mirror of
https://github.com/official-stockfish/Stockfish.git
synced 2026-07-22 12:47:08 +00:00
Sami KiminkiandJoost VandeVondele
39437f4e55
Align the TT allocation by 2M to make it huge page friendly and advise the
kernel to use huge pages.
Benchmarks on my i7-8700K (6C/12T) box: (3 runs per bench per config)
vanilla (nps) hugepages (nps) avg
==================================================================================
bench | 3012490 3024364 3036331 3071052 3067544 3071052 +1.5%
bench 16 12 20 | 19237932 19050166 19085315 19266346 19207025 19548758 +1.1%
bench 16384 12 20 | 18182313 18371581 18336838 19381275 19738012 19620225 +7.0%
On my box, huge pages have a significant perf impact when using a big
hash size. They also speed up TT initialization big time:
vanilla (s) huge pages (s) speed-up
=======================================================================
time stockfish bench 16384 1 1 | 5.37 1.48 3.6x
In practice, huge pages with auto-defrag may always be enabled in the
system, in which case this patch has no effect. This
depends on the values in /sys/kernel/mm/transparent_hugepage/enabled
and /sys/kernel/mm/transparent_hugepage/defrag.
closes https://github.com/official-stockfish/Stockfish/pull/2463
No functional change
116 lines
3.5 KiB
C++
116 lines
3.5 KiB
C++
/*
|
|
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) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, 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/>.
|
|
*/
|
|
|
|
#ifndef MISC_H_INCLUDED
|
|
#define MISC_H_INCLUDED
|
|
|
|
#include <cassert>
|
|
#include <chrono>
|
|
#include <ostream>
|
|
#include <string>
|
|
#include <vector>
|
|
|
|
#include "types.h"
|
|
|
|
const std::string engine_info(bool to_uci = false);
|
|
const std::string compiler_info();
|
|
void prefetch(void* addr);
|
|
void start_logger(const std::string& fname);
|
|
void* aligned_ttmem_alloc(size_t size, void** mem);
|
|
|
|
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
|
|
|
|
|
|
/// 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()); }
|
|
};
|
|
|
|
|
|
/// 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);
|
|
}
|
|
|
|
#endif // #ifndef MISC_H_INCLUDED
|