Const data after split point has been setup
const Position* pos;
const Search::Stack* ss;
Thread* master;
Thread* masterThread;
Depth depth;
Value beta;
int nodeType;
Move threatMove;
// Const pointers to shared data
MovePicker* mp;
SplitPoint* parent;
MovePicker* movePicker;
SplitPoint* parentSplitPoint;
// Shared data
Mutex mutex;
Position* slavesPositions[MAX_THREADS];
volatile uint64_t slavesMask;
volatile int64_t nodes;
volatile Value alpha;
@ -102,10 +101,15 @@ struct Thread {
bool is_available_to(Thread* master) const;
void wait_for(volatile const bool& b);
template <bool Fake>
void split(Position& pos, Search::Stack* ss, Value alpha, Value beta, Value* bestValue, Move* bestMove,
Depth depth, Move threatMove, int moveCount, MovePicker* movePicker, int nodeType);
SplitPoint splitPoints[MAX_SPLITPOINTS_PER_THREAD];
Material::Table materialTable;
Endgames endgames;
Pawns::Table pawnsTable;
Position* activePosition;
size_t idx;
int maxPly;
Mutex mutex;
@ -134,40 +138,28 @@ struct TimerThread : public Thread {
};
/// ThreadPool class handles all the threads related stuff like init, starting,
/// ThreadPool struct handles all the threads related stuff like init, starting,
/// parking and, the most important, launching a slave thread at a split point.
/// All the access to shared thread data is done through this class.
class ThreadPool {
struct ThreadPool : public std::vector<Thread*> {
public:
void init(); // No c'tor and d'tor, threads rely on globals that should
void exit(); // be initialized and valid during the whole thread lifetime.
Thread& operator[](size_t id) { return *threads[id]; }
size_t size() const { return threads.size(); }
MainThread* main_thread() { return static_cast<MainThread*>(threads[0]); }
TimerThread* timer_thread() { return timer; }
MainThread* main_thread() { return static_cast<MainThread*>((*this)[0]); }
void read_uci_options();
bool slave_available(Thread* master) const;
Thread* available_slave(Thread* master) const;
void wait_for_think_finished();
void start_thinking(const Position&, const Search::LimitsType&,
const std::vector<Move>&, Search::StateStackPtr&);
template <bool Fake>
Value split(Position& pos, Search::Stack* ss, Value alpha, Value beta, Value bestValue, Move* bestMove,
Depth depth, Move threatMove, int moveCount, MovePicker& mp, int nodeType);
bool sleepWhileIdle;
Depth minimumSplitDepth;
size_t maxThreadsPerSplitPoint;
Mutex mutex;
ConditionVariable sleepCondition;
private:
std::vector<Thread*> threads;
TimerThread* timer;
size_t maxThreadsPerSplitPoint;
};
extern ThreadPool Threads;

66
src/tt.cpp
View file

@ -25,35 +25,25 @@
TranspositionTable TT; // Our global transposition table
TranspositionTable::TranspositionTable() {
size = generation = 0;
entries = NULL;
}
TranspositionTable::~TranspositionTable() {
delete [] entries;
}
/// TranspositionTable::set_size() sets the size of the transposition table,
/// measured in megabytes. Transposition table consists of a power of 2 number of
/// TTCluster and each cluster consists of ClusterSize number of TTEntries. Each
/// non-empty entry contains information of exactly one position.
/// measured in megabytes. Transposition table consists of a power of 2 number
/// of clusters and each cluster consists of ClusterSize number of TTEntry.
void TranspositionTable::set_size(size_t mbSize) {
size_t newSize = 1ULL << msb((mbSize << 20) / sizeof(TTCluster));
assert(msb((mbSize << 20) / sizeof(TTEntry)) < 32);
if (newSize == size)
uint32_t size = ClusterSize << msb((mbSize << 20) / sizeof(TTEntry[ClusterSize]));
if (hashMask == size - ClusterSize)
return;
size = newSize;
delete [] entries;
entries = new (std::nothrow) TTCluster[size];
hashMask = size - ClusterSize;
delete [] table;
table = new (std::nothrow) TTEntry[size];
if (!entries)
if (!table)
{
std::cerr << "Failed to allocate " << mbSize
<< "MB for transposition table." << std::endl;
@ -70,7 +60,7 @@ void TranspositionTable::set_size(size_t mbSize) {
void TranspositionTable::clear() {
memset(entries, 0, size * sizeof(TTCluster));
memset(table, 0, (hashMask + ClusterSize) * sizeof(TTEntry));
}
@ -82,23 +72,23 @@ void TranspositionTable::clear() {
/// more valuable than a TTEntry t2 if t1 is from the current search and t2 is from
/// a previous search, or if the depth of t1 is bigger than the depth of t2.
void TranspositionTable::store(const Key posKey, Value v, Bound t, Depth d, Move m, Value statV, Value kingD) {
void TranspositionTable::store(const Key key, Value v, Bound t, Depth d, Move m, Value statV, Value kingD) {
int c1, c2, c3;
TTEntry *tte, *replace;
uint32_t posKey32 = posKey >> 32; // Use the high 32 bits as key inside the cluster
uint32_t key32 = key >> 32; // Use the high 32 bits as key inside the cluster
tte = replace = first_entry(posKey);
tte = replace = first_entry(key);
for (int i = 0; i < ClusterSize; i++, tte++)
for (unsigned i = 0; i < ClusterSize; i++, tte++)
{
if (!tte->key() || tte->key() == posKey32) // Empty or overwrite old
if (!tte->key() || tte->key() == key32) // Empty or overwrite old
{
// Preserve any existing ttMove
if (m == MOVE_NONE)
m = tte->move();
tte->save(posKey32, v, t, d, m, generation, statV, kingD);
tte->save(key32, v, t, d, m, generation, statV, kingD);
return;
}
@ -110,7 +100,7 @@ void TranspositionTable::store(const Key posKey, Value v, Bound t, Depth d, Move
if (c1 + c2 + c3 > 0)
replace = tte;
}
replace->save(posKey32, v, t, d, m, generation, statV, kingD);
replace->save(key32, v, t, d, m, generation, statV, kingD);
}
@ -118,24 +108,14 @@ void TranspositionTable::store(const Key posKey, Value v, Bound t, Depth d, Move
/// transposition table. Returns a pointer to the TTEntry or NULL if
/// position is not found.
TTEntry* TranspositionTable::probe(const Key posKey) const {
TTEntry* TranspositionTable::probe(const Key key) const {
uint32_t posKey32 = posKey >> 32;
TTEntry* tte = first_entry(posKey);
TTEntry* tte = first_entry(key);
uint32_t key32 = key >> 32;
for (int i = 0; i < ClusterSize; i++, tte++)
if (tte->key() == posKey32)
for (unsigned i = 0; i < ClusterSize; i++, tte++)
if (tte->key() == key32)
return tte;
return NULL;
}
/// TranspositionTable::new_search() is called at the beginning of every new
/// search. It increments the "generation" variable, which is used to
/// distinguish transposition table entries from previous searches from
/// entries from the current search.
void TranspositionTable::new_search() {
generation++;
}

69
src/tt.h
View file

@ -44,7 +44,7 @@
class TTEntry {
public:
void save(uint32_t k, Value v, Bound b, Depth d, Move m, int g, Value statV, Value statM) {
void save(uint32_t k, Value v, Bound b, Depth d, Move m, int g, Value ev, Value em) {
key32 = (uint32_t)k;
move16 = (uint16_t)m;
@ -52,63 +52,52 @@ public:
generation8 = (uint8_t)g;
value16 = (int16_t)v;
depth16 = (int16_t)d;
staticValue = (int16_t)statV;
staticMargin = (int16_t)statM;
evalValue = (int16_t)ev;
evalMargin = (int16_t)em;
}
void set_generation(int g) { generation8 = (uint8_t)g; }
uint32_t key() const { return key32; }
Depth depth() const { return (Depth)depth16; }
Move move() const { return (Move)move16; }
Value value() const { return (Value)value16; }
Bound type() const { return (Bound)bound; }
int generation() const { return (int)generation8; }
Value static_value() const { return (Value)staticValue; }
Value static_value_margin() const { return (Value)staticMargin; }
uint32_t key() const { return key32; }
Depth depth() const { return (Depth)depth16; }
Move move() const { return (Move)move16; }
Value value() const { return (Value)value16; }
Bound type() const { return (Bound)bound; }
int generation() const { return (int)generation8; }
Value eval_value() const { return (Value)evalValue; }
Value eval_margin() const { return (Value)evalMargin; }
private:
uint32_t key32;
uint16_t move16;
uint8_t bound, generation8;
int16_t value16, depth16, staticValue, staticMargin;
int16_t value16, depth16, evalValue, evalMargin;
};
/// This is the number of TTEntry slots for each cluster
const int ClusterSize = 4;
/// TTCluster consists of ClusterSize number of TTEntries. Size of TTCluster
/// must not be bigger than a cache line size. In case it is less, it should
/// be padded to guarantee always aligned accesses.
struct TTCluster {
TTEntry data[ClusterSize];
};
/// The transposition table class. This is basically just a huge array containing
/// TTCluster objects, and a few methods for writing and reading entries.
/// A TranspositionTable consists of a power of 2 number of clusters and each
/// cluster consists of ClusterSize number of TTEntry. Each non-empty entry
/// contains information of exactly one position. Size of a cluster shall not be
/// bigger than a cache line size. In case it is less, it should be padded to
/// guarantee always aligned accesses.
class TranspositionTable {
TranspositionTable(const TranspositionTable&);
TranspositionTable& operator=(const TranspositionTable&);
static const unsigned ClusterSize = 4; // A cluster is 64 Bytes
public:
TranspositionTable();
~TranspositionTable();
~TranspositionTable() { delete [] table; }
void new_search() { generation++; }
TTEntry* probe(const Key key) const;
TTEntry* first_entry(const Key key) const;
void refresh(const TTEntry* tte) const;
void set_size(size_t mbSize);
void clear();
void store(const Key posKey, Value v, Bound type, Depth d, Move m, Value statV, Value kingD);
TTEntry* probe(const Key posKey) const;
void new_search();
TTEntry* first_entry(const Key posKey) const;
void refresh(const TTEntry* tte) const;
void store(const Key key, Value v, Bound type, Depth d, Move m, Value statV, Value kingD);
private:
size_t size;
TTCluster* entries;
uint32_t hashMask;
TTEntry* table;
uint8_t generation; // Size must be not bigger then TTEntry::generation8
};
@ -119,9 +108,9 @@ extern TranspositionTable TT;
/// a cluster given a position. The lowest order bits of the key are used to
/// get the index of the cluster.
inline TTEntry* TranspositionTable::first_entry(const Key posKey) const {
inline TTEntry* TranspositionTable::first_entry(const Key key) const {
return entries[((uint32_t)posKey) & (size - 1)].data;
return table + ((uint32_t)key & hashMask);
}

17
src/types.h
View file

@ -489,21 +489,4 @@ inline const std::string square_to_string(Square s) {
return ch;
}
/// Our insertion sort implementation, works with pointers and iterators and is
/// guaranteed to be stable, as is needed.
template<typename T, typename K>
void sort(K begin, K end)
{
T tmp;
K p, q;
for (p = begin + 1; p < end; p++)
{
tmp = *p;
for (q = p; q != begin && *(q-1) < tmp; --q)
*q = *(q-1);
*q = tmp;
}
}
#endif // !defined(TYPES_H_INCLUDED)

4
src/uci.cpp
View file

@ -44,7 +44,7 @@ namespace {
void set_option(istringstream& up);
void set_position(Position& pos, istringstream& up);
void go(Position& pos, istringstream& up);
void go(const Position& pos, istringstream& up);
}
@ -182,7 +182,7 @@ namespace {
// the thinking time and other parameters from the input string, and starts
// the search.
void go(Position& pos, istringstream& is) {
void go(const Position& pos, istringstream& is) {
Search::LimitsType limits;
vector<Move> searchMoves;