/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
- Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
+ 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
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
+#ifdef _WIN32
+#if _WIN32_WINNT < 0x0601
+#undef _WIN32_WINNT
+#define _WIN32_WINNT 0x0601 // Force to include needed API prototypes
+#endif
-////
-//// Includes
-////
-
-#if !defined(_MSC_VER)
-
-# include <sys/time.h>
-# include <sys/types.h>
-# include <unistd.h>
-# if defined(__hpux)
-# include <sys/pstat.h>
-# endif
-
-#else
+#ifndef NOMINMAX
+#define NOMINMAX
+#endif
-#define _CRT_SECURE_NO_DEPRECATE
#include <windows.h>
-#include <sys/timeb.h>
-
+// The needed Windows API for processor groups could be missed from old Windows
+// versions, so instead of calling them directly (forcing the linker to resolve
+// the calls at compile time), try to load them at runtime. To do this we need
+// first to define the corresponding function pointers.
+extern "C" {
+typedef bool(*fun1_t)(LOGICAL_PROCESSOR_RELATIONSHIP,
+ PSYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX, PDWORD);
+typedef bool(*fun2_t)(USHORT, PGROUP_AFFINITY);
+typedef bool(*fun3_t)(HANDLE, CONST GROUP_AFFINITY*, PGROUP_AFFINITY);
+}
#endif
-#include <cassert>
-#include <cstdio>
+#include <fstream>
#include <iomanip>
#include <iostream>
#include <sstream>
+#include <vector>
-#include "bitcount.h"
#include "misc.h"
#include "thread.h"
using namespace std;
-/// Version number. If this is left empty, the current date (in the format
-/// YYMMDD) is used as a version number.
+namespace {
-static const string EngineVersion = "";
-static const string AppName = "Stockfish";
-static const string AppTag = "";
+/// Version number. If Version is left empty, then compile date in the format
+/// DD-MM-YY and show in engine_info.
+const string Version = "";
+/// Our fancy logging facility. The trick here is to replace cin.rdbuf() and
+/// cout.rdbuf() with two Tie objects that tie cin and cout to a file stream. We
+/// can toggle the logging of std::cout and std:cin at runtime whilst preserving
+/// usual I/O functionality, all without changing a single line of code!
+/// Idea from http://groups.google.com/group/comp.lang.c++/msg/1d941c0f26ea0d81
-////
-//// Variables
-////
+struct Tie: public streambuf { // MSVC requires split streambuf for cin and cout
-bool Chess960;
+ Tie(streambuf* b, streambuf* l) : buf(b), logBuf(l) {}
-uint64_t dbg_cnt0 = 0;
-uint64_t dbg_cnt1 = 0;
+ int sync() override { return logBuf->pubsync(), buf->pubsync(); }
+ int overflow(int c) override { return log(buf->sputc((char)c), "<< "); }
+ int underflow() override { return buf->sgetc(); }
+ int uflow() override { return log(buf->sbumpc(), ">> "); }
-bool dbg_show_mean = false;
-bool dbg_show_hit_rate = false;
+ streambuf *buf, *logBuf;
+ int log(int c, const char* prefix) {
-////
-//// Functions
-////
+ static int last = '\n'; // Single log file
-void dbg_hit_on(bool b) {
+ if (last == '\n')
+ logBuf->sputn(prefix, 3);
- assert(!dbg_show_mean);
- dbg_show_hit_rate = true;
- dbg_cnt0++;
- if (b)
- dbg_cnt1++;
-}
+ return last = logBuf->sputc((char)c);
+ }
+};
-void dbg_hit_on_c(bool c, bool b) {
+class Logger {
- if (c)
- dbg_hit_on(b);
-}
+ Logger() : in(cin.rdbuf(), file.rdbuf()), out(cout.rdbuf(), file.rdbuf()) {}
+ ~Logger() { start(""); }
-void dbg_before() {
+ ofstream file;
+ Tie in, out;
- assert(!dbg_show_mean);
- dbg_show_hit_rate = true;
- dbg_cnt0++;
-}
+public:
+ static void start(const std::string& fname) {
-void dbg_after() {
+ static Logger l;
- assert(!dbg_show_mean);
- dbg_show_hit_rate = true;
- dbg_cnt1++;
-}
+ if (!fname.empty() && !l.file.is_open())
+ {
+ l.file.open(fname, ifstream::out);
-void dbg_mean_of(int v) {
+ if (!l.file.is_open())
+ {
+ cerr << "Unable to open debug log file " << fname << endl;
+ exit(EXIT_FAILURE);
+ }
- assert(!dbg_show_hit_rate);
- dbg_show_mean = true;
- dbg_cnt0++;
- dbg_cnt1 += v;
-}
+ cin.rdbuf(&l.in);
+ cout.rdbuf(&l.out);
+ }
+ else if (fname.empty() && l.file.is_open())
+ {
+ cout.rdbuf(l.out.buf);
+ cin.rdbuf(l.in.buf);
+ l.file.close();
+ }
+ }
+};
-void dbg_print_hit_rate() {
+} // namespace
- cout << "Total " << dbg_cnt0 << " Hit " << dbg_cnt1
- << " hit rate (%) " << (dbg_cnt1*100)/(dbg_cnt0 ? dbg_cnt0 : 1) << endl;
-}
+/// engine_info() returns the full name of the current Stockfish version. This
+/// will be either "Stockfish <Tag> DD-MM-YY" (where DD-MM-YY is the date when
+/// the program was compiled) or "Stockfish <Version>", depending on whether
+/// Version is empty.
-void dbg_print_mean() {
+const string engine_info(bool to_uci) {
- cout << "Total " << dbg_cnt0 << " Mean "
- << (float)dbg_cnt1 / (dbg_cnt0 ? dbg_cnt0 : 1) << endl;
-}
+ const string months("Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec");
+ string month, day, year;
+ stringstream ss, date(__DATE__); // From compiler, format is "Sep 21 2008"
-void dbg_print_hit_rate(ofstream& logFile) {
+ ss << "Stockfish " << Version << setfill('0');
- logFile << "Total " << dbg_cnt0 << " Hit " << dbg_cnt1
- << " hit rate (%) " << (dbg_cnt1*100)/(dbg_cnt0 ? dbg_cnt0 : 1) << endl;
-}
+ if (Version.empty())
+ {
+ date >> month >> day >> year;
+ ss << setw(2) << day << setw(2) << (1 + months.find(month) / 4) << year.substr(2);
+ }
-void dbg_print_mean(ofstream& logFile) {
+ ss << (Is64Bit ? " 64" : "")
+ << (HasPext ? " BMI2" : (HasPopCnt ? " POPCNT" : ""))
+ << (to_uci ? "\nid author ": " by ")
+ << "T. Romstad, M. Costalba, J. Kiiski, G. Linscott";
- logFile << "Total " << dbg_cnt0 << " Mean "
- << (float)dbg_cnt1 / (dbg_cnt0 ? dbg_cnt0 : 1) << endl;
+ return ss.str();
}
-/// engine_name() returns the full name of the current Stockfish version.
-/// This will be either "Stockfish YYMMDD" (where YYMMDD is the date when the
-/// program was compiled) or "Stockfish <version number>", depending on whether
-/// the constant EngineVersion (defined in misc.h) is empty.
-
-const string engine_name() {
- const string cpu64(CpuHas64BitPath ? " 64bit" : "");
+/// Debug functions used mainly to collect run-time statistics
+static std::atomic<int64_t> hits[2], means[2];
- if (!EngineVersion.empty())
- return AppName+ " " + EngineVersion + cpu64;
+void dbg_hit_on(bool b) { ++hits[0]; if (b) ++hits[1]; }
+void dbg_hit_on(bool c, bool b) { if (c) dbg_hit_on(b); }
+void dbg_mean_of(int v) { ++means[0]; means[1] += v; }
- string date(__DATE__); // From compiler, format is "Sep 21 2008"
- string months("Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec");
+void dbg_print() {
- size_t mon = 1 + months.find(date.substr(0, 3)) / 4;
+ if (hits[0])
+ cerr << "Total " << hits[0] << " Hits " << hits[1]
+ << " hit rate (%) " << 100 * hits[1] / hits[0] << endl;
- stringstream s;
- string day = (date[4] == ' ' ? date.substr(5, 1) : date.substr(4, 2));
+ if (means[0])
+ cerr << "Total " << means[0] << " Mean "
+ << (double)means[1] / means[0] << endl;
+}
- string name = AppName + " " + AppTag + " ";
- s << name << date.substr(date.length() - 2) << setfill('0')
- << setw(2) << mon << setw(2) << day << cpu64;
+/// Used to serialize access to std::cout to avoid multiple threads writing at
+/// the same time.
- return s.str();
-}
+std::ostream& operator<<(std::ostream& os, SyncCout sc) {
+ static std::mutex m;
-/// get_system_time() returns the current system time, measured in
-/// milliseconds.
+ if (sc == IO_LOCK)
+ m.lock();
-int get_system_time() {
+ if (sc == IO_UNLOCK)
+ m.unlock();
-#if defined(_MSC_VER)
- struct _timeb t;
- _ftime(&t);
- return int(t.time*1000 + t.millitm);
-#else
- struct timeval t;
- gettimeofday(&t, NULL);
- return t.tv_sec*1000 + t.tv_usec/1000;
-#endif
+ return os;
}
-/// cpu_count() tries to detect the number of CPU cores.
+/// Trampoline helper to avoid moving Logger to misc.h
+void start_logger(const std::string& fname) { Logger::start(fname); }
-#if !defined(_MSC_VER)
-# if defined(_SC_NPROCESSORS_ONLN)
-int cpu_count() {
- return Min(sysconf(_SC_NPROCESSORS_ONLN), MAX_THREADS);
-}
-# elif defined(__hpux)
-int cpu_count() {
- struct pst_dynamic psd;
- if (pstat_getdynamic(&psd, sizeof(psd), (size_t)1, 0) == -1)
- return 1;
+/// prefetch() preloads the given address in L1/L2 cache. This is a non-blocking
+/// function that doesn't stall the CPU waiting for data to be loaded from memory,
+/// which can be quite slow.
+#ifdef NO_PREFETCH
- return Min(psd.psd_proc_cnt, MAX_THREADS);
-}
-# else
-int cpu_count() {
- return 1;
-}
-# endif
+void prefetch(void*) {}
#else
-int cpu_count() {
- SYSTEM_INFO s;
- GetSystemInfo(&s);
- return Min(s.dwNumberOfProcessors, MAX_THREADS);
+void prefetch(void* addr) {
+
+# if defined(__INTEL_COMPILER)
+ // This hack prevents prefetches from being optimized away by
+ // Intel compiler. Both MSVC and gcc seem not be affected by this.
+ __asm__ ("");
+# endif
+
+# if defined(__INTEL_COMPILER) || defined(_MSC_VER)
+ _mm_prefetch((char*)addr, _MM_HINT_T0);
+# else
+ __builtin_prefetch(addr);
+# endif
}
#endif
+namespace WinProcGroup {
-/*
- From Beowulf, from Olithink
-*/
#ifndef _WIN32
-/* Non-windows version */
-int Bioskey()
-{
- fd_set readfds;
- struct timeval timeout;
-
- FD_ZERO(&readfds);
- FD_SET(fileno(stdin), &readfds);
- /* Set to timeout immediately */
- timeout.tv_sec = 0;
- timeout.tv_usec = 0;
- select(16, &readfds, 0, 0, &timeout);
-
- return (FD_ISSET(fileno(stdin), &readfds));
-}
+
+void bindThisThread(size_t) {}
#else
-/* Windows-version */
-#include <windows.h>
-#include <conio.h>
-int Bioskey()
-{
- static int init = 0,
- pipe;
- static HANDLE inh;
- DWORD dw;
- /* If we're running under XBoard then we can't use _kbhit() as the input
- * commands are sent to us directly over the internal pipe */
-
-#if defined(FILE_CNT)
- if (stdin->_cnt > 0)
- return stdin->_cnt;
-#endif
- if (!init) {
- init = 1;
- inh = GetStdHandle(STD_INPUT_HANDLE);
- pipe = !GetConsoleMode(inh, &dw);
- if (!pipe) {
- SetConsoleMode(inh, dw & ~(ENABLE_MOUSE_INPUT | ENABLE_WINDOW_INPUT));
- FlushConsoleInputBuffer(inh);
- }
- }
- if (pipe) {
- if (!PeekNamedPipe(inh, NULL, 0, NULL, &dw, NULL))
- return 1;
- return dw;
- } else {
- // Count the number of unread input records, including keyboard,
- // mouse, and window-resizing input records.
- GetNumberOfConsoleInputEvents(inh, &dw);
- if (dw <= 0)
- return 0;
-
- // Read data from console without removing it from the buffer
- INPUT_RECORD rec[256];
- DWORD recCnt;
- if (!PeekConsoleInput(inh, rec, Min(dw, 256), &recCnt))
- return 0;
-
- // Search for at least one keyboard event
- for (DWORD i = 0; i < recCnt; i++)
- if (rec[i].EventType == KEY_EVENT)
- return 1;
-
- return 0;
- }
+
+/// best_group() retrieves logical processor information using Windows specific
+/// API and returns the best group id for the thread with index idx. Original
+/// code from Texel by Peter Österlund.
+
+int best_group(size_t idx) {
+
+ int threads = 0;
+ int nodes = 0;
+ int cores = 0;
+ DWORD returnLength = 0;
+ DWORD byteOffset = 0;
+
+ // Early exit if the needed API is not available at runtime
+ HMODULE k32 = GetModuleHandle("Kernel32.dll");
+ auto fun1 = (fun1_t)(void(*)())GetProcAddress(k32, "GetLogicalProcessorInformationEx");
+ if (!fun1)
+ return -1;
+
+ // First call to get returnLength. We expect it to fail due to null buffer
+ if (fun1(RelationAll, nullptr, &returnLength))
+ return -1;
+
+ // Once we know returnLength, allocate the buffer
+ SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX *buffer, *ptr;
+ ptr = buffer = (SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX*)malloc(returnLength);
+
+ // Second call, now we expect to succeed
+ if (!fun1(RelationAll, buffer, &returnLength))
+ {
+ free(buffer);
+ return -1;
+ }
+
+ while (byteOffset < returnLength)
+ {
+ if (ptr->Relationship == RelationNumaNode)
+ nodes++;
+
+ else if (ptr->Relationship == RelationProcessorCore)
+ {
+ cores++;
+ threads += (ptr->Processor.Flags == LTP_PC_SMT) ? 2 : 1;
+ }
+
+ assert(ptr->Size);
+ byteOffset += ptr->Size;
+ ptr = (SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX*)(((char*)ptr) + ptr->Size);
+ }
+
+ free(buffer);
+
+ std::vector<int> groups;
+
+ // Run as many threads as possible on the same node until core limit is
+ // reached, then move on filling the next node.
+ for (int n = 0; n < nodes; n++)
+ for (int i = 0; i < cores / nodes; i++)
+ groups.push_back(n);
+
+ // In case a core has more than one logical processor (we assume 2) and we
+ // have still threads to allocate, then spread them evenly across available
+ // nodes.
+ for (int t = 0; t < threads - cores; t++)
+ groups.push_back(t % nodes);
+
+ // If we still have more threads than the total number of logical processors
+ // then return -1 and let the OS to decide what to do.
+ return idx < groups.size() ? groups[idx] : -1;
+}
+
+
+/// bindThisThread() set the group affinity of the current thread
+
+void bindThisThread(size_t idx) {
+
+ // Use only local variables to be thread-safe
+ int group = best_group(idx);
+
+ if (group == -1)
+ return;
+
+ // Early exit if the needed API are not available at runtime
+ HMODULE k32 = GetModuleHandle("Kernel32.dll");
+ auto fun2 = (fun2_t)(void(*)())GetProcAddress(k32, "GetNumaNodeProcessorMaskEx");
+ auto fun3 = (fun3_t)(void(*)())GetProcAddress(k32, "SetThreadGroupAffinity");
+
+ if (!fun2 || !fun3)
+ return;
+
+ GROUP_AFFINITY affinity;
+ if (fun2(group, &affinity))
+ fun3(GetCurrentThread(), &affinity, nullptr);
}
+
#endif
+
+} // namespace WinProcGroup
projects / stockfish / blobdiff