0,0 → 1,377 |
/* |
* Copyright (C) 2005 Josef Cejka |
* 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 disclaimer. |
* - Redistributions in binary form must reproduce the above copyright |
* notice, this list of conditions and the following disclaimer in the |
* documentation and/or other materials provided with the distribution. |
* - The name of the author may not be used to endorse or promote products |
* derived from this software without specific prior written permission. |
* |
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR |
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES |
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. |
* IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, |
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT |
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF |
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
*/ |
|
#include<sftypes.h> |
#include<arithmetic.h> |
#include<comparison.h> |
|
/** Add two Float32 numbers with same signs |
*/ |
float32 addFloat32(float32 a, float32 b) |
{ |
int expdiff; |
__u32 exp1,exp2,mant1,mant2; |
|
expdiff=a.parts.exp - b.parts.exp; |
if (expdiff<0) { |
if (isFloat32NaN(b)) { |
//TODO: fix SigNaN |
if (isFloat32SigNaN(b)) { |
}; |
|
return b; |
}; |
|
if (b.parts.exp==0xFF) { |
return b; |
} |
|
mant1=b.parts.mantisa; |
exp1=b.parts.exp; |
mant2=a.parts.mantisa; |
exp2=a.parts.exp; |
expdiff*=-1; |
} else { |
if (isFloat32NaN(a)) { |
//TODO: fix SigNaN |
if ((isFloat32SigNaN(a))||(isFloat32SigNaN(b))) { |
}; |
return a; |
}; |
|
if (a.parts.exp==0xFF) { |
return a; |
} |
|
mant1=a.parts.mantisa; |
exp1=a.parts.exp; |
mant2=b.parts.mantisa; |
exp2=b.parts.exp; |
}; |
|
if (exp1==0) { |
//both are denormalized |
mant1+=mant2; |
if (mant1&0xF00000) { |
a.parts.exp=1; |
}; |
a.parts.mantisa=mant1; |
return a; |
}; |
|
// create some space for rounding |
mant1<<=6; |
mant2<<=6; |
|
mant1|=0x20000000; //add hidden bit |
|
|
if (exp2==0) { |
--expdiff; |
} else { |
mant2|=0x20000000; //hidden bit |
}; |
|
if (expdiff>24) { |
goto done; |
}; |
|
mant2>>=expdiff; |
mant1+=mant2; |
done: |
if (mant1&0x40000000) { |
++exp1; |
mant1>>=1; |
}; |
|
//rounding - if first bit after mantisa is set then round up |
mant1+=0x20; |
|
if (mant1&0x40000000) { |
++exp1; |
mant1>>=1; |
}; |
|
a.parts.exp=exp1; |
a.parts.mantisa = ((mant1&(~0x20000000))>>6); /*Clear hidden bit and shift */ |
return a; |
}; |
|
/** Subtract two float32 numbers with same signs |
*/ |
float32 subFloat32(float32 a, float32 b) |
{ |
int expdiff; |
__u32 exp1,exp2,mant1,mant2; |
float32 result; |
|
result.f = 0; |
|
expdiff=a.parts.exp - b.parts.exp; |
if ((expdiff<0)||((expdiff==0)&&(a.parts.mantisa<b.parts.mantisa))) { |
if (isFloat32NaN(b)) { |
//TODO: fix SigNaN |
if (isFloat32SigNaN(b)) { |
}; |
return b; |
}; |
|
if (b.parts.exp==0xFF) { |
b.parts.sign=!b.parts.sign; /* num -(+-inf) = -+inf */ |
return b; |
} |
|
result.parts.sign = !a.parts.sign; |
|
mant1=b.parts.mantisa; |
exp1=b.parts.exp; |
mant2=a.parts.mantisa; |
exp2=a.parts.exp; |
expdiff*=-1; |
} else { |
if (isFloat32NaN(a)) { |
//TODO: fix SigNaN |
if ((isFloat32SigNaN(a))||(isFloat32SigNaN(b))) { |
}; |
return a; |
}; |
|
if (a.parts.exp==0xFF) { |
if (b.parts.exp==0xFF) { |
/* inf - inf => nan */ |
//TODO: fix exception |
result.binary = FLOAT32_NAN; |
return result; |
}; |
return a; |
} |
|
result.parts.sign = a.parts.sign; |
|
mant1=a.parts.mantisa; |
exp1=a.parts.exp; |
mant2=b.parts.mantisa; |
exp2=b.parts.exp; |
|
|
|
}; |
|
if (exp1==0) { |
//both are denormalized |
result.parts.mantisa=mant1-mant2; |
if (result.parts.mantisa>mant1) { |
//TODO: underflow exception |
return result; |
}; |
result.parts.exp=0; |
return result; |
}; |
|
// create some space for rounding |
mant1<<=6; |
mant2<<=6; |
|
mant1|=0x20000000; //add hidden bit |
|
|
if (exp2==0) { |
--expdiff; |
} else { |
mant2|=0x20000000; //hidden bit |
}; |
|
if (expdiff>24) { |
goto done; |
}; |
|
mant1 = mant1-(mant2>>expdiff); |
done: |
|
//TODO: find first nonzero digit and shift result and detect possibly underflow |
while ((exp1>0)&&(!(mant1&0x20000000))) { |
exp1--; |
mant1 <<= 1; |
if(mant1 == 0) { |
/* Realy is it an underflow? ... */ |
/* TODO: fix underflow */ |
}; |
}; |
|
//rounding - if first bit after mantisa is set then round up |
mant1 += 0x20; |
|
if (mant1&0x40000000) { |
++exp1; |
mant1>>=1; |
}; |
|
result.parts.mantisa = ((mant1&(~0x20000000))>>6); /*Clear hidden bit and shift */ |
result.parts.exp = exp1; |
|
return result; |
}; |
|
/** Multiply two 32 bit float numbers |
* |
*/ |
float32 mulFloat32(float32 a, float32 b) |
{ |
float32 result; |
__u64 mant1, mant2; |
__s32 exp; |
|
result.parts.sign = a.parts.sign ^ b.parts.sign; |
|
if ((isFloat32NaN(a))||(isFloat32NaN(b))) { |
/* TODO: fix SigNaNs */ |
if (isFloat32SigNaN(a)) { |
result.parts.mantisa = a.parts.mantisa; |
result.parts.exp = a.parts.exp; |
return result; |
}; |
if (isFloat32SigNaN(b)) { /* TODO: fix SigNaN */ |
result.parts.mantisa = b.parts.mantisa; |
result.parts.exp = b.parts.exp; |
return result; |
}; |
/* set NaN as result */ |
result.parts.mantisa = 0x1; |
result.parts.exp = 0xFF; |
return result; |
}; |
|
if (isFloat32Infinity(a)) { |
if (isFloat32Zero(b)) { |
/* FIXME: zero * infinity */ |
result.parts.mantisa = 0x1; |
result.parts.exp = 0xFF; |
return result; |
} |
result.parts.mantisa = a.parts.mantisa; |
result.parts.exp = a.parts.exp; |
return result; |
} |
|
if (isFloat32Infinity(b)) { |
if (isFloat32Zero(a)) { |
/* FIXME: zero * infinity */ |
result.parts.mantisa = 0x1; |
result.parts.exp = 0xFF; |
return result; |
} |
result.parts.mantisa = b.parts.mantisa; |
result.parts.exp = b.parts.exp; |
return result; |
} |
|
/* exp is signed so we can easy detect underflow */ |
exp = a.parts.exp + b.parts.exp; |
exp -= FLOAT32_BIAS; |
|
if (exp >= 0xFF ) { |
/* FIXME: overflow */ |
/* set infinity as result */ |
result.parts.mantisa = 0x0; |
result.parts.exp = 0xFF; |
return result; |
}; |
|
if (exp < 0) { |
/* FIXME: underflow */ |
/* return signed zero */ |
result.parts.mantisa = 0x0; |
result.parts.exp = 0x0; |
return result; |
}; |
|
mant1 = a.parts.mantisa; |
if (a.parts.exp>0) { |
mant1 |= 0x800000; |
} else { |
++exp; |
}; |
|
mant2 = b.parts.mantisa; |
if (b.parts.exp>0) { |
mant2 |= 0x800000; |
} else { |
++exp; |
}; |
|
mant1 <<= 1; /* one bit space for rounding */ |
|
mant1 = mant1 * mant2; |
/* round and return */ |
|
while ((exp < 0xFF )&&(mant1 > 0x1FFFFFF )) { /* 0xFFFFFF is 23 bits of mantisa + one more for hidden bit (all shifted 1 bit left)*/ |
++exp; |
mant1 >>= 1; |
}; |
|
/* rounding */ |
//++mant1; /* FIXME: not works - without it is ok */ |
mant1 >>= 1; /* shift off rounding space */ |
|
if ((exp < 0xFF )&&(mant1 > 0xFFFFFF )) { |
++exp; |
mant1 >>= 1; |
}; |
|
if (exp >= 0xFF ) { |
/* TODO: fix overflow */ |
/* return infinity*/ |
result.parts.exp = 0xFF; |
result.parts.mantisa = 0x0; |
return result; |
} |
|
exp -= FLOAT32_MANTISA_SIZE; |
|
if (exp <= FLOAT32_MANTISA_SIZE) { |
/* denormalized number */ |
mant1 >>= 1; /* denormalize */ |
while ((mant1 > 0) && (exp < 0)) { |
mant1 >>= 1; |
++exp; |
}; |
if (mant1 == 0) { |
/* FIXME : underflow */ |
result.parts.exp = 0; |
result.parts.mantisa = 0; |
return result; |
}; |
}; |
result.parts.exp = exp; |
result.parts.mantisa = mant1 & 0x7FFFFF; |
|
return result; |
|
}; |
|
|