Correctly describe POPCNT compile

[stockfish] / src / bitcount.h

diff --git a/src/bitcount.h b/src/bitcount.h
index ce2a69a897024361c7797caabd6644e1917a4c45..9feed19f9755a94be0022f3e9cf5c891972589d0 100644 (file)
--- 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) 2008-2012 Marco Costalba, Joona Kiiski, Tord Romstad
+  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
@@ -18,7 +18,7 @@
   along with this program.  If not, see <http://www.gnu.org/licenses/>.
 */
 
-#if !defined(BITCOUNT_H_INCLUDED)
+#ifndef BITCOUNT_H_INCLUDED
 #define BITCOUNT_H_INCLUDED
 
 #include <cassert>
@@ -32,62 +32,56 @@ enum BitCountType {
   CNT_HW_POPCNT
 };
 
-/// Determine at compile time the best popcount<> specialization according if
-/// platform is 32 or 64 bits, to the maximum number of nonzero bits to count or
-/// use hardware popcnt instruction when available.
+/// 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 nonzero bits in a bitboard
+/// popcount() counts the number of non-zero bits in a bitboard
 template<BitCountType> inline int popcount(Bitboard);
 
 template<>
 inline int popcount<CNT_64>(Bitboard b) {
-  b -= ((b>>1) & 0x5555555555555555ULL);
-  b = ((b>>2) & 0x3333333333333333ULL) + (b & 0x3333333333333333ULL);
-  b = ((b>>4) + b) & 0x0F0F0F0F0F0F0F0FULL;
-  b *= 0x0101010101010101ULL;
-  return int(b >> 56);
+  b -=  (b >> 1) & 0x5555555555555555ULL;
+  b  = ((b >> 2) & 0x3333333333333333ULL) + (b & 0x3333333333333333ULL);
+  b  = ((b >> 4) + b) & 0x0F0F0F0F0F0F0F0FULL;
+  return (b * 0x0101010101010101ULL) >> 56;
 }
 
 template<>
 inline int popcount<CNT_64_MAX15>(Bitboard b) {
-  b -= (b>>1) & 0x5555555555555555ULL;
-  b = ((b>>2) & 0x3333333333333333ULL) + (b & 0x3333333333333333ULL);
-  b *= 0x1111111111111111ULL;
-  return int(b >> 60);
+  b -=  (b >> 1) & 0x5555555555555555ULL;
+  b  = ((b >> 2) & 0x3333333333333333ULL) + (b & 0x3333333333333333ULL);
+  return (b * 0x1111111111111111ULL) >> 60;
 }
 
 template<>
 inline int popcount<CNT_32>(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;
 }
 
 template<>
 inline int popcount<CNT_32_MAX15>(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;
 }
 
 template<>
 inline int popcount<CNT_HW_POPCNT>(Bitboard b) {
 
-#if !defined(USE_POPCNT)
+#ifndef USE_POPCNT
 
   assert(false);
   return b != 0; // Avoid 'b not used' warning
@@ -102,11 +96,9 @@ inline int popcount<CNT_HW_POPCNT>(Bitboard b) {
 
 #else
 
-  unsigned long ret;
-  __asm__("popcnt %1, %0" : "=r" (ret) : "r" (b));
-  return ret;
+  return __builtin_popcountll(b);
 
 #endif
 }
 
-#endif // !defined(BITCOUNT_H_INCLUDED)
+#endif // #ifndef BITCOUNT_H_INCLUDED