tommath/bn_fast_s_mp_sqr.c

115 lines
2.9 KiB
C
Raw Normal View History

#include <tommath_private.h>
2004-10-29 18:07:18 -04:00
#ifdef BN_FAST_S_MP_SQR_C
2003-02-28 11:08:34 -05:00
/* LibTomMath, multiple-precision integer library -- Tom St Denis
*
2003-08-04 21:24:44 -04:00
* LibTomMath is a library that provides multiple-precision
2003-02-28 11:08:34 -05:00
* integer arithmetic as well as number theoretic functionality.
*
2003-08-04 21:24:44 -04:00
* The library was designed directly after the MPI library by
2003-02-28 11:08:34 -05:00
* Michael Fromberger but has been written from scratch with
* additional optimizations in place.
*
* The library is free for all purposes without any express
* guarantee it works.
*
* Tom St Denis, tstdenis82@gmail.com, http://libtom.org
2003-02-28 11:08:34 -05:00
*/
2004-10-29 18:07:18 -04:00
/* the jist of squaring...
2005-03-12 06:55:11 -05:00
* you do like mult except the offset of the tmpx [one that
* starts closer to zero] can't equal the offset of tmpy.
* So basically you set up iy like before then you min it with
* (ty-tx) so that it never happens. You double all those
* you add in the inner loop
2004-10-29 18:07:18 -04:00
After that loop you do the squares and add them in.
*/
2004-04-11 16:46:22 -04:00
int fast_s_mp_sqr (mp_int * a, mp_int * b)
2003-02-28 11:08:34 -05:00
{
2004-10-29 18:07:18 -04:00
int olduse, res, pa, ix, iz;
mp_digit W[MP_WARRAY], *tmpx;
mp_word W1;
/* grow the destination as required */
pa = a->used + a->used;
if (b->alloc < pa) {
if ((res = mp_grow (b, pa)) != MP_OKAY) {
2003-02-28 11:08:34 -05:00
return res;
}
}
2004-10-29 18:07:18 -04:00
/* number of output digits to produce */
W1 = 0;
2004-12-22 21:40:37 -05:00
for (ix = 0; ix < pa; ix++) {
2004-10-29 18:07:18 -04:00
int tx, ty, iy;
mp_word _W;
mp_digit *tmpy;
/* clear counter */
_W = 0;
/* get offsets into the two bignums */
ty = MIN(a->used-1, ix);
tx = ix - ty;
/* setup temp aliases */
tmpx = a->dp + tx;
tmpy = a->dp + ty;
2005-03-12 06:55:11 -05:00
/* this is the number of times the loop will iterrate, essentially
2004-10-29 18:07:18 -04:00
while (tx++ < a->used && ty-- >= 0) { ... }
*/
iy = MIN(a->used-tx, ty+1);
/* now for squaring tx can never equal ty
* we halve the distance since they approach at a rate of 2x
* and we have to round because odd cases need to be executed
*/
iy = MIN(iy, ((ty-tx)+1)>>1);
2004-10-29 18:07:18 -04:00
/* execute loop */
for (iz = 0; iz < iy; iz++) {
_W += ((mp_word)*tmpx++)*((mp_word)*tmpy--);
2003-02-28 11:08:34 -05:00
}
2004-10-29 18:07:18 -04:00
/* double the inner product and add carry */
_W = _W + _W + W1;
/* even columns have the square term in them */
if ((ix&1) == 0) {
_W += ((mp_word)a->dp[ix>>1])*((mp_word)a->dp[ix>>1]);
}
/* store it */
2005-03-12 06:55:11 -05:00
W[ix] = (mp_digit)(_W & MP_MASK);
2004-10-29 18:07:18 -04:00
/* make next carry */
W1 = _W >> ((mp_word)DIGIT_BIT);
2003-02-28 11:08:34 -05:00
}
/* setup dest */
2003-05-29 09:35:26 -04:00
olduse = b->used;
2004-10-29 18:07:18 -04:00
b->used = a->used+a->used;
2003-02-28 11:08:34 -05:00
2004-10-29 18:07:18 -04:00
{
mp_digit *tmpb;
2003-05-29 09:35:26 -04:00
tmpb = b->dp;
2004-10-29 18:07:18 -04:00
for (ix = 0; ix < pa; ix++) {
*tmpb++ = W[ix] & MP_MASK;
2003-02-28 11:09:08 -05:00
}
2004-10-29 18:07:18 -04:00
/* clear unused digits [that existed in the old copy of c] */
2003-02-28 11:09:08 -05:00
for (; ix < olduse; ix++) {
*tmpb++ = 0;
}
}
2003-02-28 11:08:34 -05:00
mp_clamp (b);
return MP_OKAY;
}
2004-10-29 18:07:18 -04:00
#endif
2005-08-01 12:37:28 -04:00
2017-08-28 10:27:26 -04:00
/* ref: $Format:%D$ */
/* git commit: $Format:%H$ */
/* commit time: $Format:%ai$ */