Merge remote-tracking branch 'upstream/master' into HEAD

[stockfish] / src / syzygy / tbprobe.cpp

diff --git a/src/syzygy/tbprobe.cpp b/src/syzygy/tbprobe.cpp
index 2b7f4497246e90a7105ebd202cfe0246191402a3..c0cd04c1d4d8bd8f30607b616ae02363cb84da69 100644 (file)
--- a/src/syzygy/tbprobe.cpp
+++ b/src/syzygy/tbprobe.cpp
@@ -1,7 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1

-  Copyright (c) 2013 Ronald de Man
-  Copyright (C) 2016-2018 Marco Costalba, Lucas Braesch
+  Copyright (C) 2004-2021 The Stockfish developers (see AUTHORS file)

 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by


@@ -27,12 +26,12 @@
 #include <list>
 #include <sstream>
 #include <type_traits>
 #include <list>
 #include <sstream>
 #include <type_traits>

+#include <mutex>

 
 #include "../bitboard.h"
 #include "../movegen.h"
 #include "../position.h"
 #include "../search.h"
 
 #include "../bitboard.h"
 #include "../movegen.h"
 #include "../position.h"
 #include "../search.h"

-#include "../thread_win32.h"

 #include "../types.h"
 #include "../uci.h"
 
 #include "../types.h"
 #include "../uci.h"
 


@@ -45,26 +44,29 @@
 #include <sys/stat.h>
 #else
 #define WIN32_LEAN_AND_MEAN
 #include <sys/stat.h>
 #else
 #define WIN32_LEAN_AND_MEAN

-#define NOMINMAX
+#ifndef NOMINMAX
+#  define NOMINMAX // Disable macros min() and max()
+#endif

 #include <windows.h>
 #endif
 
 #include <windows.h>
 #endif
 

-using namespace Tablebases;
+using namespace Stockfish::Tablebases;
+
+int Stockfish::Tablebases::MaxCardinality;

 
 

-int Tablebases::MaxCardinality;
+namespace Stockfish {

 
 namespace {
 
 
 namespace {
 

-constexpr int TBPIECES = 6; // Max number of supported pieces
+constexpr int TBPIECES = 7; // Max number of supported pieces

 
 enum { BigEndian, LittleEndian };
 
 enum { BigEndian, LittleEndian };

-enum TBType { KEY, WDL, DTZ }; // Used as template parameter
+enum TBType { WDL, DTZ }; // Used as template parameter

 
 // Each table has a set of flags: all of them refer to DTZ tables, the last one to WDL tables
 
 // Each table has a set of flags: all of them refer to DTZ tables, the last one to WDL tables

-enum TBFlag { STM = 1, Mapped = 2, WinPlies = 4, LossPlies = 8, SingleValue = 128 };
+enum TBFlag { STM = 1, Mapped = 2, WinPlies = 4, LossPlies = 8, Wide = 16, SingleValue = 128 };

 
 inline WDLScore operator-(WDLScore d) { return WDLScore(-int(d)); }
 
 inline WDLScore operator-(WDLScore d) { return WDLScore(-int(d)); }

-inline Square operator^=(Square& s, int i) { return s = Square(int(s) ^ i); }

 inline Square operator^(Square s, int i) { return Square(int(s) ^ i); }
 
 const std::string PieceToChar = " PNBRQK  pnbrqk";
 inline Square operator^(Square s, int i) { return Square(int(s) ^ i); }
 
 const std::string PieceToChar = " PNBRQK  pnbrqk";


@@ -74,9 +76,9 @@ int MapB1H1H7[SQUARE_NB];
 int MapA1D1D4[SQUARE_NB];
 int MapKK[10][SQUARE_NB]; // [MapA1D1D4][SQUARE_NB]
 
 int MapA1D1D4[SQUARE_NB];
 int MapKK[10][SQUARE_NB]; // [MapA1D1D4][SQUARE_NB]
 

-int Binomial[6][SQUARE_NB];    // [k][n] k elements from a set of n elements
-int LeadPawnIdx[5][SQUARE_NB]; // [leadPawnsCnt][SQUARE_NB]
-int LeadPawnsSize[5][4];       // [leadPawnsCnt][FILE_A..FILE_D]
+int Binomial[7][SQUARE_NB];    // [k][n] k elements from a set of n elements
+int LeadPawnIdx[6][SQUARE_NB]; // [leadPawnsCnt][SQUARE_NB]
+int LeadPawnsSize[6][4];       // [leadPawnsCnt][FILE_A..FILE_D]

 
 // Comparison function to sort leading pawns in ascending MapPawns[] order
 bool pawns_comp(Square i, Square j) { return MapPawns[i] < MapPawns[j]; }
 
 // Comparison function to sort leading pawns in ascending MapPawns[] order
 bool pawns_comp(Square i, Square j) { return MapPawns[i] < MapPawns[j]; }


@@ -144,16 +146,15 @@ static_assert(sizeof(SparseEntry) == 6, "SparseEntry must be 6 bytes");
 typedef uint16_t Sym; // Huffman symbol
 
 struct LR {
 typedef uint16_t Sym; // Huffman symbol
 
 struct LR {

-    enum Side { Left, Right, Value };
+    enum Side { Left, Right };

 
     uint8_t lr[3]; // The first 12 bits is the left-hand symbol, the second 12
                    // bits is the right-hand symbol. If symbol has length 1,
 
     uint8_t lr[3]; // The first 12 bits is the left-hand symbol, the second 12
                    // bits is the right-hand symbol. If symbol has length 1,

-                   // then the first byte is the stored value.
+                   // then the left-hand symbol is the stored value.

     template<Side S>
     Sym get() {
         return S == Left  ? ((lr[1] & 0xF) << 8) | lr[0] :
     template<Side S>
     Sym get() {
         return S == Left  ? ((lr[1] & 0xF) << 8) | lr[0] :

-               S == Right ?  (lr[2] << 4) | (lr[1] >> 4) :
-               S == Value ?   lr[0] : (assert(false), Sym(-1));
+               S == Right ?  (lr[2] << 4) | (lr[1] >> 4) : (assert(false), Sym(-1));

     }
 };
 
     }
 };
 


@@ -215,38 +216,59 @@ public:
             return *baseAddress = nullptr, nullptr;
 
         fstat(fd, &statbuf);
             return *baseAddress = nullptr, nullptr;
 
         fstat(fd, &statbuf);

+
+        if (statbuf.st_size % 64 != 16)
+        {
+            std::cerr << "Corrupt tablebase file " << fname << std::endl;
+            exit(EXIT_FAILURE);
+        }
+

         *mapping = statbuf.st_size;
         *baseAddress = mmap(nullptr, statbuf.st_size, PROT_READ, MAP_SHARED, fd, 0);
         *mapping = statbuf.st_size;
         *baseAddress = mmap(nullptr, statbuf.st_size, PROT_READ, MAP_SHARED, fd, 0);

+#if defined(MADV_RANDOM)
+        madvise(*baseAddress, statbuf.st_size, MADV_RANDOM);
+#endif

         ::close(fd);
 
         ::close(fd);
 

-        if (*baseAddress == MAP_FAILED) {
+        if (*baseAddress == MAP_FAILED)
+        {

             std::cerr << "Could not mmap() " << fname << std::endl;
             std::cerr << "Could not mmap() " << fname << std::endl;

-            exit(1);
+            exit(EXIT_FAILURE);

         }
 #else
         }
 #else

+        // Note FILE_FLAG_RANDOM_ACCESS is only a hint to Windows and as such may get ignored.

         HANDLE fd = CreateFile(fname.c_str(), GENERIC_READ, FILE_SHARE_READ, nullptr,
         HANDLE fd = CreateFile(fname.c_str(), GENERIC_READ, FILE_SHARE_READ, nullptr,

-                               OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, nullptr);
+                               OPEN_EXISTING, FILE_FLAG_RANDOM_ACCESS, nullptr);

 
         if (fd == INVALID_HANDLE_VALUE)
             return *baseAddress = nullptr, nullptr;
 
         DWORD size_high;
         DWORD size_low = GetFileSize(fd, &size_high);
 
         if (fd == INVALID_HANDLE_VALUE)
             return *baseAddress = nullptr, nullptr;
 
         DWORD size_high;
         DWORD size_low = GetFileSize(fd, &size_high);

+
+        if (size_low % 64 != 16)
+        {
+            std::cerr << "Corrupt tablebase file " << fname << std::endl;
+            exit(EXIT_FAILURE);
+        }
+

         HANDLE mmap = CreateFileMapping(fd, nullptr, PAGE_READONLY, size_high, size_low, nullptr);
         CloseHandle(fd);
 
         HANDLE mmap = CreateFileMapping(fd, nullptr, PAGE_READONLY, size_high, size_low, nullptr);
         CloseHandle(fd);
 

-        if (!mmap) {
+        if (!mmap)
+        {

             std::cerr << "CreateFileMapping() failed" << std::endl;
             std::cerr << "CreateFileMapping() failed" << std::endl;

-            exit(1);
+            exit(EXIT_FAILURE);

         }
 
         *mapping = (uint64_t)mmap;
         *baseAddress = MapViewOfFile(mmap, FILE_MAP_READ, 0, 0, 0);
 
         }
 
         *mapping = (uint64_t)mmap;
         *baseAddress = MapViewOfFile(mmap, FILE_MAP_READ, 0, 0, 0);
 

-        if (!*baseAddress) {
+        if (!*baseAddress)
+        {

             std::cerr << "MapViewOfFile() failed, name = " << fname
                       << ", error = " << GetLastError() << std::endl;
             std::cerr << "MapViewOfFile() failed, name = " << fname
                       << ", error = " << GetLastError() << std::endl;

-            exit(1);
+            exit(EXIT_FAILURE);

         }
 #endif
         uint8_t* data = (uint8_t*)*baseAddress;
         }
 #endif
         uint8_t* data = (uint8_t*)*baseAddress;


@@ -254,7 +276,8 @@ public:
         constexpr uint8_t Magics[][4] = { { 0xD7, 0x66, 0x0C, 0xA5 },
                                           { 0x71, 0xE8, 0x23, 0x5D } };
 
         constexpr uint8_t Magics[][4] = { { 0xD7, 0x66, 0x0C, 0xA5 },
                                           { 0x71, 0xE8, 0x23, 0x5D } };
 

-        if (memcmp(data, Magics[type == WDL], 4)) {
+        if (memcmp(data, Magics[type == WDL], 4))
+        {

             std::cerr << "Corrupted table in file " << fname << std::endl;
             unmap(*baseAddress, *mapping);
             return *baseAddress = nullptr, nullptr;
             std::cerr << "Corrupted table in file " << fname << std::endl;
             unmap(*baseAddress, *mapping);
             return *baseAddress = nullptr, nullptr;


@@ -348,7 +371,7 @@ TBTable<WDL>::TBTable(const std::string& code) : TBTable() {
     hasPawns = pos.pieces(PAWN);
 
     hasUniquePieces = false;
     hasPawns = pos.pieces(PAWN);
 
     hasUniquePieces = false;

-    for (Color c = WHITE; c <= BLACK; ++c)
+    for (Color c : { WHITE, BLACK })

         for (PieceType pt = PAWN; pt < KING; ++pt)
             if (popcount(pos.pieces(c, pt)) == 1)
                 hasUniquePieces = true;
         for (PieceType pt = PAWN; pt < KING; ++pt)
             if (popcount(pos.pieces(c, pt)) == 1)
                 hasUniquePieces = true;


@@ -383,34 +406,57 @@ TBTable<DTZ>::TBTable(const TBTable<WDL>& wdl) : TBTable() {
 // at init time, accessed at probe time.
 class TBTables {
 
 // at init time, accessed at probe time.
 class TBTables {
 

-    typedef std::tuple<Key, TBTable<WDL>*, TBTable<DTZ>*> Entry;
+    struct Entry
+    {
+        Key key;
+        TBTable<WDL>* wdl;
+        TBTable<DTZ>* dtz;

 
 

-    static const int Size = 1 << 12; // 4K table, indexed by key's 12 lsb
+        template <TBType Type>
+        TBTable<Type>* get() const {
+            return (TBTable<Type>*)(Type == WDL ? (void*)wdl : (void*)dtz);
+        }
+    };

 
 

-    Entry hashTable[Size];
+    static constexpr int Size = 1 << 12; // 4K table, indexed by key's 12 lsb
+    static constexpr int Overflow = 1;  // Number of elements allowed to map to the last bucket
+
+    Entry hashTable[Size + Overflow];

 
     std::deque<TBTable<WDL>> wdlTable;
     std::deque<TBTable<DTZ>> dtzTable;
 
     void insert(Key key, TBTable<WDL>* wdl, TBTable<DTZ>* dtz) {
 
     std::deque<TBTable<WDL>> wdlTable;
     std::deque<TBTable<DTZ>> dtzTable;
 
     void insert(Key key, TBTable<WDL>* wdl, TBTable<DTZ>* dtz) {

-        Entry* entry = &hashTable[(uint32_t)key & (Size - 1)];
+        uint32_t homeBucket = (uint32_t)key & (Size - 1);
+        Entry entry{ key, wdl, dtz };

 
         // Ensure last element is empty to avoid overflow when looking up
 
         // Ensure last element is empty to avoid overflow when looking up

-        for ( ; entry - hashTable < Size - 1; ++entry)
-            if (std::get<KEY>(*entry) == key || !std::get<WDL>(*entry)) {
-                *entry = std::make_tuple(key, wdl, dtz);
+        for (uint32_t bucket = homeBucket; bucket < Size + Overflow - 1; ++bucket) {
+            Key otherKey = hashTable[bucket].key;
+            if (otherKey == key || !hashTable[bucket].get<WDL>()) {
+                hashTable[bucket] = entry;

                 return;
             }
                 return;
             }

+
+            // Robin Hood hashing: If we've probed for longer than this element,
+            // insert here and search for a new spot for the other element instead.
+            uint32_t otherHomeBucket = (uint32_t)otherKey & (Size - 1);
+            if (otherHomeBucket > homeBucket) {
+                std::swap(entry, hashTable[bucket]);
+                key = otherKey;
+                homeBucket = otherHomeBucket;
+            }
+        }

         std::cerr << "TB hash table size too low!" << std::endl;
         std::cerr << "TB hash table size too low!" << std::endl;

-        exit(1);
+        exit(EXIT_FAILURE);

     }
 
 public:
     template<TBType Type>
     TBTable<Type>* get(Key key) {
         for (const Entry* entry = &hashTable[(uint32_t)key & (Size - 1)]; ; ++entry) {
     }
 
 public:
     template<TBType Type>
     TBTable<Type>* get(Key key) {
         for (const Entry* entry = &hashTable[(uint32_t)key & (Size - 1)]; ; ++entry) {

-            if (std::get<KEY>(*entry) == key || !std::get<Type>(*entry))
-                return std::get<Type>(*entry);
+            if (entry->key == key || !entry->get<Type>())
+                return entry->get<Type>();

         }
     }
 
         }
     }
 


@@ -487,7 +533,7 @@ int decompress_pairs(PairsData* d, uint64_t idx) {
     //       I(k) = k * d->span + d->span / 2      (1)
 
     // First step is to get the 'k' of the I(k) nearest to our idx, using definition (1)
     //       I(k) = k * d->span + d->span / 2      (1)
 
     // First step is to get the 'k' of the I(k) nearest to our idx, using definition (1)

-    uint32_t k = idx / d->span;
+    uint32_t k = uint32_t(idx / d->span);

 
     // Then we read the corresponding SparseIndex[] entry
     uint32_t block = number<uint32_t, LittleEndian>(&d->sparseIndex[k].block);
 
     // Then we read the corresponding SparseIndex[] entry
     uint32_t block = number<uint32_t, LittleEndian>(&d->sparseIndex[k].block);


@@ -512,7 +558,7 @@ int decompress_pairs(PairsData* d, uint64_t idx) {
         offset -= d->blockLength[block++] + 1;
 
     // Finally, we find the start address of our block of canonical Huffman symbols
         offset -= d->blockLength[block++] + 1;
 
     // Finally, we find the start address of our block of canonical Huffman symbols

-    uint32_t* ptr = (uint32_t*)(d->data + block * d->sizeofBlock);
+    uint32_t* ptr = (uint32_t*)(d->data + ((uint64_t)block * d->sizeofBlock));

 
     // Read the first 64 bits in our block, this is a (truncated) sequence of
     // unknown number of symbols of unknown length but we know the first one
 
     // Read the first 64 bits in our block, this is a (truncated) sequence of
     // unknown number of symbols of unknown length but we know the first one


@@ -533,7 +579,7 @@ int decompress_pairs(PairsData* d, uint64_t idx) {
         // All the symbols of a given length are consecutive integers (numerical
         // sequence property), so we can compute the offset of our symbol of
         // length len, stored at the beginning of buf64.
         // All the symbols of a given length are consecutive integers (numerical
         // sequence property), so we can compute the offset of our symbol of
         // length len, stored at the beginning of buf64.

-        sym = (buf64 - d->base64[len]) >> (64 - len - d->minSymLen);
+        sym = Sym((buf64 - d->base64[len]) >> (64 - len - d->minSymLen));

 
         // Now add the value of the lowest symbol of length len to get our symbol
         sym += number<Sym, LittleEndian>(&d->lowestSym[len]);
 
         // Now add the value of the lowest symbol of length len to get our symbol
         sym += number<Sym, LittleEndian>(&d->lowestSym[len]);


@@ -575,7 +621,7 @@ int decompress_pairs(PairsData* d, uint64_t idx) {
         }
     }
 
         }
     }
 

-    return d->btree[sym].get<LR::Value>();
+    return d->btree[sym].get<LR::Left>();

 }
 
 bool check_dtz_stm(TBTable<WDL>*, int, File) { return true; }
 }
 
 bool check_dtz_stm(TBTable<WDL>*, int, File) { return true; }


@@ -601,8 +647,12 @@ int map_score(TBTable<DTZ>* entry, File f, int value, WDLScore wdl) {
 
     uint8_t* map = entry->map;
     uint16_t* idx = entry->get(0, f)->map_idx;
 
     uint8_t* map = entry->map;
     uint16_t* idx = entry->get(0, f)->map_idx;

-    if (flags & TBFlag::Mapped)
-        value = map[idx[WDLMap[wdl + 2]] + value];
+    if (flags & TBFlag::Mapped) {
+        if (flags & TBFlag::Wide)
+            value = ((uint16_t *)map)[idx[WDLMap[wdl + 2]] + value];
+        else
+            value = map[idx[WDLMap[wdl + 2]] + value];
+    }

 
     // DTZ tables store distance to zero in number of moves or plies. We
     // want to return plies, so we have convert to plies when needed.
 
     // DTZ tables store distance to zero in number of moves or plies. We
     // want to return plies, so we have convert to plies when needed.


@@ -645,7 +695,7 @@ Ret do_probe_table(const Position& pos, T* entry, WDLScore wdl, ProbeState* resu
     bool blackStronger = (pos.material_key() != entry->key);
 
     int flipColor   = (symmetricBlackToMove || blackStronger) * 8;
     bool blackStronger = (pos.material_key() != entry->key);
 
     int flipColor   = (symmetricBlackToMove || blackStronger) * 8;

-    int flipSquares = (symmetricBlackToMove || blackStronger) * 070;
+    int flipSquares = (symmetricBlackToMove || blackStronger) * 56;

     int stm         = (symmetricBlackToMove || blackStronger) ^ pos.side_to_move();
 
     // For pawns, TB files store 4 separate tables according if leading pawn is on
     int stm         = (symmetricBlackToMove || blackStronger) ^ pos.side_to_move();
 
     // For pawns, TB files store 4 separate tables according if leading pawn is on


@@ -668,9 +718,7 @@ Ret do_probe_table(const Position& pos, T* entry, WDLScore wdl, ProbeState* resu
 
         std::swap(squares[0], *std::max_element(squares, squares + leadPawnsCnt, pawns_comp));
 
 
         std::swap(squares[0], *std::max_element(squares, squares + leadPawnsCnt, pawns_comp));
 

-        tbFile = file_of(squares[0]);
-        if (tbFile > FILE_D)
-            tbFile = file_of(squares[0] ^ 7); // Horizontal flip: SQ_H1 -> SQ_A1
+        tbFile = File(edge_distance(file_of(squares[0])));

     }
 
     // DTZ tables are one-sided, i.e. they store positions only for white to
     }
 
     // DTZ tables are one-sided, i.e. they store positions only for white to


@@ -694,8 +742,8 @@ Ret do_probe_table(const Position& pos, T* entry, WDLScore wdl, ProbeState* resu
 
     // Then we reorder the pieces to have the same sequence as the one stored
     // in pieces[i]: the sequence that ensures the best compression.
 
     // Then we reorder the pieces to have the same sequence as the one stored
     // in pieces[i]: the sequence that ensures the best compression.

-    for (int i = leadPawnsCnt; i < size; ++i)
-        for (int j = i; j < size; ++j)
+    for (int i = leadPawnsCnt; i < size - 1; ++i)
+        for (int j = i + 1; j < size; ++j)

             if (d->pieces[i] == pieces[j])
             {
                 std::swap(pieces[i], pieces[j]);
             if (d->pieces[i] == pieces[j])
             {
                 std::swap(pieces[i], pieces[j]);


@@ -707,14 +755,14 @@ Ret do_probe_table(const Position& pos, T* entry, WDLScore wdl, ProbeState* resu
     // the triangle A1-D1-D4.
     if (file_of(squares[0]) > FILE_D)
         for (int i = 0; i < size; ++i)
     // the triangle A1-D1-D4.
     if (file_of(squares[0]) > FILE_D)
         for (int i = 0; i < size; ++i)

-            squares[i] ^= 7; // Horizontal flip: SQ_H1 -> SQ_A1
+            squares[i] = flip_file(squares[i]);

 
     // Encode leading pawns starting with the one with minimum MapPawns[] and
     // proceeding in ascending order.
     if (entry->hasPawns) {
         idx = LeadPawnIdx[leadPawnsCnt][squares[0]];
 
 
     // Encode leading pawns starting with the one with minimum MapPawns[] and
     // proceeding in ascending order.
     if (entry->hasPawns) {
         idx = LeadPawnIdx[leadPawnsCnt][squares[0]];
 

-        std::sort(squares + 1, squares + leadPawnsCnt, pawns_comp);
+        std::stable_sort(squares + 1, squares + leadPawnsCnt, pawns_comp);

 
         for (int i = 1; i < leadPawnsCnt; ++i)
             idx += Binomial[i][MapPawns[squares[i]]];
 
         for (int i = 1; i < leadPawnsCnt; ++i)
             idx += Binomial[i][MapPawns[squares[i]]];


@@ -726,7 +774,7 @@ Ret do_probe_table(const Position& pos, T* entry, WDLScore wdl, ProbeState* resu
     // piece is below RANK_5.
     if (rank_of(squares[0]) > RANK_4)
         for (int i = 0; i < size; ++i)
     // piece is below RANK_5.
     if (rank_of(squares[0]) > RANK_4)
         for (int i = 0; i < size; ++i)

-            squares[i] ^= 070; // Vertical flip: SQ_A8 -> SQ_A1
+            squares[i] = flip_rank(squares[i]);

 
     // Look for the first piece of the leading group not on the A1-D4 diagonal
     // and ensure it is mapped below the diagonal.
 
     // Look for the first piece of the leading group not on the A1-D4 diagonal
     // and ensure it is mapped below the diagonal.


@@ -734,7 +782,7 @@ Ret do_probe_table(const Position& pos, T* entry, WDLScore wdl, ProbeState* resu
         if (!off_A1H8(squares[i]))
             continue;
 
         if (!off_A1H8(squares[i]))
             continue;
 

-        if (off_A1H8(squares[i]) > 0) // A1-H8 diagonal flip: SQ_A3 -> SQ_C3
+        if (off_A1H8(squares[i]) > 0) // A1-H8 diagonal flip: SQ_A3 -> SQ_C1

             for (int j = i; j < size; ++j)
                 squares[j] = Square(((squares[j] >> 3) | (squares[j] << 3)) & 63);
         break;
             for (int j = i; j < size; ++j)
                 squares[j] = Square(((squares[j] >> 3) | (squares[j] << 3)) & 63);
         break;


@@ -815,7 +863,7 @@ encode_remaining:
 
     while (d->groupLen[++next])
     {
 
     while (d->groupLen[++next])
     {

-        std::sort(groupSq, groupSq + d->groupLen[next]);
+        std::stable_sort(groupSq, groupSq + d->groupLen[next]);

         uint64_t n = 0;
 
         // Map down a square if "comes later" than a square in the previous
         uint64_t n = 0;
 
         // Map down a square if "comes later" than a square in the previous


@@ -939,7 +987,7 @@ uint8_t* set_sizes(PairsData* d, uint8_t* data) {
 
     d->sizeofBlock = 1ULL << *data++;
     d->span = 1ULL << *data++;
 
     d->sizeofBlock = 1ULL << *data++;
     d->span = 1ULL << *data++;

-    d->sparseIndexSize = (tbSize + d->span - 1) / d->span; // Round up
+    d->sparseIndexSize = size_t((tbSize + d->span - 1) / d->span); // Round up

     auto padding = number<uint8_t, LittleEndian>(data++);
     d->blocksNum = number<uint32_t, LittleEndian>(data); data += sizeof(uint32_t);
     d->blockLengthSize = d->blocksNum + padding; // Padded to ensure SparseIndex[]
     auto padding = number<uint8_t, LittleEndian>(data++);
     d->blocksNum = number<uint32_t, LittleEndian>(data); data += sizeof(uint32_t);
     d->blockLengthSize = d->blocksNum + padding; // Padded to ensure SparseIndex[]


@@ -954,7 +1002,7 @@ uint8_t* set_sizes(PairsData* d, uint8_t* data) {
     // so that d->lowestSym[i] >= d->lowestSym[i+1] (when read as LittleEndian).
     // Starting from this we compute a base64[] table indexed by symbol length
     // and containing 64 bit values so that d->base64[i] >= d->base64[i+1].
     // so that d->lowestSym[i] >= d->lowestSym[i+1] (when read as LittleEndian).
     // Starting from this we compute a base64[] table indexed by symbol length
     // and containing 64 bit values so that d->base64[i] >= d->base64[i+1].

-    // See http://www.eecs.harvard.edu/~michaelm/E210/huffman.pdf
+    // See https://en.wikipedia.org/wiki/Huffman_coding

     for (int i = d->base64.size() - 2; i >= 0; --i) {
         d->base64[i] = (d->base64[i + 1] + number<Sym, LittleEndian>(&d->lowestSym[i])
                                          - number<Sym, LittleEndian>(&d->lowestSym[i + 1])) / 2;
     for (int i = d->base64.size() - 2; i >= 0; --i) {
         d->base64[i] = (d->base64[i + 1] + number<Sym, LittleEndian>(&d->lowestSym[i])
                                          - number<Sym, LittleEndian>(&d->lowestSym[i + 1])) / 2;


@@ -973,7 +1021,7 @@ uint8_t* set_sizes(PairsData* d, uint8_t* data) {
     d->symlen.resize(number<uint16_t, LittleEndian>(data)); data += sizeof(uint16_t);
     d->btree = (LR*)data;
 
     d->symlen.resize(number<uint16_t, LittleEndian>(data)); data += sizeof(uint16_t);
     d->btree = (LR*)data;
 

-    // The comrpession scheme used is "Recursive Pairing", that replaces the most
+    // The compression scheme used is "Recursive Pairing", that replaces the most

     // frequent adjacent pair of symbols in the source message by a new symbol,
     // reevaluating the frequencies of all of the symbol pairs with respect to
     // the extended alphabet, and then repeating the process.
     // frequent adjacent pair of symbols in the source message by a new symbol,
     // reevaluating the frequencies of all of the symbol pairs with respect to
     // the extended alphabet, and then repeating the process.


@@ -994,11 +1042,22 @@ uint8_t* set_dtz_map(TBTable<DTZ>& e, uint8_t* data, File maxFile) {
     e.map = data;
 
     for (File f = FILE_A; f <= maxFile; ++f) {
     e.map = data;
 
     for (File f = FILE_A; f <= maxFile; ++f) {

-        if (e.get(0, f)->flags & TBFlag::Mapped)
-            for (int i = 0; i < 4; ++i) { // Sequence like 3,x,x,x,1,x,0,2,x,x
-                e.get(0, f)->map_idx[i] = (uint16_t)(data - e.map + 1);
-                data += *data + 1;
+        auto flags = e.get(0, f)->flags;
+        if (flags & TBFlag::Mapped) {
+            if (flags & TBFlag::Wide) {
+                data += (uintptr_t)data & 1;  // Word alignment, we may have a mixed table
+                for (int i = 0; i < 4; ++i) { // Sequence like 3,x,x,x,1,x,0,2,x,x
+                    e.get(0, f)->map_idx[i] = (uint16_t)((uint16_t *)data - (uint16_t *)e.map + 1);
+                    data += 2 * number<uint16_t, LittleEndian>(data) + 2;
+                }
+            }
+            else {
+                for (int i = 0; i < 4; ++i) {
+                    e.get(0, f)->map_idx[i] = (uint16_t)(data - e.map + 1);
+                    data += *data + 1;
+                }

             }
             }

+        }

     }
 
     return data += (uintptr_t)data & 1; // Word alignment
     }
 
     return data += (uintptr_t)data & 1; // Word alignment


@@ -1013,8 +1072,8 @@ void set(T& e, uint8_t* data) {
 
     enum { Split = 1, HasPawns = 2 };
 
 
     enum { Split = 1, HasPawns = 2 };
 

-    assert(e.hasPawns        == !!(*data & HasPawns));
-    assert((e.key != e.key2) == !!(*data & Split));
+    assert(e.hasPawns        == bool(*data & HasPawns));
+    assert((e.key != e.key2) == bool(*data & Split));

 
     data++; // First byte stores flags
 
 
     data++; // First byte stores flags
 


@@ -1077,14 +1136,14 @@ void set(T& e, uint8_t* data) {
 template<TBType Type>
 void* mapped(TBTable<Type>& e, const Position& pos) {
 
 template<TBType Type>
 void* mapped(TBTable<Type>& e, const Position& pos) {
 

-    static Mutex mutex;
+    static std::mutex mutex;

 
 

-    // Use 'aquire' to avoid a thread reads 'ready' == true while another is
-    // still working, this could happen due to compiler reordering.
+    // Use 'acquire' to avoid a thread reading 'ready' == true while
+    // another is still working. (compiler reordering may cause this).

     if (e.ready.load(std::memory_order_acquire))
     if (e.ready.load(std::memory_order_acquire))

-        return e.baseAddress; // Could be nullptr if file does not exsist
+        return e.baseAddress; // Could be nullptr if file does not exist

 
 

-    std::unique_lock<Mutex> lk(mutex);
+    std::scoped_lock<std::mutex> lk(mutex);

 
     if (e.ready.load(std::memory_order_relaxed)) // Recheck under lock
         return e.baseAddress;
 
     if (e.ready.load(std::memory_order_relaxed)) // Recheck under lock
         return e.baseAddress;


@@ -1144,7 +1203,7 @@ WDLScore search(Position& pos, ProbeState* result) {
     auto moveList = MoveList<LEGAL>(pos);
     size_t totalCount = moveList.size(), moveCount = 0;
 
     auto moveList = MoveList<LEGAL>(pos);
     size_t totalCount = moveList.size(), moveCount = 0;
 

-    for (const Move& move : moveList)
+    for (const Move move : moveList)

     {
         if (   !pos.capture(move)
             && (!CheckZeroingMoves || type_of(pos.moved_piece(move)) != PAWN))
     {
         if (   !pos.capture(move)
             && (!CheckZeroingMoves || type_of(pos.moved_piece(move)) != PAWN))


@@ -1249,7 +1308,7 @@ void Tablebases::init(const std::string& paths) {
                         continue; // First on diagonal, second above
 
                     else if (!off_A1H8(s1) && !off_A1H8(s2))
                         continue; // First on diagonal, second above
 
                     else if (!off_A1H8(s1) && !off_A1H8(s2))

-                        bothOnDiagonal.push_back(std::make_pair(idx, s2));
+                        bothOnDiagonal.emplace_back(idx, s2);

 
                     else
                         MapKK[idx][s2] = code++;
 
                     else
                         MapKK[idx][s2] = code++;


@@ -1264,7 +1323,7 @@ void Tablebases::init(const std::string& paths) {
     Binomial[0][0] = 1;
 
     for (int n = 1; n < 64; n++) // Squares
     Binomial[0][0] = 1;
 
     for (int n = 1; n < 64; n++) // Squares

-        for (int k = 0; k < 6 && k <= n; ++k) // Pieces
+        for (int k = 0; k < 7 && k <= n; ++k) // Pieces

             Binomial[k][n] =  (k > 0 ? Binomial[k - 1][n - 1] : 0)
                             + (k < n ? Binomial[k    ][n - 1] : 0);
 
             Binomial[k][n] =  (k > 0 ? Binomial[k - 1][n - 1] : 0)
                             + (k < n ? Binomial[k    ][n - 1] : 0);
 


@@ -1274,9 +1333,9 @@ void Tablebases::init(const std::string& paths) {
     // among pawns with same file, the one with lowest rank.
     int availableSquares = 47; // Available squares when lead pawn is in a2
 
     // among pawns with same file, the one with lowest rank.
     int availableSquares = 47; // Available squares when lead pawn is in a2
 

-    // Init the tables for the encoding of leading pawns group: with 6-men TB we
-    // can have up to 4 leading pawns (KPPPPK).
-    for (int leadPawnsCnt = 1; leadPawnsCnt <= 4; ++leadPawnsCnt)
+    // Init the tables for the encoding of leading pawns group: with 7-men TB we
+    // can have up to 5 leading pawns (KPPPPPK).
+    for (int leadPawnsCnt = 1; leadPawnsCnt <= 5; ++leadPawnsCnt)

         for (File f = FILE_A; f <= FILE_D; ++f)
         {
             // Restart the index at every file because TB table is splitted
         for (File f = FILE_A; f <= FILE_D; ++f)
         {
             // Restart the index at every file because TB table is splitted


@@ -1297,7 +1356,7 @@ void Tablebases::init(const std::string& paths) {
                 if (leadPawnsCnt == 1)
                 {
                     MapPawns[sq] = availableSquares--;
                 if (leadPawnsCnt == 1)
                 {
                     MapPawns[sq] = availableSquares--;

-                    MapPawns[sq ^ 7] = availableSquares--; // Horizontal flip
+                    MapPawns[flip_file(sq)] = availableSquares--;

                 }
                 LeadPawnIdx[leadPawnsCnt][sq] = idx;
                 idx += Binomial[leadPawnsCnt - 1][MapPawns[sq]];
                 }
                 LeadPawnIdx[leadPawnsCnt][sq] = idx;
                 idx += Binomial[leadPawnsCnt - 1][MapPawns[sq]];


@@ -1306,7 +1365,7 @@ void Tablebases::init(const std::string& paths) {
             LeadPawnsSize[leadPawnsCnt][f] = idx;
         }
 
             LeadPawnsSize[leadPawnsCnt][f] = idx;
         }
 

-    // Add entries in TB tables if the corresponding ".rtbw" file exsists
+    // Add entries in TB tables if the corresponding ".rtbw" file exists

     for (PieceType p1 = PAWN; p1 < KING; ++p1) {
         TBTables.add({KING, p1, KING});
 
     for (PieceType p1 = PAWN; p1 < KING; ++p1) {
         TBTables.add({KING, p1, KING});
 


@@ -1320,11 +1379,22 @@ void Tablebases::init(const std::string& paths) {
             for (PieceType p3 = PAWN; p3 <= p2; ++p3) {
                 TBTables.add({KING, p1, p2, p3, KING});
 
             for (PieceType p3 = PAWN; p3 <= p2; ++p3) {
                 TBTables.add({KING, p1, p2, p3, KING});
 

-                for (PieceType p4 = PAWN; p4 <= p3; ++p4)
+                for (PieceType p4 = PAWN; p4 <= p3; ++p4) {

                     TBTables.add({KING, p1, p2, p3, p4, KING});
 
                     TBTables.add({KING, p1, p2, p3, p4, KING});
 

-                for (PieceType p4 = PAWN; p4 < KING; ++p4)
+                    for (PieceType p5 = PAWN; p5 <= p4; ++p5)
+                        TBTables.add({KING, p1, p2, p3, p4, p5, KING});
+
+                    for (PieceType p5 = PAWN; p5 < KING; ++p5)
+                        TBTables.add({KING, p1, p2, p3, p4, KING, p5});
+                }
+
+                for (PieceType p4 = PAWN; p4 < KING; ++p4) {

                     TBTables.add({KING, p1, p2, p3, KING, p4});
                     TBTables.add({KING, p1, p2, p3, KING, p4});

+
+                    for (PieceType p5 = PAWN; p5 <= p4; ++p5)
+                        TBTables.add({KING, p1, p2, p3, KING, p4, p5});
+                }

             }
 
             for (PieceType p3 = PAWN; p3 <= p1; ++p3)
             }
 
             for (PieceType p3 = PAWN; p3 <= p1; ++p3)


@@ -1372,7 +1442,7 @@ WDLScore Tablebases::probe_wdl(Position& pos, ProbeState* result) {
 // If n = 100 immediately after a capture or pawn move, then the position
 // is also certainly a win, and during the whole phase until the next
 // capture or pawn move, the inequality to be preserved is
 // If n = 100 immediately after a capture or pawn move, then the position
 // is also certainly a win, and during the whole phase until the next
 // capture or pawn move, the inequality to be preserved is

-// dtz + 50-movecounter <= 100.
+// dtz + 50-move-counter <= 100.

 //
 // In short, if a move is available resulting in dtz + 50-move-counter <= 99,
 // then do not accept moves leading to dtz + 50-move-counter == 100.
 //
 // In short, if a move is available resulting in dtz + 50-move-counter <= 99,
 // then do not accept moves leading to dtz + 50-move-counter == 100.


@@ -1402,7 +1472,7 @@ int Tablebases::probe_dtz(Position& pos, ProbeState* result) {
     StateInfo st;
     int minDTZ = 0xFFFF;
 
     StateInfo st;
     int minDTZ = 0xFFFF;
 

-    for (const Move& move : MoveList<LEGAL>(pos))
+    for (const Move move : MoveList<LEGAL>(pos))

     {
         bool zeroing = pos.capture(move) || type_of(pos.moved_piece(move)) == PAWN;
 
     {
         bool zeroing = pos.capture(move) || type_of(pos.moved_piece(move)) == PAWN;
 


@@ -1491,12 +1561,12 @@ bool Tablebases::root_probe(Position& pos, Search::RootMoves& rootMoves) {
         int r =  dtz > 0 ? (dtz + cnt50 <= 99 && !rep ? 1000 : 1000 - (dtz + cnt50))
                : dtz < 0 ? (-dtz * 2 + cnt50 < 100 ? -1000 : -1000 + (-dtz + cnt50))
                : 0;
         int r =  dtz > 0 ? (dtz + cnt50 <= 99 && !rep ? 1000 : 1000 - (dtz + cnt50))
                : dtz < 0 ? (-dtz * 2 + cnt50 < 100 ? -1000 : -1000 + (-dtz + cnt50))
                : 0;

-        m.TBRank = r;
+        m.tbRank = r;

 
         // Determine the score to be displayed for this move. Assign at least
         // 1 cp to cursed wins and let it grow to 49 cp as the positions gets
         // closer to a real win.
 
         // Determine the score to be displayed for this move. Assign at least
         // 1 cp to cursed wins and let it grow to 49 cp as the positions gets
         // closer to a real win.

-        m.TBScore =  r >= bound ? VALUE_MATE - MAX_PLY - 1
+        m.tbScore =  r >= bound ? VALUE_MATE - MAX_PLY - 1

                    : r >  0     ? Value((std::max( 3, r - 800) * int(PawnValueEg)) / 200)
                    : r == 0     ? VALUE_DRAW
                    : r > -bound ? Value((std::min(-3, r + 800) * int(PawnValueEg)) / 200)
                    : r >  0     ? Value((std::max( 3, r - 800) * int(PawnValueEg)) / 200)
                    : r == 0     ? VALUE_DRAW
                    : r > -bound ? Value((std::min(-3, r + 800) * int(PawnValueEg)) / 200)


@@ -1532,13 +1602,15 @@ bool Tablebases::root_probe_wdl(Position& pos, Search::RootMoves& rootMoves) {
         if (result == FAIL)
             return false;
 
         if (result == FAIL)
             return false;
 

-        m.TBRank = WDL_to_rank[wdl + 2];
+        m.tbRank = WDL_to_rank[wdl + 2];

 
         if (!rule50)
             wdl =  wdl > WDLDraw ? WDLWin
                  : wdl < WDLDraw ? WDLLoss : WDLDraw;
 
         if (!rule50)
             wdl =  wdl > WDLDraw ? WDLWin
                  : wdl < WDLDraw ? WDLLoss : WDLDraw;

-        m.TBScore = WDL_to_value[wdl + 2];
+        m.tbScore = WDL_to_value[wdl + 2];

     }
 
     return true;
 }
     }
 
     return true;
 }

+
+} // namespace Stockfish