X-Git-Url: https://git.sesse.net/?p=stockfish;a=blobdiff_plain;f=src%2Fmisc.cpp;h=449e07cee9b28530b6c1f646be50043c7a905a33;hp=7a23b7a9c930791ce41c254ac3834fbff2941978;hb=8df1cd10df0869f05916d1eb19e33b66127fcb86;hpb=2ec626ddae9c81af6338f72296df8a1465b5e036

diff --git a/src/misc.cpp b/src/misc.cpp
index 7a23b7a9..449e07ce 100644
--- a/src/misc.cpp
+++ b/src/misc.cpp
@@ -2,7 +2,7 @@
   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) 2015-2016 Marco Costalba, Joona Kiiski, Gary Linscott, 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
@@ -23,6 +23,11 @@
 #undef  _WIN32_WINNT
 #define _WIN32_WINNT 0x0601 // Force to include needed API prototypes
 #endif
+
+#ifndef NOMINMAX
+#define NOMINMAX
+#endif
+
 #include <windows.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
@@ -63,10 +68,10 @@ struct Tie: public streambuf { // MSVC requires split streambuf for cin and cout
 
   Tie(streambuf* b, streambuf* l) : buf(b), logBuf(l) {}
 
-  int sync() { return logBuf->pubsync(), buf->pubsync(); }
-  int overflow(int c) { return log(buf->sputc((char)c), "<< "); }
-  int underflow() { return buf->sgetc(); }
-  int uflow() { return log(buf->sbumpc(), ">> "); }
+  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(), ">> "); }
 
   streambuf *buf, *logBuf;
 
@@ -205,6 +210,12 @@ void prefetch(void* addr) {
 
 #endif
 
+void prefetch2(void* addr) {
+
+  prefetch(addr);
+  prefetch((uint8_t*)addr + 64);
+}
+
 namespace WinProcGroup {
 
 #ifndef _WIN32
@@ -213,11 +224,11 @@ void bindThisThread(size_t) {}
 
 #else
 
-/// get_group() retrieves logical processor information using Windows specific
+/// 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 get_group(size_t idx) {
+int best_group(size_t idx) {
 
   int threads = 0;
   int nodes = 0;
@@ -227,7 +238,7 @@ int get_group(size_t idx) {
 
   // Early exit if the needed API is not available at runtime
   HMODULE k32 = GetModuleHandle("Kernel32.dll");
-  auto fun1 = (fun1_t)GetProcAddress(k32, "GetLogicalProcessorInformationEx");
+  auto fun1 = (fun1_t)(void(*)())GetProcAddress(k32, "GetLogicalProcessorInformationEx");
   if (!fun1)
       return -1;
 
@@ -246,7 +257,7 @@ int get_group(size_t idx) {
       return -1;
   }
 
-  while (ptr->Size > 0 && byteOffset + ptr->Size <= returnLength)
+  while (byteOffset < returnLength)
   {
       if (ptr->Relationship == RelationNumaNode)
           nodes++;
@@ -257,6 +268,7 @@ int get_group(size_t idx) {
           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);
   }
@@ -287,24 +299,16 @@ int get_group(size_t idx) {
 
 void bindThisThread(size_t idx) {
 
-  // If OS already scheduled us on a different group than 0 then don't overwrite
-  // the choice, eventually we are one of many one-threaded processes running on
-  // some Windows NUMA hardware, for instance in fishtest. To make it simple,
-  // just check if running threads are below a threshold, in this case all this
-  // NUMA machinery is not needed.
-  if (Threads.size() < 8)
-      return;
-
   // Use only local variables to be thread-safe
-  int group = get_group(idx);
+  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)GetProcAddress(k32, "GetNumaNodeProcessorMaskEx");
-  auto fun3 = (fun3_t)GetProcAddress(k32, "SetThreadGroupAffinity");
+  auto fun2 = (fun2_t)(void(*)())GetProcAddress(k32, "GetNumaNodeProcessorMaskEx");
+  auto fun3 = (fun3_t)(void(*)())GetProcAddress(k32, "SetThreadGroupAffinity");
 
   if (!fun2 || !fun3)
       return;