softfloat/source/f128M_rem.c

   1
   2 /*============================================================================
   3
   4 This C source file is part of the SoftFloat IEEE Floating-Point Arithmetic
   5 Package, Release 3e, by John R. Hauser.
   6
   7 Copyright 2011, 2012, 2013, 2014 The Regents of the University of California.
   8 All rights reserved.
   9
  10 Redistribution and use in source and binary forms, with or without
  11 modification, are permitted provided that the following conditions are met:
  12
  13  1. Redistributions of source code must retain the above copyright notice,
  14     this list of conditions, and the following disclaimer.
  15
  16  2. Redistributions in binary form must reproduce the above copyright notice,
  17     this list of conditions, and the following disclaimer in the documentation
  18     and/or other materials provided with the distribution.
  19
  20  3. Neither the name of the University nor the names of its contributors may
  21     be used to endorse or promote products derived from this software without
  22     specific prior written permission.
  23
  24 THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS "AS IS", AND ANY
  25 EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
  26 WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, ARE
  27 DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE FOR ANY
  28 DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
  29 (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
  30 LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
  31 ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  32 (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
  33 SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  34
  35 =============================================================================*/
  36
  37 #include <stdbool.h>
  38 #include <stdint.h>
  39 #include "platform.h"
  40 #include "internals.h"
  41 #include "specialize.h"
  42 #include "softfloat.h"
  43
  44 #ifdef SOFTFLOAT_FAST_INT64
  45
  46 void
  47  f128M_rem( const float128_t *aPtr, const float128_t *bPtr, float128_t *zPtr )
  48 {
  49
  50     *zPtr = f128_rem( *aPtr, *bPtr );
  51
  52 }
  53
  54 #else
  55
  56 void
  57  f128M_rem( const float128_t *aPtr, const float128_t *bPtr, float128_t *zPtr )
  58 {
  59     const uint32_t *aWPtr, *bWPtr;
  60     uint32_t *zWPtr, uiA96;
  61     int32_t expA, expB;
  62     uint32_t x[4], rem1[5], *remPtr;
  63     bool signRem;
  64     int32_t expDiff;
  65     uint32_t q, recip32;
  66     uint64_t q64;
  67     uint32_t rem2[5], *altRemPtr, *newRemPtr, wordMeanRem;
  68
  69     /*------------------------------------------------------------------------
  70     *------------------------------------------------------------------------*/
  71     aWPtr = (const uint32_t *) aPtr;
  72     bWPtr = (const uint32_t *) bPtr;
  73     zWPtr = (uint32_t *) zPtr;
  74     /*------------------------------------------------------------------------
  75     *------------------------------------------------------------------------*/
  76     uiA96 = aWPtr[indexWordHi( 4 )];
  77     expA = expF128UI96( uiA96 );
  78     expB = expF128UI96( bWPtr[indexWordHi( 4 )] );
  79     /*------------------------------------------------------------------------
  80     *------------------------------------------------------------------------*/
  81     if ( (expA == 0x7FFF) || (expB == 0x7FFF) ) {
  82         if ( softfloat_tryPropagateNaNF128M( aWPtr, bWPtr, zWPtr ) ) return;
  83         if ( expA == 0x7FFF ) goto invalid;
  84         goto copyA;
  85     }
  86     /*------------------------------------------------------------------------
  87     *------------------------------------------------------------------------*/
  88     if ( expA < expB - 1 ) goto copyA;
  89     /*------------------------------------------------------------------------
  90     *------------------------------------------------------------------------*/
  91     expB = softfloat_shiftNormSigF128M( bWPtr, 13, x );
  92     if ( expB == -128 ) goto invalid;
  93     remPtr = &rem1[indexMultiwordLo( 5, 4 )];
  94     expA = softfloat_shiftNormSigF128M( aWPtr, 13, remPtr );
  95     if ( expA == -128 ) goto copyA;
  96     signRem = signF128UI96( uiA96 );
  97     /*------------------------------------------------------------------------
  98     *------------------------------------------------------------------------*/
  99     expDiff = expA - expB;
 100     if ( expDiff < 1 ) {
 101         if ( expDiff < -1 ) goto copyA;
 102         if ( expDiff ) {
 103             --expB;
 104             softfloat_add128M( x, x, x );
 105             q = 0;
 106         } else {
 107             q = (softfloat_compare128M( x, remPtr ) <= 0);
 108             if ( q ) softfloat_sub128M( remPtr, x, remPtr );
 109         }
 110     } else {
 111         recip32 =
 112             softfloat_approxRecip32_1(
 113                 ((uint64_t) x[indexWord( 4, 3 )]<<32 | x[indexWord( 4, 2 )])
 114                     >>30
 115             );
 116         expDiff -= 30;
 117         for (;;) {
 118             q64 = (uint64_t) remPtr[indexWordHi( 4 )] * recip32;
 119             if ( expDiff < 0 ) break;
 120             q = (q64 + 0x80000000)>>32;
 121             softfloat_remStep128MBy32( remPtr, 29, x, q, remPtr );
 122             if ( remPtr[indexWordHi( 4 )] & 0x80000000 ) {
 123                 softfloat_add128M( remPtr, x, remPtr );
 124             }
 125             expDiff -= 29;
 126         }
 127         /*--------------------------------------------------------------------
 128         | (`expDiff' cannot be less than -29 here.)
 129         *--------------------------------------------------------------------*/
 130         q = (uint32_t) (q64>>32)>>(~expDiff & 31);
 131         softfloat_remStep128MBy32( remPtr, expDiff + 30, x, q, remPtr );
 132         if ( remPtr[indexWordHi( 4 )] & 0x80000000 ) {
 133             altRemPtr = &rem2[indexMultiwordLo( 5, 4 )];
 134             softfloat_add128M( remPtr, x, altRemPtr );
 135             goto selectRem;
 136         }
 137     }
 138     /*------------------------------------------------------------------------
 139     *------------------------------------------------------------------------*/
 140     altRemPtr = &rem2[indexMultiwordLo( 5, 4 )];
 141     do {
 142         ++q;
 143         newRemPtr = altRemPtr;
 144         softfloat_sub128M( remPtr, x, newRemPtr );
 145         altRemPtr = remPtr;
 146         remPtr = newRemPtr;
 147     } while ( ! (remPtr[indexWordHi( 4 )] & 0x80000000) );
 148  selectRem:
 149     softfloat_add128M( remPtr, altRemPtr, x );
 150     wordMeanRem = x[indexWordHi( 4 )];
 151     if (
 152         (wordMeanRem & 0x80000000)
 153             || (! wordMeanRem && (q & 1) && ! x[indexWord( 4, 0 )]
 154                     && ! (x[indexWord( 4, 2 )] | x[indexWord( 4, 1 )]))
 155     ) {
 156         remPtr = altRemPtr;
 157     }
 158     if ( remPtr[indexWordHi( 4 )] & 0x80000000 ) {
 159         signRem = ! signRem;
 160         softfloat_negX128M( remPtr );
 161     }
 162     remPtr -= indexMultiwordLo( 5, 4 );
 163     remPtr[indexWordHi( 5 )] = 0;
 164     softfloat_normRoundPackMToF128M( signRem, expB + 18, remPtr, zWPtr );
 165     return;
 166     /*------------------------------------------------------------------------
 167     *------------------------------------------------------------------------*/
 168  invalid:
 169     softfloat_invalidF128M( zWPtr );
 170     return;
 171     /*------------------------------------------------------------------------
 172     *------------------------------------------------------------------------*/
 173  copyA:
 174     zWPtr[indexWordHi( 4 )] = uiA96;
 175     zWPtr[indexWord( 4, 2 )] = aWPtr[indexWord( 4, 2 )];
 176     zWPtr[indexWord( 4, 1 )] = aWPtr[indexWord( 4, 1 )];
 177     zWPtr[indexWord( 4, 0 )] = aWPtr[indexWord( 4, 0 )];
 178
 179 }
 180
 181 #endif
 182