X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=src%2Fmisc.cpp;h=d8895d87e53e5928c598768ce5f695c45195ff44;hb=487c21b1aa64dcc09dd95b845a66f39ae3c3754e;hp=4ff52f48288f0f50834809f5fb032835efb69111;hpb=4004bcf3a48c47450344e93fcf186e3eb5d32165;p=stockfish diff --git a/src/misc.cpp b/src/misc.cpp index 4ff52f48..d8895d87 100644 --- a/src/misc.cpp +++ b/src/misc.cpp @@ -1,6 +1,6 @@ /* Stockfish, a UCI chess playing engine derived from Glaurung 2.1 - Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file) + Copyright (C) 2004-2023 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 @@ -32,35 +32,48 @@ // 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); +using fun1_t = bool(*)(LOGICAL_PROCESSOR_RELATIONSHIP, + PSYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX, PDWORD); +using fun2_t = bool(*)(USHORT, PGROUP_AFFINITY); +using fun3_t = bool(*)(HANDLE, CONST GROUP_AFFINITY*, PGROUP_AFFINITY); +using fun4_t = bool(*)(USHORT, PGROUP_AFFINITY, USHORT, PUSHORT); +using fun5_t = WORD(*)(); +using fun6_t = bool(*)(HANDLE, DWORD, PHANDLE); +using fun7_t = bool(*)(LPCSTR, LPCSTR, PLUID); +using fun8_t = bool(*)(HANDLE, BOOL, PTOKEN_PRIVILEGES, DWORD, PTOKEN_PRIVILEGES, PDWORD); } #endif +#include +#include #include #include #include #include +#include #include -#include #if defined(__linux__) && !defined(__ANDROID__) #include #include #endif +#if defined(__APPLE__) || defined(__ANDROID__) || defined(__OpenBSD__) || (defined(__GLIBCXX__) && !defined(_GLIBCXX_HAVE_ALIGNED_ALLOC) && !defined(_WIN32)) || defined(__e2k__) +#define POSIXALIGNEDALLOC +#include +#endif + #include "misc.h" #include "thread.h" using namespace std; +namespace Stockfish { + namespace { -/// Version number. If Version is left empty, then compile date in the format -/// DD-MM-YY and show in engine_info. -const string Version = ""; +/// Version number or dev. +constexpr string_view version = "dev"; /// 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 @@ -103,7 +116,14 @@ public: static Logger l; - if (!fname.empty() && !l.file.is_open()) + if (l.file.is_open()) + { + cout.rdbuf(l.out.buf); + cin.rdbuf(l.in.buf); + l.file.close(); + } + + if (!fname.empty()) { l.file.open(fname, ifstream::out); @@ -116,34 +136,40 @@ public: 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(); - } } }; } // namespace -/// 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. - -const string engine_info(bool to_uci) { - 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" +/// engine_info() returns the full name of the current Stockfish version. +/// For local dev compiles we try to append the commit sha and commit date +/// from git if that fails only the local compilation date is set and "nogit" is specified: +/// Stockfish dev-YYYYMMDD-SHA +/// or +/// Stockfish dev-YYYYMMDD-nogit +/// +/// For releases (non dev builds) we only include the version number: +/// Stockfish version - ss << "Stockfish " << Version << setfill('0'); +string engine_info(bool to_uci) { + stringstream ss; + ss << "Stockfish " << version << setfill('0'); - if (Version.empty()) + if constexpr (version == "dev") { + ss << "-"; + #ifdef GIT_DATE + ss << stringify(GIT_DATE); + #else + constexpr string_view months("Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec"); + string month, day, year; + stringstream date(__DATE__); // From compiler, format is "Sep 21 2008" + date >> month >> day >> year; - ss << setw(2) << day << setw(2) << (1 + months.find(month) / 4) << year.substr(2); + ss << year << setw(2) << setfill('0') << (1 + months.find(month) / 4) << setw(2) << setfill('0') << day; + #endif + ss << "-asn"; } @@ -156,10 +182,8 @@ const string engine_info(bool to_uci) { /// compiler_info() returns a string trying to describe the compiler we use -const std::string compiler_info() { +std::string compiler_info() { - #define stringify2(x) #x - #define stringify(x) stringify2(x) #define make_version_string(major, minor, patch) stringify(major) "." stringify(minor) "." stringify(patch) /// Predefined macros hell: @@ -185,6 +209,18 @@ const std::string compiler_info() { compiler += "(version "; compiler += stringify(_MSC_FULL_VER) "." stringify(_MSC_BUILD); compiler += ")"; + #elif defined(__e2k__) && defined(__LCC__) + #define dot_ver2(n) \ + compiler += (char)'.'; \ + compiler += (char)('0' + (n) / 10); \ + compiler += (char)('0' + (n) % 10); + + compiler += "MCST LCC "; + compiler += "(version "; + compiler += std::to_string(__LCC__ / 100); + dot_ver2(__LCC__ % 100) + dot_ver2(__LCC_MINOR__) + compiler += ")"; #elif __GNUC__ compiler += "g++ (GNUC) "; compiler += make_version_string(__GNUC__, __GNUC_MINOR__, __GNUC_PATCHLEVEL__); @@ -215,26 +251,33 @@ const std::string compiler_info() { compiler += "\nCompilation settings include: "; compiler += (Is64Bit ? " 64bit" : " 32bit"); + #if defined(USE_VNNI) + compiler += " VNNI"; + #endif #if defined(USE_AVX512) compiler += " AVX512"; #endif + compiler += (HasPext ? " BMI2" : ""); #if defined(USE_AVX2) compiler += " AVX2"; #endif - #if defined(USE_SSE42) - compiler += " SSE42"; - #endif #if defined(USE_SSE41) compiler += " SSE41"; #endif #if defined(USE_SSSE3) compiler += " SSSE3"; #endif - #if defined(USE_SSE3) - compiler += " SSE3"; + #if defined(USE_SSE2) + compiler += " SSE2"; + #endif + compiler += (HasPopCnt ? " POPCNT" : ""); + #if defined(USE_MMX) + compiler += " MMX"; + #endif + #if defined(USE_NEON) + compiler += " NEON"; #endif - compiler += (HasPext ? " BMI2" : ""); - compiler += (HasPopCnt ? " POPCNT" : ""); + #if !defined(NDEBUG) compiler += " DEBUG"; #endif @@ -252,21 +295,94 @@ const std::string compiler_info() { /// Debug functions used mainly to collect run-time statistics -static std::atomic hits[2], means[2]; +constexpr int MaxDebugSlots = 32; + +namespace { + +template +struct DebugInfo { + std::atomic data[N] = { 0 }; + + constexpr inline std::atomic& operator[](int index) { return data[index]; } +}; + +DebugInfo<2> hit[MaxDebugSlots]; +DebugInfo<2> mean[MaxDebugSlots]; +DebugInfo<3> stdev[MaxDebugSlots]; +DebugInfo<6> correl[MaxDebugSlots]; + +} // namespace + +void dbg_hit_on(bool cond, int slot) { + + ++hit[slot][0]; + if (cond) + ++hit[slot][1]; +} + +void dbg_mean_of(int64_t value, int slot) { + + ++mean[slot][0]; + mean[slot][1] += value; +} -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_stdev_of(int64_t value, int slot) { + + ++stdev[slot][0]; + stdev[slot][1] += value; + stdev[slot][2] += value * value; +} + +void dbg_correl_of(int64_t value1, int64_t value2, int slot) { + + ++correl[slot][0]; + correl[slot][1] += value1; + correl[slot][2] += value1 * value1; + correl[slot][3] += value2; + correl[slot][4] += value2 * value2; + correl[slot][5] += value1 * value2; +} void dbg_print() { - if (hits[0]) - cerr << "Total " << hits[0] << " Hits " << hits[1] - << " hit rate (%) " << 100 * hits[1] / hits[0] << endl; + int64_t n; + auto E = [&n](int64_t x) { return double(x) / n; }; + auto sqr = [](double x) { return x * x; }; + + for (int i = 0; i < MaxDebugSlots; ++i) + if ((n = hit[i][0])) + std::cerr << "Hit #" << i + << ": Total " << n << " Hits " << hit[i][1] + << " Hit Rate (%) " << 100.0 * E(hit[i][1]) + << std::endl; + + for (int i = 0; i < MaxDebugSlots; ++i) + if ((n = mean[i][0])) + { + std::cerr << "Mean #" << i + << ": Total " << n << " Mean " << E(mean[i][1]) + << std::endl; + } - if (means[0]) - cerr << "Total " << means[0] << " Mean " - << (double)means[1] / means[0] << endl; + for (int i = 0; i < MaxDebugSlots; ++i) + if ((n = stdev[i][0])) + { + double r = sqrtl(E(stdev[i][2]) - sqr(E(stdev[i][1]))); + std::cerr << "Stdev #" << i + << ": Total " << n << " Stdev " << r + << std::endl; + } + + for (int i = 0; i < MaxDebugSlots; ++i) + if ((n = correl[i][0])) + { + double r = (E(correl[i][5]) - E(correl[i][1]) * E(correl[i][3])) + / ( sqrtl(E(correl[i][2]) - sqr(E(correl[i][1]))) + * sqrtl(E(correl[i][4]) - sqr(E(correl[i][3])))); + std::cerr << "Correl. #" << i + << ": Total " << n << " Coefficient " << r + << std::endl; + } } @@ -317,49 +433,47 @@ void prefetch(void* addr) { #endif -/// Wrappers for systems where the c++17 implementation doesn't guarantee the availability of aligned_alloc. -/// Memory allocated with std_aligned_alloc must be freed with std_aligned_free. -/// + +/// std_aligned_alloc() is our wrapper for systems where the c++17 implementation +/// does not guarantee the availability of aligned_alloc(). Memory allocated with +/// std_aligned_alloc() must be freed with std_aligned_free(). void* std_aligned_alloc(size_t alignment, size_t size) { -#if defined(__APPLE__) - return aligned_alloc(alignment, size); -#elif defined(_WIN32) + +#if defined(POSIXALIGNEDALLOC) + void *mem; + return posix_memalign(&mem, alignment, size) ? nullptr : mem; +#elif defined(_WIN32) && !defined(_M_ARM) && !defined(_M_ARM64) return _mm_malloc(size, alignment); +#elif defined(_WIN32) + return _aligned_malloc(size, alignment); #else return std::aligned_alloc(alignment, size); #endif } void std_aligned_free(void* ptr) { -#if defined(__APPLE__) + +#if defined(POSIXALIGNEDALLOC) free(ptr); -#elif defined(_WIN32) +#elif defined(_WIN32) && !defined(_M_ARM) && !defined(_M_ARM64) _mm_free(ptr); +#elif defined(_WIN32) + _aligned_free(ptr); #else free(ptr); #endif } -/// aligned_ttmem_alloc() will return suitably aligned memory, and if possible use large pages. -/// The returned pointer is the aligned one, while the mem argument is the one that needs -/// to be passed to free. With c++17 some of this functionality could be simplified. +/// aligned_large_pages_alloc() will return suitably aligned memory, if possible using large pages. -#if defined(__linux__) && !defined(__ANDROID__) +#if defined(_WIN32) -void* aligned_ttmem_alloc(size_t allocSize, void*& mem) { +static void* aligned_large_pages_alloc_windows([[maybe_unused]] size_t allocSize) { - constexpr size_t alignment = 2 * 1024 * 1024; // assumed 2MB page sizes - size_t size = ((allocSize + alignment - 1) / alignment) * alignment; // multiple of alignment - if (posix_memalign(&mem, alignment, size)) - mem = nullptr; - madvise(mem, allocSize, MADV_HUGEPAGE); - return mem; -} - -#elif defined(_WIN64) - -static void* aligned_ttmem_alloc_large_pages(size_t allocSize) { + #if !defined(_WIN64) + return nullptr; + #else HANDLE hProcessToken { }; LUID luid { }; @@ -369,11 +483,30 @@ static void* aligned_ttmem_alloc_large_pages(size_t allocSize) { if (!largePageSize) return nullptr; - // We need SeLockMemoryPrivilege, so try to enable it for the process - if (!OpenProcessToken(GetCurrentProcess(), TOKEN_ADJUST_PRIVILEGES | TOKEN_QUERY, &hProcessToken)) + // Dynamically link OpenProcessToken, LookupPrivilegeValue and AdjustTokenPrivileges + + HMODULE hAdvapi32 = GetModuleHandle(TEXT("advapi32.dll")); + + if (!hAdvapi32) + hAdvapi32 = LoadLibrary(TEXT("advapi32.dll")); + + auto fun6 = (fun6_t)(void(*)())GetProcAddress(hAdvapi32, "OpenProcessToken"); + if (!fun6) return nullptr; + auto fun7 = (fun7_t)(void(*)())GetProcAddress(hAdvapi32, "LookupPrivilegeValueA"); + if (!fun7) + return nullptr; + auto fun8 = (fun8_t)(void(*)())GetProcAddress(hAdvapi32, "AdjustTokenPrivileges"); + if (!fun8) + return nullptr; + + // We need SeLockMemoryPrivilege, so try to enable it for the process + if (!fun6( // OpenProcessToken() + GetCurrentProcess(), TOKEN_ADJUST_PRIVILEGES | TOKEN_QUERY, &hProcessToken)) + return nullptr; - if (LookupPrivilegeValue(NULL, SE_LOCK_MEMORY_NAME, &luid)) + if (fun7( // LookupPrivilegeValue(nullptr, SE_LOCK_MEMORY_NAME, &luid) + nullptr, "SeLockMemoryPrivilege", &luid)) { TOKEN_PRIVILEGES tp { }; TOKEN_PRIVILEGES prevTp { }; @@ -385,83 +518,82 @@ static void* aligned_ttmem_alloc_large_pages(size_t allocSize) { // Try to enable SeLockMemoryPrivilege. Note that even if AdjustTokenPrivileges() succeeds, // we still need to query GetLastError() to ensure that the privileges were actually obtained. - if (AdjustTokenPrivileges( + if (fun8( // AdjustTokenPrivileges() hProcessToken, FALSE, &tp, sizeof(TOKEN_PRIVILEGES), &prevTp, &prevTpLen) && GetLastError() == ERROR_SUCCESS) { // Round up size to full pages and allocate allocSize = (allocSize + largePageSize - 1) & ~size_t(largePageSize - 1); mem = VirtualAlloc( - NULL, allocSize, MEM_RESERVE | MEM_COMMIT | MEM_LARGE_PAGES, PAGE_READWRITE); + nullptr, allocSize, MEM_RESERVE | MEM_COMMIT | MEM_LARGE_PAGES, PAGE_READWRITE); // Privilege no longer needed, restore previous state - AdjustTokenPrivileges(hProcessToken, FALSE, &prevTp, 0, NULL, NULL); + fun8( // AdjustTokenPrivileges () + hProcessToken, FALSE, &prevTp, 0, nullptr, nullptr); } } CloseHandle(hProcessToken); return mem; -} -void* aligned_ttmem_alloc(size_t allocSize, void*& mem) { + #endif +} - static bool firstCall = true; +void* aligned_large_pages_alloc(size_t allocSize) { // Try to allocate large pages - mem = aligned_ttmem_alloc_large_pages(allocSize); - - // Suppress info strings on the first call. The first call occurs before 'uci' - // is received and in that case this output confuses some GUIs. - if (!firstCall) - { - if (mem) - sync_cout << "info string Hash table allocation: Windows large pages used." << sync_endl; - else - sync_cout << "info string Hash table allocation: Windows large pages not used." << sync_endl; - } - firstCall = false; + void* mem = aligned_large_pages_alloc_windows(allocSize); // Fall back to regular, page aligned, allocation if necessary if (!mem) - mem = VirtualAlloc(NULL, allocSize, MEM_RESERVE | MEM_COMMIT, PAGE_READWRITE); + mem = VirtualAlloc(nullptr, allocSize, MEM_RESERVE | MEM_COMMIT, PAGE_READWRITE); return mem; } #else -void* aligned_ttmem_alloc(size_t allocSize, void*& mem) { +void* aligned_large_pages_alloc(size_t allocSize) { - constexpr size_t alignment = 64; // assumed cache line size - size_t size = allocSize + alignment - 1; // allocate some extra space - mem = malloc(size); - void* ret = reinterpret_cast((uintptr_t(mem) + alignment - 1) & ~uintptr_t(alignment - 1)); - return ret; +#if defined(__linux__) + constexpr size_t alignment = 2 * 1024 * 1024; // assumed 2MB page size +#else + constexpr size_t alignment = 4096; // assumed small page size +#endif + + // round up to multiples of alignment + size_t size = ((allocSize + alignment - 1) / alignment) * alignment; + void *mem = std_aligned_alloc(alignment, size); +#if defined(MADV_HUGEPAGE) + madvise(mem, size, MADV_HUGEPAGE); +#endif + return mem; } #endif -/// aligned_ttmem_free() will free the previously allocated ttmem +/// aligned_large_pages_free() will free the previously allocated ttmem -#if defined(_WIN64) +#if defined(_WIN32) -void aligned_ttmem_free(void* mem) { +void aligned_large_pages_free(void* mem) { if (mem && !VirtualFree(mem, 0, MEM_RELEASE)) { DWORD err = GetLastError(); - std::cerr << "Failed to free transposition table. Error code: 0x" << - std::hex << err << std::dec << std::endl; + std::cerr << "Failed to free large page memory. Error code: 0x" + << std::hex << err + << std::dec << std::endl; exit(EXIT_FAILURE); } } #else -void aligned_ttmem_free(void *mem) { - free(mem); +void aligned_large_pages_free(void *mem) { + std_aligned_free(mem); } #endif @@ -475,11 +607,11 @@ void bindThisThread(size_t) {} #else -/// best_group() retrieves logical processor information using Windows specific -/// API and returns the best group id for the thread with index idx. Original +/// best_node() retrieves logical processor information using Windows specific +/// API and returns the best node id for the thread with index idx. Original /// code from Texel by Peter Österlund. -int best_group(size_t idx) { +static int best_node(size_t idx) { int threads = 0; int nodes = 0; @@ -488,12 +620,13 @@ int best_group(size_t idx) { DWORD byteOffset = 0; // Early exit if the needed API is not available at runtime - HMODULE k32 = GetModuleHandle("Kernel32.dll"); + HMODULE k32 = GetModuleHandle(TEXT("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 + // First call to GetLogicalProcessorInformationEx() to get returnLength. + // We expect the call to fail due to null buffer. if (fun1(RelationAll, nullptr, &returnLength)) return -1; @@ -501,7 +634,7 @@ int best_group(size_t idx) { SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX *buffer, *ptr; ptr = buffer = (SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX*)malloc(returnLength); - // Second call, now we expect to succeed + // Second call to GetLogicalProcessorInformationEx(), now we expect to succeed if (!fun1(RelationAll, buffer, &returnLength)) { free(buffer); @@ -551,24 +684,98 @@ int best_group(size_t idx) { void bindThisThread(size_t idx) { // Use only local variables to be thread-safe - int group = best_group(idx); + int node = best_node(idx); - if (group == -1) + if (node == -1) return; // Early exit if the needed API are not available at runtime - HMODULE k32 = GetModuleHandle("Kernel32.dll"); + HMODULE k32 = GetModuleHandle(TEXT("Kernel32.dll")); auto fun2 = (fun2_t)(void(*)())GetProcAddress(k32, "GetNumaNodeProcessorMaskEx"); auto fun3 = (fun3_t)(void(*)())GetProcAddress(k32, "SetThreadGroupAffinity"); + auto fun4 = (fun4_t)(void(*)())GetProcAddress(k32, "GetNumaNodeProcessorMask2"); + auto fun5 = (fun5_t)(void(*)())GetProcAddress(k32, "GetMaximumProcessorGroupCount"); if (!fun2 || !fun3) return; - GROUP_AFFINITY affinity; - if (fun2(group, &affinity)) - fun3(GetCurrentThread(), &affinity, nullptr); + if (!fun4 || !fun5) + { + GROUP_AFFINITY affinity; + if (fun2(node, &affinity)) // GetNumaNodeProcessorMaskEx + fun3(GetCurrentThread(), &affinity, nullptr); // SetThreadGroupAffinity + } + else + { + // If a numa node has more than one processor group, we assume they are + // sized equal and we spread threads evenly across the groups. + USHORT elements, returnedElements; + elements = fun5(); // GetMaximumProcessorGroupCount + GROUP_AFFINITY *affinity = (GROUP_AFFINITY*)malloc(elements * sizeof(GROUP_AFFINITY)); + if (fun4(node, affinity, elements, &returnedElements)) // GetNumaNodeProcessorMask2 + fun3(GetCurrentThread(), &affinity[idx % returnedElements], nullptr); // SetThreadGroupAffinity + free(affinity); + } } #endif } // namespace WinProcGroup + +#ifdef _WIN32 +#include +#define GETCWD _getcwd +#else +#include +#define GETCWD getcwd +#endif + +namespace CommandLine { + +string argv0; // path+name of the executable binary, as given by argv[0] +string binaryDirectory; // path of the executable directory +string workingDirectory; // path of the working directory + +void init([[maybe_unused]] int argc, char* argv[]) { + string pathSeparator; + + // extract the path+name of the executable binary + argv0 = argv[0]; + +#ifdef _WIN32 + pathSeparator = "\\"; + #ifdef _MSC_VER + // Under windows argv[0] may not have the extension. Also _get_pgmptr() had + // issues in some windows 10 versions, so check returned values carefully. + char* pgmptr = nullptr; + if (!_get_pgmptr(&pgmptr) && pgmptr != nullptr && *pgmptr) + argv0 = pgmptr; + #endif +#else + pathSeparator = "/"; +#endif + + // extract the working directory + workingDirectory = ""; + char buff[40000]; + char* cwd = GETCWD(buff, 40000); + if (cwd) + workingDirectory = cwd; + + // extract the binary directory path from argv0 + binaryDirectory = argv0; + size_t pos = binaryDirectory.find_last_of("\\/"); + if (pos == std::string::npos) + binaryDirectory = "." + pathSeparator; + else + binaryDirectory.resize(pos + 1); + + // pattern replacement: "./" at the start of path is replaced by the working directory + if (binaryDirectory.find("." + pathSeparator) == 0) + binaryDirectory.replace(0, 1, workingDirectory); +} + + +} // namespace CommandLine + +} // namespace Stockfish