cds/algo/bit_reversal.h

   1 /*
   2     This file is a part of libcds - Concurrent Data Structures library
   3
   4     (C) Copyright Maxim Khizhinsky (libcds.dev@gmail.com) 2006-2017
   5
   6     Source code repo: http://github.com/khizmax/libcds/
   7     Download: http://sourceforge.net/projects/libcds/files/
   8
   9     Redistribution and use in source and binary forms, with or without
  10     modification, are permitted provided that the following conditions are met:
  11
  12     * Redistributions of source code must retain the above copyright notice, this
  13       list of conditions and the following disclaimer.
  14
  15     * Redistributions in binary form must reproduce the above copyright notice,
  16       this list of conditions and the following disclaimer in the documentation
  17       and/or other materials provided with the distribution.
  18
  19     THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
  20     AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  21     IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
  22     DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
  23     FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  24     DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
  25     SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
  26     CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
  27     OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
  28     OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  29 */
  30
  31 #ifndef CDSLIB_ALGO_BIT_REVERSAL_H
  32 #define CDSLIB_ALGO_BIT_REVERSAL_H
  33
  34 #include <cds/algo/base.h>
  35
  36     // Source: http://stackoverflow.com/questions/746171/best-algorithm-for-bit-reversal-from-msb-lsb-to-lsb-msb-in-c
  37 namespace cds { namespace algo {
  38
  39     /// Bit reversal algorithms
  40     namespace bit_reversal {
  41
  42         /// SWAR algorithm (source: http://aggregate.org/MAGIC/#Bit%20Reversal)
  43         struct swar {
  44             /// 32bit
  45             uint32_t operator()( uint32_t x ) const
  46             {
  47                 x = ( ( ( x & 0xaaaaaaaa ) >> 1 ) | ( ( x & 0x55555555 ) << 1 ));
  48                 x = ( ( ( x & 0xcccccccc ) >> 2 ) | ( ( x & 0x33333333 ) << 2 ));
  49                 x = ( ( ( x & 0xf0f0f0f0 ) >> 4 ) | ( ( x & 0x0f0f0f0f ) << 4 ));
  50                 x = ( ( ( x & 0xff00ff00 ) >> 8 ) | ( ( x & 0x00ff00ff ) << 8 ));
  51                 return( ( x >> 16 ) | ( x << 16 ));
  52             }
  53
  54             /// 64bit
  55             uint64_t operator()( uint64_t x ) const
  56             {
  57                 return ( static_cast<uint64_t>( operator()( static_cast<uint32_t>( x )) ) << 32 )  // low 32bit
  58                     | ( static_cast<uint64_t>( operator()( static_cast<uint32_t>( x >> 32 )) ));  // high 32bit
  59             }
  60         };
  61
  62         /// Lookup table algorithm
  63         struct lookup {
  64             /// 32bit
  65             uint32_t operator()( uint32_t x ) const
  66             {
  67                 static uint8_t const table[] = {
  68                     0x00, 0x80, 0x40, 0xC0, 0x20, 0xA0, 0x60, 0xE0, 0x10, 0x90, 0x50, 0xD0, 0x30, 0xB0, 0x70, 0xF0,
  69                     0x08, 0x88, 0x48, 0xC8, 0x28, 0xA8, 0x68, 0xE8, 0x18, 0x98, 0x58, 0xD8, 0x38, 0xB8, 0x78, 0xF8,
  70                     0x04, 0x84, 0x44, 0xC4, 0x24, 0xA4, 0x64, 0xE4, 0x14, 0x94, 0x54, 0xD4, 0x34, 0xB4, 0x74, 0xF4,
  71                     0x0C, 0x8C, 0x4C, 0xCC, 0x2C, 0xAC, 0x6C, 0xEC, 0x1C, 0x9C, 0x5C, 0xDC, 0x3C, 0xBC, 0x7C, 0xFC,
  72                     0x02, 0x82, 0x42, 0xC2, 0x22, 0xA2, 0x62, 0xE2, 0x12, 0x92, 0x52, 0xD2, 0x32, 0xB2, 0x72, 0xF2,
  73                     0x0A, 0x8A, 0x4A, 0xCA, 0x2A, 0xAA, 0x6A, 0xEA, 0x1A, 0x9A, 0x5A, 0xDA, 0x3A, 0xBA, 0x7A, 0xFA,
  74                     0x06, 0x86, 0x46, 0xC6, 0x26, 0xA6, 0x66, 0xE6, 0x16, 0x96, 0x56, 0xD6, 0x36, 0xB6, 0x76, 0xF6,
  75                     0x0E, 0x8E, 0x4E, 0xCE, 0x2E, 0xAE, 0x6E, 0xEE, 0x1E, 0x9E, 0x5E, 0xDE, 0x3E, 0xBE, 0x7E, 0xFE,
  76                     0x01, 0x81, 0x41, 0xC1, 0x21, 0xA1, 0x61, 0xE1, 0x11, 0x91, 0x51, 0xD1, 0x31, 0xB1, 0x71, 0xF1,
  77                     0x09, 0x89, 0x49, 0xC9, 0x29, 0xA9, 0x69, 0xE9, 0x19, 0x99, 0x59, 0xD9, 0x39, 0xB9, 0x79, 0xF9,
  78                     0x05, 0x85, 0x45, 0xC5, 0x25, 0xA5, 0x65, 0xE5, 0x15, 0x95, 0x55, 0xD5, 0x35, 0xB5, 0x75, 0xF5,
  79                     0x0D, 0x8D, 0x4D, 0xCD, 0x2D, 0xAD, 0x6D, 0xED, 0x1D, 0x9D, 0x5D, 0xDD, 0x3D, 0xBD, 0x7D, 0xFD,
  80                     0x03, 0x83, 0x43, 0xC3, 0x23, 0xA3, 0x63, 0xE3, 0x13, 0x93, 0x53, 0xD3, 0x33, 0xB3, 0x73, 0xF3,
  81                     0x0B, 0x8B, 0x4B, 0xCB, 0x2B, 0xAB, 0x6B, 0xEB, 0x1B, 0x9B, 0x5B, 0xDB, 0x3B, 0xBB, 0x7B, 0xFB,
  82                     0x07, 0x87, 0x47, 0xC7, 0x27, 0xA7, 0x67, 0xE7, 0x17, 0x97, 0x57, 0xD7, 0x37, 0xB7, 0x77, 0xF7,
  83                     0x0F, 0x8F, 0x4F, 0xCF, 0x2F, 0xAF, 0x6F, 0xEF, 0x1F, 0x9F, 0x5F, 0xDF, 0x3F, 0xBF, 0x7F, 0xFF
  84                 };
  85                 static_assert( sizeof( table ) / sizeof( table[0] ) == 256, "Table size mismatch" );
  86
  87                 return ( static_cast<uint32_t>( table[x & 0xff] ) << 24 ) |
  88                     ( static_cast<uint32_t>( table[( x >> 8 ) & 0xff] ) << 16 ) |
  89                     ( static_cast<uint32_t>( table[( x >> 16 ) & 0xff] ) << 8 ) |
  90                     ( static_cast<uint32_t>( table[( x >> 24 ) & 0xff] ));
  91             }
  92
  93             /// 64bit
  94             uint64_t operator()( uint64_t x ) const
  95             {
  96                 return ( static_cast<uint64_t>( operator()( static_cast<uint32_t>( x )) ) << 32 ) |
  97                     static_cast<uint64_t>( operator()( static_cast<uint32_t>( x >> 32 )) );
  98             }
  99         };
 100
 101
 102         /// Mul-Div algorithm for 32bit architectire
 103
 104         /// Mul-Div algorithm
 105         struct muldiv {
 106             //@cond
 107             static uint8_t muldiv32_byte( uint8_t b )
 108             {
 109                 return static_cast<uint8_t>( ( ( b * 0x0802LU & 0x22110LU ) | ( b * 0x8020LU & 0x88440LU )) * 0x10101LU >> 16 );
 110             }
 111
 112             static uint8_t muldiv64_byte( uint8_t b )
 113             {
 114                 return static_cast<uint8_t>( ( b * 0x0202020202ULL & 0x010884422010ULL ) % 1023 );
 115             }
 116
 117             // for 32bit architecture
 118             static uint32_t muldiv32( uint32_t x )
 119             {
 120                 return static_cast<uint32_t>( muldiv32_byte( static_cast<uint8_t>( x >> 24 )) )
 121                     | ( static_cast<uint32_t>( muldiv32_byte( static_cast<uint8_t>( x >> 16 )) ) << 8 )
 122                     | ( static_cast<uint32_t>( muldiv32_byte( static_cast<uint8_t>( x >> 8 )) ) << 16 )
 123                     | ( static_cast<uint32_t>( muldiv32_byte( static_cast<uint8_t>( x )) ) << 24 );
 124             }
 125
 126             static uint64_t muldiv32( uint64_t x )
 127             {
 128                 return static_cast<uint64_t>( muldiv32_byte( static_cast<uint8_t>( x >> 56 )) )
 129                     | ( static_cast<uint64_t>( muldiv32_byte( static_cast<uint8_t>( x >> 48 )) ) << 8 )
 130                     | ( static_cast<uint64_t>( muldiv32_byte( static_cast<uint8_t>( x >> 40 )) ) << 16 )
 131                     | ( static_cast<uint64_t>( muldiv32_byte( static_cast<uint8_t>( x >> 32 )) ) << 24 )
 132                     | ( static_cast<uint64_t>( muldiv32_byte( static_cast<uint8_t>( x >> 24 )) ) << 32 )
 133                     | ( static_cast<uint64_t>( muldiv32_byte( static_cast<uint8_t>( x >> 16 )) ) << 40 )
 134                     | ( static_cast<uint64_t>( muldiv32_byte( static_cast<uint8_t>( x >> 8 )) ) << 48 )
 135                     | ( static_cast<uint64_t>( muldiv32_byte( static_cast<uint8_t>( x )) ) << 56 );
 136             }
 137
 138             /// for 64bit architectire
 139             static uint32_t muldiv64( uint32_t x )
 140             {
 141                 return static_cast<uint32_t>( muldiv64_byte( static_cast<uint8_t>( x >> 24 )) )
 142                     | ( static_cast<uint32_t>( muldiv64_byte( static_cast<uint8_t>( x >> 16 )) ) << 8 )
 143                     | ( static_cast<uint32_t>( muldiv64_byte( static_cast<uint8_t>( x >> 8 )) ) << 16 )
 144                     | ( static_cast<uint32_t>( muldiv64_byte( static_cast<uint8_t>( x )) ) << 24 );
 145             }
 146
 147             static uint64_t muldiv64( uint64_t x )
 148             {
 149                 return static_cast<uint64_t>( muldiv64_byte( static_cast<uint8_t>( x >> 56 )) )
 150                     | ( static_cast<uint64_t>( muldiv64_byte( static_cast<uint8_t>( x >> 48 )) ) << 8 )
 151                     | ( static_cast<uint64_t>( muldiv64_byte( static_cast<uint8_t>( x >> 40 )) ) << 16 )
 152                     | ( static_cast<uint64_t>( muldiv64_byte( static_cast<uint8_t>( x >> 32 )) ) << 24 )
 153                     | ( static_cast<uint64_t>( muldiv64_byte( static_cast<uint8_t>( x >> 24 )) ) << 32 )
 154                     | ( static_cast<uint64_t>( muldiv64_byte( static_cast<uint8_t>( x >> 16 )) ) << 40 )
 155                     | ( static_cast<uint64_t>( muldiv64_byte( static_cast<uint8_t>( x >> 8 )) ) << 48 )
 156                     | ( static_cast<uint64_t>( muldiv64_byte( static_cast<uint8_t>( x )) ) << 56 );
 157             }
 158             //@endcond
 159
 160             /// 32bit
 161             uint32_t operator()( uint32_t x ) const
 162             {
 163 #           if CDS_BUILD_BITS == 32
 164                 return muldiv32( x );
 165 #           else
 166                 return muldiv64( x );
 167 #           endif
 168             }
 169
 170             /// 64bit
 171             uint64_t operator()( uint64_t x ) const
 172             {
 173 #           if CDS_BUILD_BITS == 32
 174                 return muldiv32( x );
 175 #           else
 176                 return muldiv64( x );
 177 #           endif
 178             }
 179         };
 180
 181     } // namespace bit_reversal
 182 }} // namespace cds::algo
 183
 184 #endif // #ifndef CDSLIB_ALGO_BIT_REVERSAL_H