softfloat.c

来自「基于4个mips核的noc设计」· C语言 代码 · 共 1,846 行 · 第 1/5 页

float32 float32_sqrt( float32 a )
{
    flag aSign;
    int16 aExp, zExp;
    bits32 aSig, zSig, rem0, rem1, term0, term1;

    aSig = extractFloat32Frac( a );
    aExp = extractFloat32Exp( a );
    aSign = extractFloat32Sign( a );
    if ( aExp == 0xFF ) {
        if ( aSig ) return propagateFloat32NaN( a, 0 );
        if ( ! aSign ) return a;
        float_raise( float_flag_invalid );
        return float32_default_nan;
    }
    if ( aSign ) {
        if ( ( aExp | aSig ) == 0 ) return a;
        float_raise( float_flag_invalid );
        return float32_default_nan;
    }
    if ( aExp == 0 ) {
        if ( aSig == 0 ) return 0;
        normalizeFloat32Subnormal( aSig, &aExp, &aSig );
    }
    zExp = ( ( aExp - 0x7F )>>1 ) + 0x7E;
    aSig = ( aSig | 0x00800000 )<<8;
    zSig = estimateSqrt32( aExp, aSig ) + 2;
    if ( ( zSig & 0x7F ) <= 5 ) {
        if ( zSig < 2 ) {
            zSig = 0x7FFFFFFF;
            goto roundAndPack;
        }
        else {
            aSig >>= aExp & 1;
            mul32To64( zSig, zSig, &term0, &term1 );
            sub64( aSig, 0, term0, term1, &rem0, &rem1 );
            while ( (sbits32) rem0 < 0 ) {
                --zSig;
                shortShift64Left( 0, zSig, 1, &term0, &term1 );
                term1 |= 1;
                add64( rem0, rem1, term0, term1, &rem0, &rem1 );
            }
            zSig |= ( ( rem0 | rem1 ) != 0 );
        }
    }
    shift32RightJamming( zSig, 1, &zSig );
 roundAndPack:
    return roundAndPackFloat32( 0, zExp, zSig );

}

/*----------------------------------------------------------------------------
| Returns 1 if the single-precision floating-point value `a' is equal to
| the corresponding value `b', and 0 otherwise.  The comparison is performed
| according to the IEC/IEEE Standard for Binary Floating-Point Arithmetic.
*----------------------------------------------------------------------------*/

flag float32_eq( float32 a, float32 b )
{

    if (    ( ( extractFloat32Exp( a ) == 0xFF ) && extractFloat32Frac( a ) )
         || ( ( extractFloat32Exp( b ) == 0xFF ) && extractFloat32Frac( b ) )
       ) {
        if ( float32_is_signaling_nan( a ) || float32_is_signaling_nan( b ) ) {
            float_raise( float_flag_invalid );
        }
        return 0;
    }
    return ( a == b ) || ( (bits32) ( ( a | b )<<1 ) == 0 );

}

/*----------------------------------------------------------------------------
| Returns 1 if the single-precision floating-point value `a' is less than
| or equal to the corresponding value `b', and 0 otherwise.  The comparison
| is performed according to the IEC/IEEE Standard for Binary Floating-Point
| Arithmetic.
*----------------------------------------------------------------------------*/

flag float32_le( float32 a, float32 b )
{
    flag aSign, bSign;

    if (    ( ( extractFloat32Exp( a ) == 0xFF ) && extractFloat32Frac( a ) )
         || ( ( extractFloat32Exp( b ) == 0xFF ) && extractFloat32Frac( b ) )
       ) {
        float_raise( float_flag_invalid );
        return 0;
    }
    aSign = extractFloat32Sign( a );
    bSign = extractFloat32Sign( b );
    if ( aSign != bSign ) return aSign || ( (bits32) ( ( a | b )<<1 ) == 0 );
    return ( a == b ) || ( aSign ^ ( a < b ) );

}

/*----------------------------------------------------------------------------
| Returns 1 if the single-precision floating-point value `a' is less than
| the corresponding value `b', and 0 otherwise.  The comparison is performed
| according to the IEC/IEEE Standard for Binary Floating-Point Arithmetic.
*----------------------------------------------------------------------------*/

flag float32_lt( float32 a, float32 b )
{
    flag aSign, bSign;

    if (    ( ( extractFloat32Exp( a ) == 0xFF ) && extractFloat32Frac( a ) )
         || ( ( extractFloat32Exp( b ) == 0xFF ) && extractFloat32Frac( b ) )
       ) {
        float_raise( float_flag_invalid );
        return 0;
    }
    aSign = extractFloat32Sign( a );
    bSign = extractFloat32Sign( b );
    if ( aSign != bSign ) return aSign && ( (bits32) ( ( a | b )<<1 ) != 0 );
    return ( a != b ) && ( aSign ^ ( a < b ) );

}

/*----------------------------------------------------------------------------
| Returns 1 if the single-precision floating-point value `a' is equal to
| the corresponding value `b', and 0 otherwise.  The invalid exception is
| raised if either operand is a NaN.  Otherwise, the comparison is performed
| according to the IEC/IEEE Standard for Binary Floating-Point Arithmetic.
*----------------------------------------------------------------------------*/

flag float32_eq_signaling( float32 a, float32 b )
{

    if (    ( ( extractFloat32Exp( a ) == 0xFF ) && extractFloat32Frac( a ) )
         || ( ( extractFloat32Exp( b ) == 0xFF ) && extractFloat32Frac( b ) )
       ) {
        float_raise( float_flag_invalid );
        return 0;
    }
    return ( a == b ) || ( (bits32) ( ( a | b )<<1 ) == 0 );

}

/*----------------------------------------------------------------------------
| Returns 1 if the single-precision floating-point value `a' is less than or
| equal to the corresponding value `b', and 0 otherwise.  Quiet NaNs do not
| cause an exception.  Otherwise, the comparison is performed according to the
| IEC/IEEE Standard for Binary Floating-Point Arithmetic.
*----------------------------------------------------------------------------*/

flag float32_le_quiet( float32 a, float32 b )
{
    flag aSign, bSign;
    int16 aExp, bExp;

    if (    ( ( extractFloat32Exp( a ) == 0xFF ) && extractFloat32Frac( a ) )
         || ( ( extractFloat32Exp( b ) == 0xFF ) && extractFloat32Frac( b ) )
       ) {
        if ( float32_is_signaling_nan( a ) || float32_is_signaling_nan( b ) ) {
            float_raise( float_flag_invalid );
        }
        return 0;
    }
    aSign = extractFloat32Sign( a );
    bSign = extractFloat32Sign( b );
    if ( aSign != bSign ) return aSign || ( (bits32) ( ( a | b )<<1 ) == 0 );
    return ( a == b ) || ( aSign ^ ( a < b ) );

}

/*----------------------------------------------------------------------------
| Returns 1 if the single-precision floating-point value `a' is less than
| the corresponding value `b', and 0 otherwise.  Quiet NaNs do not cause an
| exception.  Otherwise, the comparison is performed according to the IEC/IEEE
| Standard for Binary Floating-Point Arithmetic.
*----------------------------------------------------------------------------*/

flag float32_lt_quiet( float32 a, float32 b )
{
    flag aSign, bSign;

    if (    ( ( extractFloat32Exp( a ) == 0xFF ) && extractFloat32Frac( a ) )
         || ( ( extractFloat32Exp( b ) == 0xFF ) && extractFloat32Frac( b ) )
       ) {
        if ( float32_is_signaling_nan( a ) || float32_is_signaling_nan( b ) ) {
            float_raise( float_flag_invalid );
        }
        return 0;
    }
    aSign = extractFloat32Sign( a );
    bSign = extractFloat32Sign( b );
    if ( aSign != bSign ) return aSign && ( (bits32) ( ( a | b )<<1 ) != 0 );
    return ( a != b ) && ( aSign ^ ( a < b ) );

}

/*----------------------------------------------------------------------------
| Returns the result of converting the double-precision floating-point value
| `a' to the 32-bit two's complement integer format.  The conversion is
| performed according to the IEC/IEEE Standard for Binary Floating-Point
| Arithmetic---which means in particular that the conversion is rounded
| according to the current rounding mode.  If `a' is a NaN, the largest
| positive integer is returned.  Otherwise, if the conversion overflows, the
| largest integer with the same sign as `a' is returned.
*----------------------------------------------------------------------------*/

int32 float64_to_int32( float64 a )
{
    flag aSign;
    int16 aExp, shiftCount;
    bits32 aSig0, aSig1, absZ, aSigExtra;
    int32 z;
    int8 roundingMode;

    aSig1 = extractFloat64Frac1( a );
    aSig0 = extractFloat64Frac0( a );
    aExp = extractFloat64Exp( a );
    aSign = extractFloat64Sign( a );
    shiftCount = aExp - 0x413;
    if ( 0 <= shiftCount ) {
        if ( 0x41E < aExp ) {
            if ( ( aExp == 0x7FF ) && ( aSig0 | aSig1 ) ) aSign = 0;
            goto invalid;
        }
        shortShift64Left(
            aSig0 | 0x00100000, aSig1, shiftCount, &absZ, &aSigExtra );
        if ( 0x80000000 < absZ ) goto invalid;
    }
    else {
        aSig1 = ( aSig1 != 0 );
        if ( aExp < 0x3FE ) {
            aSigExtra = aExp | aSig0 | aSig1;
            absZ = 0;
        }
        else {
            aSig0 |= 0x00100000;
            aSigExtra = ( aSig0<<( shiftCount & 31 ) ) | aSig1;
            absZ = aSig0>>( - shiftCount );
        }
    }
    roundingMode = float_rounding_mode;
    if ( roundingMode == float_round_nearest_even ) {
        if ( (sbits32) aSigExtra < 0 ) {
            ++absZ;
            if ( (bits32) ( aSigExtra<<1 ) == 0 ) absZ &= ~1;
        }
        z = aSign ? - absZ : absZ;
    }
    else {
        aSigExtra = ( aSigExtra != 0 );
        if ( aSign ) {
            z = - (   absZ
                    + ( ( roundingMode == float_round_down ) & aSigExtra ) );
        }
        else {
            z = absZ + ( ( roundingMode == float_round_up ) & aSigExtra );
        }
    }
    if ( ( aSign ^ ( z < 0 ) ) && z ) {
 invalid:
        float_raise( float_flag_invalid );
        return aSign ? (sbits32) 0x80000000 : 0x7FFFFFFF;
    }
    if ( aSigExtra ) float_exception_flags |= float_flag_inexact;
    return z;

}

/*----------------------------------------------------------------------------
| Returns the result of converting the double-precision floating-point value
| `a' to the 32-bit two's complement integer format.  The conversion is
| performed according to the IEC/IEEE Standard for Binary Floating-Point
| Arithmetic, except that the conversion is always rounded toward zero.
| If `a' is a NaN, the largest positive integer is returned.  Otherwise, if
| the conversion overflows, the largest integer with the same sign as `a' is
| returned.
*----------------------------------------------------------------------------*/

int32 float64_to_int32_round_to_zero( float64 a )
{
    flag aSign;
    int16 aExp, shiftCount;
    bits32 aSig0, aSig1, absZ, aSigExtra;
    int32 z;

    aSig1 = extractFloat64Frac1( a );
    aSig0 = extractFloat64Frac0( a );
    aExp = extractFloat64Exp( a );
    aSign = extractFloat64Sign( a );
    shiftCount = aExp - 0x413;
    if ( 0 <= shiftCount ) {
        if ( 0x41E < aExp ) {
            if ( ( aExp == 0x7FF ) && ( aSig0 | aSig1 ) ) aSign = 0;
            goto invalid;
        }
        shortShift64Left(
            aSig0 | 0x00100000, aSig1, shiftCount, &absZ, &aSigExtra );
    }
    else {
        if ( aExp < 0x3FF ) {
            if ( aExp | aSig0 | aSig1 ) {
                float_exception_flags |= float_flag_inexact;
            }
            return 0;
        }
        aSig0 |= 0x00100000;
        aSigExtra = ( aSig0<<( shiftCount & 31 ) ) | aSig1;
        absZ = aSig0>>( - shiftCount );
    }
    z = aSign ? - absZ : absZ;
    if ( ( aSign ^ ( z < 0 ) ) && z ) {
 invalid:
        float_raise( float_flag_invalid );
        return aSign ? (sbits32) 0x80000000 : 0x7FFFFFFF;
    }
    if ( aSigExtra ) float_exception_flags |= float_flag_inexact;
    return z;

}

/*----------------------------------------------------------------------------
| Returns the result of converting the double-precision floating-point value
| `a' to the single-precision floating-point format.  The conversion is
| performed according to the IEC/IEEE Standard for Binary Floating-Point
| Arithmetic.
*----------------------------------------------------------------------------*/

float32 float64_to_float32( float64 a )
{
    flag aSign;
    int16 aExp;
    bits32 aSig0, aSig1, zSig;
    bits32 allZero;

    aSig1 = extractFloat64Frac1( a );
    aSig0 = extractFloat64Frac0( a );
    aExp = extractFloat64Exp( a );
    aSign = extractFloat64Sign( a );
    if ( aExp == 0x7FF ) {
        if ( aSig0 | aSig1 ) {
            return commonNaNToFloat32( float64ToCommonNaN( a ) );
        }
        return packFloat32( aSign, 0xFF, 0 );
    }
    shift64RightJamming( aSig0, aSig1, 22, &allZero, &zSig );
    if ( aExp ) zSig |= 0x40000000;
    return roundAndPackFloat32( aSign, aExp - 0x381, zSig );

}

/*----------------------------------------------------------------------------
| Rounds the double-precision floating-point value `a' to an integer,
| and returns the result as a double-precision floating-point value.  The
| operation is performed according to the IEC/IEEE Standard for Binary
| Floating-Point Arithmetic.
*----------------------------------------------------------------------------*/

float64 float64_round_to_int( float64 a )
{
    flag aSign;
    int16 aExp;
    bits32 lastBitMask, roundBitsMask;
    int8 roundingMode;
    float64 z;

    aExp = extractFloat64Exp( a );
    if ( 0x413 <= aExp ) {
        if ( 0x433 <= aExp ) {
            if (    ( aExp == 0x7FF )
                 && ( extractFloat64Frac0( a ) | extractFloat64Frac1( a ) ) ) {
                return propagateFloat64NaN( a, a );
            }
            return a;
        }