From a87ea9846d4adc8d2ed8107c072c43b5ef53e4fe Mon Sep 17 00:00:00 2001
From: Marco Costalba <mcostalba@gmail.com>
Date: Fri, 3 Jul 2009 08:28:13 +0100
Subject: [PATCH] Use bsfq asm instruction to count bits

On 64 bit systems we can use bsfq instruction to count
set bits in a bitboard.

This is a patch for GCC and Intel compilers to take advantage
of that and get a 2% speed up.

Original patch from Heinz van Saanen, adapted to current tree
by me.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
---
 src/bitboard.cpp | 10 +++++-----
 src/bitboard.h   | 26 +++++++++++++++++++++-----
 2 files changed, 26 insertions(+), 10 deletions(-)

diff --git a/src/bitboard.cpp b/src/bitboard.cpp
index ba03d66e..0bdb9d3a 100644
--- a/src/bitboard.cpp
+++ b/src/bitboard.cpp
@@ -161,7 +161,7 @@ const int RShift[64] = {
   21, 22, 22, 22, 22, 22, 22, 21, 20, 21, 21, 21, 21, 21, 21, 20
 };
 
-#endif
+#endif // defined(IS_64BIT)
 
 
 Bitboard RMask[64];
@@ -245,16 +245,16 @@ void init_bitboards() {
 /// pop_1st_bit() finds and clears the least significant nonzero bit in a
 /// nonzero bitboard.
 
-#if defined(IS_64BIT)
+#if defined(IS_64BIT) && !defined(USE_BSFQ)
 
-Square pop_1st_bit(Bitboard *b) {
+Square pop_1st_bit(Bitboard* b) {
   Bitboard bb = *b ^ (*b - 1);
   uint32_t fold = int(bb) ^ int(bb >> 32);
   *b &= (*b - 1);
   return Square(BitTable[(fold * 0x783a9b23) >> 26]);
 }
 
-#else
+#elif !defined(USE_BSFQ)
 
 // Use type-punning
 union b_union {
@@ -267,7 +267,7 @@ union b_union {
 };
 
 // WARNING: Needs -fno-strict-aliasing compiler option
-Square pop_1st_bit(Bitboard *bb) {
+Square pop_1st_bit(Bitboard* bb) {
 
   b_union u;
   uint32_t b;
diff --git a/src/bitboard.h b/src/bitboard.h
index 3429c42e..d5c16a91 100644
--- a/src/bitboard.h
+++ b/src/bitboard.h
@@ -36,6 +36,11 @@
 #define IS_64BIT
 #endif
 
+#if defined(IS_64BIT) && (defined(__GNUC__) || defined(__INTEL_COMPILER))
+#define USE_BSFQ
+#endif
+
+
 ////
 //// Includes
 ////
@@ -383,14 +388,24 @@ inline Bitboard isolated_pawn_mask(Square s) {
 
 
 /// first_1() finds the least significant nonzero bit in a nonzero bitboard.
+/// pop_1st_bit() finds and clears the least significant nonzero bit in a
+/// nonzero bitboard.
 
-#if defined(IS_64BIT)
+#if defined(USE_BSFQ) // Assembly code by Heinz van Saanen
 
-inline Square first_1(Bitboard b) {
-  return Square(BitTable[((b & -b) * 0x218a392cd3d5dbfULL) >> 58]);
+inline Square __attribute__((always_inline)) first_1(Bitboard b) {
+  Bitboard dummy;
+  __asm__("bsfq %1, %0": "=r"(dummy): "rm"(b) );
+  return (Square)(dummy);
+}
+
+inline Square __attribute__((always_inline)) pop_1st_bit(Bitboard* b) {
+  const Square s = first_1(*b);
+  *b &= ~(1ULL<<s);
+  return s;
 }
 
-#else
+#else // if !defined(USE_BSFQ)
 
 inline Square first_1(Bitboard b) {
   b ^= (b - 1);
@@ -398,6 +413,8 @@ inline Square first_1(Bitboard b) {
   return Square(BitTable[(fold * 0x783a9b23) >> 26]);
 }
 
+extern Square pop_1st_bit(Bitboard* b);
+
 #endif
 
 
@@ -407,7 +424,6 @@ inline Square first_1(Bitboard b) {
 
 extern void print_bitboard(Bitboard b);
 extern void init_bitboards();
-extern Square pop_1st_bit(Bitboard *b);
 
 
 #endif // !defined(BITBOARD_H_INCLUDED)
-- 
2.39.2