mirror of
https://github.com/urbit/shrub.git
synced 2024-12-24 03:14:30 +03:00
187 lines
7.0 KiB
C
187 lines
7.0 KiB
C
|
|
||
|
/*============================================================================
|
||
|
|
||
|
This C source file is part of the SoftFloat IEEE Floating-Point Arithmetic
|
||
|
Package, Release 3, by John R. Hauser.
|
||
|
|
||
|
Copyright 2011, 2012, 2013, 2014 The Regents of the University of California
|
||
|
(Regents). All Rights Reserved. Redistribution and use in source and binary
|
||
|
forms, with or without modification, are permitted provided that the following
|
||
|
conditions are met:
|
||
|
|
||
|
Redistributions of source code must retain the above copyright notice,
|
||
|
this list of conditions, and the following two paragraphs of disclaimer.
|
||
|
Redistributions in binary form must reproduce the above copyright notice,
|
||
|
this list of conditions, and the following two paragraphs of disclaimer in the
|
||
|
documentation and/or other materials provided with the distribution. Neither
|
||
|
the name of the Regents nor the names of its contributors may be used to
|
||
|
endorse or promote products derived from this software without specific prior
|
||
|
written permission.
|
||
|
|
||
|
IN NO EVENT SHALL REGENTS BE LIABLE TO ANY PARTY FOR DIRECT, INDIRECT,
|
||
|
SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING LOST PROFITS, ARISING
|
||
|
OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN IF REGENTS HAS
|
||
|
BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||
|
|
||
|
REGENTS SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING, BUT NOT LIMITED
|
||
|
TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||
|
PURPOSE. THE SOFTWARE AND ACCOMPANYING DOCUMENTATION, IF ANY, PROVIDED
|
||
|
HEREUNDER IS PROVIDED "AS IS". REGENTS HAS NO OBLIGATION TO PROVIDE
|
||
|
MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
|
||
|
|
||
|
=============================================================================*/
|
||
|
|
||
|
#include <stdbool.h>
|
||
|
#include <stdint.h>
|
||
|
#include "platform.h"
|
||
|
#include "internals.h"
|
||
|
#include "specialize.h"
|
||
|
#include "softfloat.h"
|
||
|
|
||
|
float64_t f64_rem( float64_t a, float64_t b )
|
||
|
{
|
||
|
union ui64_f64 uA;
|
||
|
uint_fast64_t uiA;
|
||
|
bool signA;
|
||
|
int_fast16_t expA;
|
||
|
uint_fast64_t sigA;
|
||
|
union ui64_f64 uB;
|
||
|
uint_fast64_t uiB;
|
||
|
int_fast16_t expB;
|
||
|
uint_fast64_t sigB;
|
||
|
struct exp16_sig64 normExpSig;
|
||
|
uint64_t rem;
|
||
|
int_fast16_t expDiff;
|
||
|
uint32_t q, recip32;
|
||
|
uint_fast64_t q64;
|
||
|
uint64_t altRem, meanRem;
|
||
|
bool signRem;
|
||
|
uint_fast64_t uiZ;
|
||
|
union ui64_f64 uZ;
|
||
|
|
||
|
/*------------------------------------------------------------------------
|
||
|
*------------------------------------------------------------------------*/
|
||
|
uA.f = a;
|
||
|
uiA = uA.ui;
|
||
|
signA = signF64UI( uiA );
|
||
|
expA = expF64UI( uiA );
|
||
|
sigA = fracF64UI( uiA );
|
||
|
uB.f = b;
|
||
|
uiB = uB.ui;
|
||
|
expB = expF64UI( uiB );
|
||
|
sigB = fracF64UI( uiB );
|
||
|
/*------------------------------------------------------------------------
|
||
|
*------------------------------------------------------------------------*/
|
||
|
if ( expA == 0x7FF ) {
|
||
|
if ( sigA || ((expB == 0x7FF) && sigB) ) goto propagateNaN;
|
||
|
goto invalid;
|
||
|
}
|
||
|
if ( expB == 0x7FF ) {
|
||
|
if ( sigB ) goto propagateNaN;
|
||
|
return a;
|
||
|
}
|
||
|
/*------------------------------------------------------------------------
|
||
|
*------------------------------------------------------------------------*/
|
||
|
if ( expA < expB - 1 ) return a;
|
||
|
/*------------------------------------------------------------------------
|
||
|
*------------------------------------------------------------------------*/
|
||
|
if ( ! expB ) {
|
||
|
if ( ! sigB ) goto invalid;
|
||
|
normExpSig = softfloat_normSubnormalF64Sig( sigB );
|
||
|
expB = normExpSig.exp;
|
||
|
sigB = normExpSig.sig;
|
||
|
}
|
||
|
if ( ! expA ) {
|
||
|
if ( ! sigA ) return a;
|
||
|
normExpSig = softfloat_normSubnormalF64Sig( sigA );
|
||
|
expA = normExpSig.exp;
|
||
|
sigA = normExpSig.sig;
|
||
|
}
|
||
|
/*------------------------------------------------------------------------
|
||
|
*------------------------------------------------------------------------*/
|
||
|
rem = sigA | UINT64_C( 0x0010000000000000 );
|
||
|
sigB |= UINT64_C( 0x0010000000000000 );
|
||
|
expDiff = expA - expB;
|
||
|
if ( expDiff < 1 ) {
|
||
|
if ( expDiff < -1 ) return a;
|
||
|
sigB <<= 9;
|
||
|
if ( expDiff ) {
|
||
|
rem <<= 8;
|
||
|
q = 0;
|
||
|
} else {
|
||
|
rem <<= 9;
|
||
|
q = (sigB <= rem);
|
||
|
if ( q ) rem -= sigB;
|
||
|
}
|
||
|
} else {
|
||
|
recip32 = softfloat_approxRecip32_1( sigB>>21 );
|
||
|
/*--------------------------------------------------------------------
|
||
|
| Changing the shift of `rem' here requires also changing the initial
|
||
|
| subtraction from `expDiff'.
|
||
|
*--------------------------------------------------------------------*/
|
||
|
rem <<= 9;
|
||
|
expDiff -= 30;
|
||
|
/*--------------------------------------------------------------------
|
||
|
| The scale of `sigB' affects how many bits are obtained during each
|
||
|
| cycle of the loop. Currently this is 29 bits per loop iteration,
|
||
|
| the maximum possible.
|
||
|
*--------------------------------------------------------------------*/
|
||
|
sigB <<= 9;
|
||
|
for (;;) {
|
||
|
q64 = (uint32_t) (rem>>32) * (uint_fast64_t) recip32;
|
||
|
if ( expDiff < 0 ) break;
|
||
|
q = (q64 + 0x80000000)>>32;
|
||
|
#ifdef SOFTFLOAT_FAST_INT64
|
||
|
rem <<= 29;
|
||
|
#else
|
||
|
rem = (uint_fast64_t) (uint32_t) (rem>>3)<<32;
|
||
|
#endif
|
||
|
rem -= q * (uint64_t) sigB;
|
||
|
if ( rem & UINT64_C( 0x8000000000000000 ) ) rem += sigB;
|
||
|
expDiff -= 29;
|
||
|
}
|
||
|
/*--------------------------------------------------------------------
|
||
|
| (`expDiff' cannot be less than -29 here.)
|
||
|
*--------------------------------------------------------------------*/
|
||
|
q = (uint32_t) (q64>>32)>>(~expDiff & 31);
|
||
|
rem = (rem<<(expDiff + 30)) - q * (uint64_t) sigB;
|
||
|
if ( rem & UINT64_C( 0x8000000000000000 ) ) {
|
||
|
altRem = rem + sigB;
|
||
|
goto selectRem;
|
||
|
}
|
||
|
}
|
||
|
/*------------------------------------------------------------------------
|
||
|
*------------------------------------------------------------------------*/
|
||
|
do {
|
||
|
altRem = rem;
|
||
|
++q;
|
||
|
rem -= sigB;
|
||
|
} while ( ! (rem & UINT64_C( 0x8000000000000000 )) );
|
||
|
selectRem:
|
||
|
meanRem = rem + altRem;
|
||
|
if (
|
||
|
(meanRem & UINT64_C( 0x8000000000000000 )) || (! meanRem && (q & 1))
|
||
|
) {
|
||
|
rem = altRem;
|
||
|
}
|
||
|
signRem = signA;
|
||
|
if ( rem & UINT64_C( 0x8000000000000000 ) ) {
|
||
|
signRem = ! signRem;
|
||
|
rem = -rem;
|
||
|
}
|
||
|
return softfloat_normRoundPackToF64( signRem, expB, rem );
|
||
|
/*------------------------------------------------------------------------
|
||
|
*------------------------------------------------------------------------*/
|
||
|
propagateNaN:
|
||
|
uiZ = softfloat_propagateNaNF64UI( uiA, uiB );
|
||
|
goto uiZ;
|
||
|
invalid:
|
||
|
softfloat_raiseFlags( softfloat_flag_invalid );
|
||
|
uiZ = defaultNaNF64UI;
|
||
|
uiZ:
|
||
|
uZ.ui = uiZ;
|
||
|
return uZ.f;
|
||
|
|
||
|
}
|
||
|
|