Optimize the most common update accumalator cases w/o tiling

[stockfish] / src / misc.h

diff --git a/src/misc.h b/src/misc.h
index 59ca6e376bda6b12a736d61f3ee92b9c621d1b6e..52595fb906bdd448472332de91e6e63f515c0482 100644 (file)
--- a/src/misc.h
+++ b/src/misc.h
@@ -1,6 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2021 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
@@ -21,12 +21,13 @@
 
 #include <cassert>
 #include <chrono>
-#include <ostream>
-#include <string>
-#include <vector>
+#include <cstddef>
 #include <cstdint>
+#include <iosfwd>
+#include <string>
 
-#include "types.h"
+#define stringify2(x) #x
+#define stringify(x) stringify2(x)
 
 namespace Stockfish {
 
@@ -39,26 +40,19 @@ void std_aligned_free(void* ptr);
 void* aligned_large_pages_alloc(size_t size); // memory aligned by page size, min alignment: 4096 bytes
 void aligned_large_pages_free(void* mem); // nop if mem == nullptr
 
-void dbg_hit_on(bool b);
-void dbg_hit_on(bool c, bool b);
-void dbg_mean_of(int v);
+void dbg_hit_on(bool cond, int slot = 0);
+void dbg_mean_of(int64_t value, int slot = 0);
+void dbg_stdev_of(int64_t value, int slot = 0);
+void dbg_correl_of(int64_t value1, int64_t value2, int slot = 0);
 void dbg_print();
 
-typedef std::chrono::milliseconds::rep TimePoint; // A value in milliseconds
+using TimePoint = std::chrono::milliseconds::rep; // A value in milliseconds
 static_assert(sizeof(TimePoint) == sizeof(int64_t), "TimePoint should be 64 bits");
 inline TimePoint now() {
   return std::chrono::duration_cast<std::chrono::milliseconds>
         (std::chrono::steady_clock::now().time_since_epoch()).count();
 }
 
-template<class Entry, int Size>
-struct HashTable {
-  Entry* operator[](Key key) { return &table[(uint32_t)key & (Size - 1)]; }
-
-private:
-  std::vector<Entry> table = std::vector<Entry>(Size); // Allocate on the heap
-};
-
 
 enum SyncCout { IO_LOCK, IO_UNLOCK };
 std::ostream& operator<<(std::ostream&, SyncCout);
@@ -66,9 +60,10 @@ std::ostream& operator<<(std::ostream&, SyncCout);
 #define sync_cout std::cout << IO_LOCK
 #define sync_endl std::endl << IO_UNLOCK
 
-// `ptr` must point to an array of size at least
-// `sizeof(T) * N + alignment` bytes, where `N` is the
-// number of elements in the array.
+
+// align_ptr_up() : get the first aligned element of an array.
+// ptr must point to an array of size at least `sizeof(T) * N + alignment` bytes,
+// where N is the number of elements in the array.
 template <uintptr_t Alignment, typename T>
 T* align_ptr_up(T* ptr)
 {
@@ -78,49 +73,28 @@ T* align_ptr_up(T* ptr)
   return reinterpret_cast<T*>(reinterpret_cast<char*>((ptrint + (Alignment - 1)) / Alignment * Alignment));
 }
 
-template <typename T>
-class ValueListInserter {
-public:
-  ValueListInserter(T* v, std::size_t& s) :
-    values(v),
-    size(&s)
-  {
-  }
 
-  void push_back(const T& value) { values[(*size)++] = value; }
-private:
-  T* values;
-  std::size_t* size;
-};
+// IsLittleEndian : true if and only if the binary is compiled on a little endian machine
+static inline const union { uint32_t i; char c[4]; } Le = { 0x01020304 };
+static inline const bool IsLittleEndian = (Le.c[0] == 4);
+
 
 template <typename T, std::size_t MaxSize>
 class ValueList {
 
 public:
   std::size_t size() const { return size_; }
-  void resize(std::size_t newSize) { size_ = newSize; }
   void push_back(const T& value) { values_[size_++] = value; }
-  T& operator[](std::size_t index) { return values_[index]; }
-  T* begin() { return values_; }
-  T* end() { return values_ + size_; }
-  const T& operator[](std::size_t index) const { return values_[index]; }
   const T* begin() const { return values_; }
   const T* end() const { return values_ + size_; }
-  operator ValueListInserter<T>() { return ValueListInserter(values_, size_); }
-
-  void swap(ValueList& other) {
-    const std::size_t maxSize = std::max(size_, other.size_);
-    for (std::size_t i = 0; i < maxSize; ++i) {
-      std::swap(values_[i], other.values_[i]);
-    }
-    std::swap(size_, other.size_);
-  }
+  const T& operator[](int index) const { return values_[index]; }
 
 private:
   T values_[MaxSize];
   std::size_t size_ = 0;
 };
 
+
 /// xorshift64star Pseudo-Random Number Generator
 /// This class is based on original code written and dedicated
 /// to the public domain by Sebastiano Vigna (2014).
@@ -159,14 +133,14 @@ public:
 
 inline uint64_t mul_hi64(uint64_t a, uint64_t b) {
 #if defined(__GNUC__) && defined(IS_64BIT)
-    __extension__ typedef unsigned __int128 uint128;
-    return ((uint128)a * (uint128)b) >> 64;
+    __extension__ using uint128 = unsigned __int128;
+    return (uint128(a) * uint128(b)) >> 64;
 #else
-    uint64_t aL = (uint32_t)a, aH = a >> 32;
-    uint64_t bL = (uint32_t)b, bH = b >> 32;
+    uint64_t aL = uint32_t(a), aH = a >> 32;
+    uint64_t bL = uint32_t(b), bH = b >> 32;
     uint64_t c1 = (aL * bL) >> 32;
     uint64_t c2 = aH * bL + c1;
-    uint64_t c3 = aL * bH + (uint32_t)c2;
+    uint64_t c3 = aL * bH + uint32_t(c2);
     return aH * bH + (c2 >> 32) + (c3 >> 32);
 #endif
 }