X-Git-Url: https://git.sesse.net/?p=stockfish;a=blobdiff_plain;f=src%2Fbitcount.h;h=9feed19f9755a94be0022f3e9cf5c891972589d0;hp=9a3b481169b667943dedbb28cf538f977e02ca5e;hb=c014444f09ace05e908909d9c5c60127e998b538;hpb=08f3aac97c9ee9d9a4cf0a3232ac0bfa4c320896
diff --git a/src/bitcount.h b/src/bitcount.h
index 9a3b4811..9feed19f 100644
--- a/src/bitcount.h
+++ b/src/bitcount.h
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
- Copyright (C) 2008-2009 Marco Costalba
+ Copyright (C) 2008-2014 Marco Costalba, Joona Kiiski, 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
@@ -18,147 +18,87 @@
along with this program. If not, see .
*/
-
-#if !defined(BITCOUNT_H_INCLUDED)
+#ifndef BITCOUNT_H_INCLUDED
#define BITCOUNT_H_INCLUDED
-// To disable POPCNT support uncomment NO_POPCNT define. You should do it only
-// in PGO compiling to exercise the default fallback path. Don't forget to
-// re-comment the line for the final optimized compile though ;-)
-
-//#define NO_POPCNT
-
-
+#include
#include "types.h"
-// Select type of intrinsic bit count instruction to use
-
-#if defined(_MSC_VER) && defined(IS_64BIT) && !defined(NO_POPCNT) // Microsoft compiler
-
-#include
-
-inline bool cpu_has_popcnt() {
-
- int CPUInfo[4] = {-1};
- __cpuid(CPUInfo, 0x00000001);
- return (CPUInfo[2] >> 23) & 1;
+enum BitCountType {
+ CNT_64,
+ CNT_64_MAX15,
+ CNT_32,
+ CNT_32_MAX15,
+ CNT_HW_POPCNT
+};
+
+/// Determine at compile time the best popcount<> specialization according to
+/// whether the platform is 32 or 64 bit, the maximum number of non-zero
+/// bits to count and if the hardware popcnt instruction is available.
+const BitCountType Full = HasPopCnt ? CNT_HW_POPCNT : Is64Bit ? CNT_64 : CNT_32;
+const BitCountType Max15 = HasPopCnt ? CNT_HW_POPCNT : Is64Bit ? CNT_64_MAX15 : CNT_32_MAX15;
+
+
+/// popcount() counts the number of non-zero bits in a bitboard
+template inline int popcount(Bitboard);
+
+template<>
+inline int popcount(Bitboard b) {
+ b -= (b >> 1) & 0x5555555555555555ULL;
+ b = ((b >> 2) & 0x3333333333333333ULL) + (b & 0x3333333333333333ULL);
+ b = ((b >> 4) + b) & 0x0F0F0F0F0F0F0F0FULL;
+ return (b * 0x0101010101010101ULL) >> 56;
}
-// Define a dummy template to workaround a compile error if __popcnt64() is not defined.
-//
-// If __popcnt64() is defined in it will be choosen first due to
-// C++ overload rules that always prefer a function to a template with the same name.
-// If not, we avoid a compile error and because cpu_has_popcnt() should return false,
-// our templetized __popcnt64() is never called anyway.
-template unsigned __popcnt64(T) { return 0; } // Is never called
-
-#define POPCNT_INTRINSIC(x) __popcnt64(x)
-
-#elif defined(__INTEL_COMPILER) && defined(IS_64BIT) && !defined(NO_POPCNT) // Intel compiler
-
-#include
-
-inline bool cpu_has_popcnt() {
-
- int CPUInfo[4] = {-1};
- __cpuid(CPUInfo, 0x00000001);
- return (CPUInfo[2] >> 23) & 1;
+template<>
+inline int popcount(Bitboard b) {
+ b -= (b >> 1) & 0x5555555555555555ULL;
+ b = ((b >> 2) & 0x3333333333333333ULL) + (b & 0x3333333333333333ULL);
+ return (b * 0x1111111111111111ULL) >> 60;
}
-// See comment of __popcnt64<>() few lines above for an explanation.
-template unsigned _mm_popcnt_u64(T) { return 0; } // Is never called
-
-#define POPCNT_INTRINSIC(x) _mm_popcnt_u64(x)
-
-#else // Safe fallback for unsupported compilers or when NO_POPCNT is defined
-
-inline bool cpu_has_popcnt() { return false; }
-
-#define POPCNT_INTRINSIC(x) 0
-
-#endif // cpu_has_popcnt() and POPCNT_INTRINSIC() definitions
-
-
-/// Software implementation of bit count functions
-
-#if defined(IS_64BIT)
-
-inline int count_1s(Bitboard b) {
- b -= ((b>>1) & 0x5555555555555555ULL);
- b = ((b>>2) & 0x3333333333333333ULL) + (b & 0x3333333333333333ULL);
- b = ((b>>4) + b) & 0x0F0F0F0F0F0F0F0FULL;
- b *= 0x0101010101010101ULL;
- return int(b >> 56);
-}
-
-inline int count_1s_max_15(Bitboard b) {
- b -= (b>>1) & 0x5555555555555555ULL;
- b = ((b>>2) & 0x3333333333333333ULL) + (b & 0x3333333333333333ULL);
- b *= 0x1111111111111111ULL;
- return int(b >> 60);
-}
-
-#else // if !defined(IS_64BIT)
-
-inline int count_1s(Bitboard b) {
+template<>
+inline int popcount(Bitboard b) {
unsigned w = unsigned(b >> 32), v = unsigned(b);
- v -= (v >> 1) & 0x55555555; // 0-2 in 2 bits
- w -= (w >> 1) & 0x55555555;
- v = ((v >> 2) & 0x33333333) + (v & 0x33333333); // 0-4 in 4 bits
- w = ((w >> 2) & 0x33333333) + (w & 0x33333333);
- v = ((v >> 4) + v) & 0x0F0F0F0F; // 0-8 in 8 bits
- v += (((w >> 4) + w) & 0x0F0F0F0F); // 0-16 in 8 bits
- v *= 0x01010101; // mul is fast on amd procs
- return int(v >> 24);
+ v -= (v >> 1) & 0x55555555; // 0-2 in 2 bits
+ w -= (w >> 1) & 0x55555555;
+ v = ((v >> 2) & 0x33333333) + (v & 0x33333333); // 0-4 in 4 bits
+ w = ((w >> 2) & 0x33333333) + (w & 0x33333333);
+ v = ((v >> 4) + v + (w >> 4) + w) & 0x0F0F0F0F;
+ return (v * 0x01010101) >> 24;
}
-inline int count_1s_max_15(Bitboard b) {
+template<>
+inline int popcount(Bitboard b) {
unsigned w = unsigned(b >> 32), v = unsigned(b);
- v -= (v >> 1) & 0x55555555; // 0-2 in 2 bits
- w -= (w >> 1) & 0x55555555;
- v = ((v >> 2) & 0x33333333) + (v & 0x33333333); // 0-4 in 4 bits
- w = ((w >> 2) & 0x33333333) + (w & 0x33333333);
- v += w; // 0-8 in 4 bits
- v *= 0x11111111;
- return int(v >> 28);
+ v -= (v >> 1) & 0x55555555; // 0-2 in 2 bits
+ w -= (w >> 1) & 0x55555555;
+ v = ((v >> 2) & 0x33333333) + (v & 0x33333333); // 0-4 in 4 bits
+ w = ((w >> 2) & 0x33333333) + (w & 0x33333333);
+ return ((v + w) * 0x11111111) >> 28;
}
-#endif // BITCOUNT
+template<>
+inline int popcount(Bitboard b) {
+#ifndef USE_POPCNT
-/// count_1s() counts the number of nonzero bits in a bitboard.
-/// If template parameter is true an intrinsic is called, otherwise
-/// we fallback on a software implementation.
+ assert(false);
+ return b != 0; // Avoid 'b not used' warning
-template
-inline int count_1s(Bitboard b) {
-
- return UseIntrinsic ? POPCNT_INTRINSIC(b) : count_1s(b);
-}
+#elif defined(_MSC_VER) && defined(__INTEL_COMPILER)
-template
-inline int count_1s_max_15(Bitboard b) {
+ return _mm_popcnt_u64(b);
- return UseIntrinsic ? POPCNT_INTRINSIC(b) : count_1s_max_15(b);
-}
+#elif defined(_MSC_VER)
+ return (int)__popcnt64(b);
-// Global constant initialized at startup that is set to true if
-// CPU on which application runs supports POPCNT intrinsic. Unless
-// NO_POPCNT is defined.
-#if defined(NO_POPCNT)
-const bool CpuHasPOPCNT = false;
#else
-const bool CpuHasPOPCNT = cpu_has_popcnt();
-#endif
+ return __builtin_popcountll(b);
-// Global constant used to print info about the use of 64 optimized
-// functions to verify that a 64 bit compile has been correctly built.
-#if defined(IS_64BIT)
-const bool CpuHas64BitPath = true;
-#else
-const bool CpuHas64BitPath = false;
#endif
+}
-#endif // !defined(BITCOUNT_H_INCLUDED)
+#endif // #ifndef BITCOUNT_H_INCLUDED