X-Git-Url: https://git.sesse.net/?p=stockfish;a=blobdiff_plain;f=src%2Fmisc.cpp;h=fefa8d4f33cbd2dd7a74ff57f6a481e14468025b;hp=0fc151c51352600148848b4ef246669a9d623f5c;hb=215cd19108d97376284192c29790b42a0b0e618a;hpb=7f367e6019289e65b902a5ebf7e3686db5cefdef diff --git a/src/misc.cpp b/src/misc.cpp index 0fc151c5..6f908fd2 100644 --- a/src/misc.cpp +++ b/src/misc.cpp @@ -1,7 +1,8 @@ /* 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) 2015-2019 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 @@ -17,254 +18,307 @@ along with this program. If not, see . */ -#if !defined(_MSC_VER) - -# include -# include -# include -# if defined(__hpux) -# include -# endif - -#else - -#define _CRT_SECURE_NO_DEPRECATE -#include -#include +#ifdef _WIN32 +#if _WIN32_WINNT < 0x0601 +#undef _WIN32_WINNT +#define _WIN32_WINNT 0x0601 // Force to include needed API prototypes +#endif +#ifndef NOMINMAX +#define NOMINMAX #endif -#if !defined(NO_PREFETCH) -# include +#include +// 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 -#include +#include #include #include #include +#include -#include "bitcount.h" #include "misc.h" #include "thread.h" using namespace std; -/// Version number. If EngineVersion is left empty, then AppTag plus -/// current date (in the format YYMMDD) is used as a version number. +namespace { -static const string AppName = "Stockfish"; -static const string EngineVersion = ""; -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 -/// 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 ", depending -/// on whether the constant EngineVersion is empty. +struct Tie: public streambuf { // MSVC requires split streambuf for cin and cout -const string engine_name() { + Tie(streambuf* b, streambuf* l) : buf(b), logBuf(l) {} - const string months("Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec"); - const string cpu64(CpuIs64Bit ? " 64bit" : ""); + 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(), ">> "); } - if (!EngineVersion.empty()) - return AppName + " " + EngineVersion + cpu64; + streambuf *buf, *logBuf; - stringstream s, date(__DATE__); // From compiler, format is "Sep 21 2008" - string month, day, year; + int log(int c, const char* prefix) { - date >> month >> day >> year; + static int last = '\n'; // Single log file - s << setfill('0') << AppName + " " + AppTag + " " - << year.substr(2, 2) << setw(2) - << (1 + months.find(month) / 4) << setw(2) - << day << cpu64; + if (last == '\n') + logBuf->sputn(prefix, 3); - return s.str(); -} + return last = logBuf->sputc((char)c); + } +}; +class Logger { -/// Our brave developers! Required by UCI + Logger() : in(cin.rdbuf(), file.rdbuf()), out(cout.rdbuf(), file.rdbuf()) {} + ~Logger() { start(""); } -const string engine_authors() { + ofstream file; + Tie in, out; - return "Tord Romstad, Marco Costalba and Joona Kiiski"; -} +public: + static void start(const std::string& fname) { + + static Logger l; + if (!fname.empty() && !l.file.is_open()) + { + l.file.open(fname, ifstream::out); -/// Debug stuff. Helper functions used mainly for debugging purposes + if (!l.file.is_open()) + { + cerr << "Unable to open debug log file " << fname << endl; + exit(EXIT_FAILURE); + } -static uint64_t dbg_hit_cnt0; -static uint64_t dbg_hit_cnt1; -static uint64_t dbg_mean_cnt0; -static uint64_t dbg_mean_cnt1; + 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 - if (dbg_hit_cnt0) - cout << "Total " << dbg_hit_cnt0 << " Hit " << dbg_hit_cnt1 - << " hit rate (%) " << 100 * dbg_hit_cnt1 / (dbg_hit_cnt0 + 1) << endl; -} +/// engine_info() returns the full name of the current Stockfish version. This +/// will be either "Stockfish DD-MM-YY" (where DD-MM-YY is the date when +/// the program was compiled) or "Stockfish ", depending on whether +/// Version is empty. -void dbg_print_mean() { +const string engine_info(bool to_uci) { - if (dbg_mean_cnt0) - cout << "Total " << dbg_mean_cnt0 << " Mean " - << (float)dbg_mean_cnt1 / (dbg_mean_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" + + ss << "Stockfish " << Version << setfill('0'); -void dbg_mean_of(int v) { + if (Version.empty()) + { + date >> month >> day >> year; + ss << setw(2) << day << setw(2) << (1 + months.find(month) / 4) << year.substr(2); + } - dbg_mean_cnt0++; - dbg_mean_cnt1 += v; + ss << (Is64Bit ? " 64" : "") + << (HasPext ? " BMI2" : (HasPopCnt ? " POPCNT" : "")) + << (to_uci ? "\nid author ": " by ") + << "T. Romstad, M. Costalba, J. Kiiski, G. Linscott"; + + return ss.str(); } -void dbg_hit_on(bool b) { - dbg_hit_cnt0++; - if (b) - dbg_hit_cnt1++; +/// Debug functions used mainly to collect run-time statistics +static std::atomic hits[2], means[2]; + +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; } + +void dbg_print() { + + if (hits[0]) + cerr << "Total " << hits[0] << " Hits " << hits[1] + << " hit rate (%) " << 100 * hits[1] / hits[0] << endl; + + if (means[0]) + cerr << "Total " << means[0] << " Mean " + << (double)means[1] / means[0] << endl; } -void dbg_hit_on_c(bool c, bool b) { if (c) dbg_hit_on(b); } -void dbg_before() { dbg_hit_on(false); } -void dbg_after() { dbg_hit_on(true); dbg_hit_cnt0--; } +/// Used to serialize access to std::cout to avoid multiple threads writing at +/// the same time. -/// get_system_time() returns the current system time, measured in milliseconds +std::ostream& operator<<(std::ostream& os, SyncCout sc) { -int get_system_time() { + static std::mutex m; -#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 + if (sc == IO_LOCK) + m.lock(); + + if (sc == IO_UNLOCK) + m.unlock(); + + 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); } + + +/// 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 -int cpu_count() { +void prefetch(void*) {} -#if defined(_MSC_VER) - SYSTEM_INFO s; - GetSystemInfo(&s); - return Min(s.dwNumberOfProcessors, MAX_THREADS); #else -# if defined(_SC_NPROCESSORS_ONLN) - return Min(sysconf(_SC_NPROCESSORS_ONLN), MAX_THREADS); -# elif defined(__hpux) - struct pst_dynamic psd; - if (pstat_getdynamic(&psd, sizeof(psd), (size_t)1, 0) == -1) - return 1; - return Min(psd.psd_proc_cnt, MAX_THREADS); -# else - return 1; +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 -#endif +# if defined(__INTEL_COMPILER) || defined(_MSC_VER) + _mm_prefetch((char*)addr, _MM_HINT_T0); +# else + __builtin_prefetch(addr); +# endif } +#endif -/// Check for console input. Original code from Beowulf, Olithink and Greko +namespace WinProcGroup { #ifndef _WIN32 -int input_available() { +void bindThisThread(size_t) {} - fd_set readfds; - struct timeval timeout; +#else - FD_ZERO(&readfds); - FD_SET(fileno(stdin), &readfds); - timeout.tv_sec = 0; // Set to timeout immediately - timeout.tv_usec = 0; - select(16, &readfds, 0, 0, &timeout); +/// 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. - return (FD_ISSET(fileno(stdin), &readfds)); -} +int best_group(size_t idx) { -#else + 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; -int input_available() { + // First call to get returnLength. We expect it to fail due to null buffer + if (fun1(RelationAll, nullptr, &returnLength)) + return -1; - static HANDLE inh = NULL; - static bool usePipe = false; - INPUT_RECORD rec[256]; - DWORD nchars, recCnt; + // Once we know returnLength, allocate the buffer + SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX *buffer, *ptr; + ptr = buffer = (SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX*)malloc(returnLength); - if (!inh) + // Second call, now we expect to succeed + if (!fun1(RelationAll, buffer, &returnLength)) { - inh = GetStdHandle(STD_INPUT_HANDLE); - if (GetConsoleMode(inh, &nchars)) + free(buffer); + return -1; + } + + while (byteOffset < returnLength) + { + if (ptr->Relationship == RelationNumaNode) + nodes++; + + else if (ptr->Relationship == RelationProcessorCore) { - SetConsoleMode(inh, nchars & ~(ENABLE_MOUSE_INPUT | ENABLE_WINDOW_INPUT)); - FlushConsoleInputBuffer(inh); - } else - usePipe = true; + 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); } - // When using Standard C input functions, also check if there - // is anything in the buffer. After a call to such functions, - // the input waiting in the pipe will be copied to the buffer, - // and the call to PeekNamedPipe can indicate no input available. - // Setting stdin to unbuffered was not enough. [from Greko] - if (stdin->_cnt > 0) - return 1; - - // When running under a GUI the input commands are sent to us - // directly over the internal pipe. If PeekNamedPipe() returns 0 - // then something went wrong. Probably the parent program exited. - // Returning 1 will make the next call to the input function - // return EOF, where this should be catched then. - if (usePipe) - return PeekNamedPipe(inh, NULL, 0, NULL, &nchars, NULL) ? nchars : 1; - - // Count the number of unread input records, including keyboard, - // mouse, and window-resizing input records. - GetNumberOfConsoleInputEvents(inh, &nchars); - - // Read data from console without removing it from the buffer - if (nchars <= 0 || !PeekConsoleInput(inh, rec, Min(nchars, 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; + free(buffer); + + std::vector 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; } -#endif +/// bindThisThread() set the group affinity of the current thread -/// prefetch() preloads the given address in L1/L2 cache. This is a non -/// blocking function and do not stalls the CPU waiting for data to be -/// loaded from memory, that can be quite slow. -#if defined(NO_PREFETCH) +void bindThisThread(size_t idx) { -void prefetch(char*) {} + // Use only local variables to be thread-safe + int group = best_group(idx); -#else + if (group == -1) + return; -void prefetch(char* addr) { + // 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 defined(__INTEL_COMPILER) || defined(__ICL) - // This hack prevents prefetches to be optimized away by - // Intel compiler. Both MSVC and gcc seems not affected. - __asm__ (""); -#endif + if (!fun2 || !fun3) + return; - _mm_prefetch(addr, _MM_HINT_T2); - _mm_prefetch(addr+64, _MM_HINT_T2); // 64 bytes ahead + GROUP_AFFINITY affinity; + if (fun2(group, &affinity)) + fun3(GetCurrentThread(), &affinity, nullptr); } #endif + +} // namespace WinProcGroup