X-Git-Url: https://git.sesse.net/?p=stockfish;a=blobdiff_plain;f=src%2Fmisc.cpp;h=4d6483e73d20f78d352429bbc24289a648fc3649;hp=484d0b210cb6ddeeed9ac315f6bc930b063b31f1;hb=b8c00efa2767ebf74545d2ba4bd344ef7c963319;hpb=7ed817d7e4847196a5ba0a3aded0519000074be7

diff --git a/src/misc.cpp b/src/misc.cpp
index 484d0b21..4d6483e7 100644
--- a/src/misc.cpp
+++ b/src/misc.cpp
@@ -47,6 +47,11 @@ typedef bool(*fun3_t)(HANDLE, CONST GROUP_AFFINITY*, PGROUP_AFFINITY);
 #include <sstream>
 #include <vector>
 
+#if defined(__linux__) && !defined(__ANDROID__)
+#include <stdlib.h>
+#include <sys/mman.h>
+#endif
+
 #include "misc.h"
 #include "thread.h"
 
@@ -190,7 +195,7 @@ const std::string compiler_info() {
      compiler += "(unknown version)";
   #endif
 
-  #if defined(__APPLE__) 
+  #if defined(__APPLE__)
      compiler += " on Apple";
   #elif defined(__CYGWIN__)
      compiler += " on Cygwin";
@@ -288,6 +293,35 @@ void prefetch(void* addr) {
 
 #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 can be simplified.
+#if defined(__linux__) && !defined(__ANDROID__)
+
+void* aligned_ttmem_alloc(size_t allocSize, void*& mem) {
+
+  constexpr size_t alignment = 2 * 1024 * 1024; // assumed 2MB page sizes
+  size_t size = ((allocSize + alignment - 1) / alignment) * alignment; // multiple of alignment
+  mem = aligned_alloc(alignment, size);
+  madvise(mem, allocSize, MADV_HUGEPAGE);
+  return mem;
+}
+
+#else
+
+void* aligned_ttmem_alloc(size_t allocSize, void*& mem) {
+
+  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<void*>((uintptr_t(mem) + alignment - 1) & ~uintptr_t(alignment - 1));
+  return ret;
+}
+
+#endif
+
+
 namespace WinProcGroup {
 
 #ifndef _WIN32