Files
Stockfish/src/misc.h
Tomasz Sobczyk b748b46714 Cleanup and simplify NNUE code.
A lot of optimizations happend since the NNUE was introduced
and since then some parts of the code were left unused. This
got to the point where asserts were have to be made just to
let people know that modifying something will not have any
effects or may even break everything due to the assumptions
being made. Removing these parts removes those inexisting
"false dependencies". Additionally:

 * append_changed_indices now takes the king pos and stateinfo
   explicitly, no more misleading pos parameter
 * IndexList is removed in favor of a generic ValueList.
   Feature transformer just instantiates the type it needs.
 * The update cost and refresh requirement is deferred to the
   feature set once again, but now doesn't go through the whole
   FeatureSet machinery and just calls HalfKP directly.
 * accumulator no longer has a singular dimension.
 * The PS constants and the PieceSquareIndex array are made local
   to the HalfKP feature set because they are specific to it and
   DO differ for other feature sets.
 * A few names are changed to more descriptive

Passed STC non-regression:
https://tests.stockfishchess.org/tests/view/608421dd95e7f1852abd2790
LLR: 2.95 (-2.94,2.94) <-2.50,0.50>
Total: 180008 W: 16186 L: 16258 D: 147564
Ptnml(0-2): 587, 12593, 63725, 12503, 596

closes https://github.com/official-stockfish/Stockfish/pull/3441

No functional change
2021-04-25 13:16:30 +02:00

194 lines
5.8 KiB
C++

/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
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
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/>.
*/
#ifndef MISC_H_INCLUDED
#define MISC_H_INCLUDED
#include <cassert>
#include <chrono>
#include <ostream>
#include <string>
#include <vector>
#include <cstdint>
#include "types.h"
namespace Stockfish {
std::string engine_info(bool to_uci = false);
std::string compiler_info();
void prefetch(void* addr);
void start_logger(const std::string& fname);
void* std_aligned_alloc(size_t alignment, size_t size);
void std_aligned_free(void* ptr);
void* aligned_large_pages_alloc(size_t size); // memory aligned by page size, min alignment: 4096 bytes
void aligned_large_pages_free(void* mem); // nop if mem == nullptr
void dbg_hit_on(bool b);
void dbg_hit_on(bool c, bool b);
void dbg_mean_of(int v);
void dbg_print();
typedef std::chrono::milliseconds::rep TimePoint; // A value in milliseconds
static_assert(sizeof(TimePoint) == sizeof(int64_t), "TimePoint should be 64 bits");
inline TimePoint now() {
return std::chrono::duration_cast<std::chrono::milliseconds>
(std::chrono::steady_clock::now().time_since_epoch()).count();
}
template<class Entry, int Size>
struct HashTable {
Entry* operator[](Key key) { return &table[(uint32_t)key & (Size - 1)]; }
private:
std::vector<Entry> table = std::vector<Entry>(Size); // Allocate on the heap
};
enum SyncCout { IO_LOCK, IO_UNLOCK };
std::ostream& operator<<(std::ostream&, SyncCout);
#define sync_cout std::cout << IO_LOCK
#define sync_endl std::endl << IO_UNLOCK
// `ptr` must point to an array of size at least
// `sizeof(T) * N + alignment` bytes, where `N` is the
// number of elements in the array.
template <uintptr_t Alignment, typename T>
T* align_ptr_up(T* ptr)
{
static_assert(alignof(T) < Alignment);
const uintptr_t ptrint = reinterpret_cast<uintptr_t>(reinterpret_cast<char*>(ptr));
return reinterpret_cast<T*>(reinterpret_cast<char*>((ptrint + (Alignment - 1)) / Alignment * Alignment));
}
template <typename T>
class ValueListInserter {
public:
ValueListInserter(T* v, std::size_t& s) :
values(v),
size(&s)
{
}
void push_back(const T& value) { values[(*size)++] = value; }
private:
T* values;
std::size_t* size;
};
template <typename T, std::size_t MaxSize>
class ValueList {
public:
std::size_t size() const { return size_; }
void resize(std::size_t newSize) { size_ = newSize; }
void push_back(const T& value) { values_[size_++] = value; }
T& operator[](std::size_t index) { return values_[index]; }
T* begin() { return values_; }
T* end() { return values_ + size_; }
const T& operator[](std::size_t index) const { return values_[index]; }
const T* begin() const { return values_; }
const T* end() const { return values_ + size_; }
operator ValueListInserter<T>() { return ValueListInserter(values_, size_); }
void swap(ValueList& other) {
const std::size_t maxSize = std::max(size_, other.size_);
for (std::size_t i = 0; i < maxSize; ++i) {
std::swap(values_[i], other.values_[i]);
}
std::swap(size_, other.size_);
}
private:
T values_[MaxSize];
std::size_t size_ = 0;
};
/// xorshift64star Pseudo-Random Number Generator
/// This class is based on original code written and dedicated
/// to the public domain by Sebastiano Vigna (2014).
/// It has the following characteristics:
///
/// - Outputs 64-bit numbers
/// - Passes Dieharder and SmallCrush test batteries
/// - Does not require warm-up, no zeroland to escape
/// - Internal state is a single 64-bit integer
/// - Period is 2^64 - 1
/// - Speed: 1.60 ns/call (Core i7 @3.40GHz)
///
/// For further analysis see
/// <http://vigna.di.unimi.it/ftp/papers/xorshift.pdf>
class PRNG {
uint64_t s;
uint64_t rand64() {
s ^= s >> 12, s ^= s << 25, s ^= s >> 27;
return s * 2685821657736338717LL;
}
public:
PRNG(uint64_t seed) : s(seed) { assert(seed); }
template<typename T> T rand() { return T(rand64()); }
/// Special generator used to fast init magic numbers.
/// Output values only have 1/8th of their bits set on average.
template<typename T> T sparse_rand()
{ return T(rand64() & rand64() & rand64()); }
};
inline uint64_t mul_hi64(uint64_t a, uint64_t b) {
#if defined(__GNUC__) && defined(IS_64BIT)
__extension__ typedef unsigned __int128 uint128;
return ((uint128)a * (uint128)b) >> 64;
#else
uint64_t aL = (uint32_t)a, aH = a >> 32;
uint64_t bL = (uint32_t)b, bH = b >> 32;
uint64_t c1 = (aL * bL) >> 32;
uint64_t c2 = aH * bL + c1;
uint64_t c3 = aL * bH + (uint32_t)c2;
return aH * bH + (c2 >> 32) + (c3 >> 32);
#endif
}
/// Under Windows it is not possible for a process to run on more than one
/// logical processor group. This usually means to be limited to use max 64
/// cores. To overcome this, some special platform specific API should be
/// called to set group affinity for each thread. Original code from Texel by
/// Peter Österlund.
namespace WinProcGroup {
void bindThisThread(size_t idx);
}
namespace CommandLine {
void init(int argc, char* argv[]);
extern std::string binaryDirectory; // path of the executable directory
extern std::string workingDirectory; // path of the working directory
}
} // namespace Stockfish
#endif // #ifndef MISC_H_INCLUDED