/* -*-c-*-
*
- * $Id: mpx-ksqr.c,v 1.4 2000/07/29 17:04:02 mdw Exp $
+ * $Id$
*
* Karatsuba-based squaring algorithm
*
* (c) 1999 Straylight/Edgeware
*/
-/*----- Licensing notice --------------------------------------------------*
+/*----- Licensing notice --------------------------------------------------*
*
* This file is part of Catacomb.
*
* it under the terms of the GNU Library General Public License as
* published by the Free Software Foundation; either version 2 of the
* License, or (at your option) any later version.
- *
+ *
* Catacomb is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU Library General Public License for more details.
- *
+ *
* You should have received a copy of the GNU Library General Public
* License along with Catacomb; if not, write to the Free
* Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
* MA 02111-1307, USA.
*/
-/*----- Revision history --------------------------------------------------*
- *
- * $Log: mpx-ksqr.c,v $
- * Revision 1.4 2000/07/29 17:04:02 mdw
- * Remove useless header `mpscan.h'.
- *
- * Revision 1.3 2000/06/17 11:42:54 mdw
- * Moved the Karatsuba macros into a separate file for better sharing.
- * Fixed some comments. Use an improved technique so that all the
- * operations are squarings.
- *
- * Revision 1.2 1999/12/13 15:35:01 mdw
- * Simplify and improve.
- *
- * Revision 1.1 1999/12/11 10:57:43 mdw
- * Karatsuba squaring algorithm.
- *
- */
-
/*----- Header files ------------------------------------------------------*/
#include <assert.h>
#include <stdio.h>
#include "mpx.h"
-#include "mpx-kmac.h"
+#include "karatsuba.h"
/*----- Tweakables --------------------------------------------------------*/
#ifdef TEST_RIG
-# undef KARATSUBA_CUTOFF
-# define KARATSUBA_CUTOFF 2
+# undef MPK_THRESH
+# define MPK_THRESH 4
#endif
/*----- Main code ---------------------------------------------------------*/
* large numbers, although more expensive on small ones, and
* rather simpler than full-blown Karatsuba multiplication.
*
- * The destination must be twice as large as the argument. The
- * scratch space must be twice as large as the argument, plus
- * the magic number @KARATSUBA_SLOP@.
+ * The destination must be three times as large as the larger
+ * argument. The scratch space must be five times as large as
+ * the larger argument.
*/
void mpx_ksqr(mpw *dv, mpw *dvl,
MPX_SHRINK(av, avl);
- if (avl - av <= KARATSUBA_CUTOFF) {
+ if (avl - av <= MPK_THRESH) {
mpx_usqr(dv, dvl, av, avl);
return;
}
m = (avl - av + 1) >> 1;
avm = av + m;
- assert(((void)"Destination too small for Karatsuba square",
- dvl - dv >= 4 * m));
- assert(((void)"Not enough workspace for Karatsuba square",
- svl - sv >= 4 * m));
-
/* --- Sort out everything --- */
{
mpw *tdv = dv + m;
mpw *rdv = tdv + m;
+ assert(rdv + m + 4 < dvl);
+ assert(ssv < svl);
UADD2(sv, svm, av, avm, avm, avl);
- if (m > KARATSUBA_CUTOFF)
+ if (m > MPK_THRESH)
mpx_ksqr(tdv, rdv + m + 4, sv, svm + 1, ssv, svl);
else
mpx_usqr(tdv, rdv + m + 4, sv, svm + 1);
- if (m > KARATSUBA_CUTOFF)
+ if (m > MPK_THRESH)
mpx_ksqr(sv, ssv, avm, avl, ssv, svl);
else
mpx_usqr(sv, ssv, avm, avl);
MPX_COPY(rdv + m + 1, dvl, svm + 1, svn);
UADD(rdv, sv, svm + 1);
USUB(tdv, sv, svn);
-
- if (m > KARATSUBA_CUTOFF)
+
+ if (m > MPK_THRESH)
mpx_ksqr(sv, ssv, av, avm, ssv, svl);
else
mpx_usqr(sv, ssv, av, avm);
#include <mLib/alloc.h>
#include <mLib/testrig.h>
-#define ALLOC(v, vl, sz) do { \
- size_t _sz = (sz); \
- mpw *_vv = xmalloc(MPWS(_sz)); \
- mpw *_vvl = _vv + _sz; \
- (v) = _vv; \
- (vl) = _vvl; \
+#define ALLOC(v, vl, sz) do { \
+ size_t _sz = (sz); \
+ mpw *_vv = xmalloc(MPWS(_sz)); \
+ mpw *_vvl = _vv + _sz; \
+ (v) = _vv; \
+ (vl) = _vvl; \
} while (0)
-#define LOAD(v, vl, d) do { \
- const dstr *_d = (d); \
- mpw *_v, *_vl; \
- ALLOC(_v, _vl, MPW_RQ(_d->len)); \
- mpx_loadb(_v, _vl, _d->buf, _d->len); \
- (v) = _v; \
- (vl) = _vl; \
+#define LOAD(v, vl, d) do { \
+ const dstr *_d = (d); \
+ mpw *_v, *_vl; \
+ ALLOC(_v, _vl, MPW_RQ(_d->len)); \
+ mpx_loadb(_v, _vl, _d->buf, _d->len); \
+ (v) = _v; \
+ (vl) = _vl; \
} while (0)
#define MAX(x, y) ((x) > (y) ? (x) : (y))
LOAD(a, al, &v[0]);
LOAD(c, cl, &v[1]);
m = al - a + 1;
- ALLOC(d, dl, 2 * m);
- ALLOC(s, sl, 2 * m + 32);
+ ALLOC(d, dl, 3 * m);
+ ALLOC(s, sl, 5 * m);
mpx_ksqr(d, dl, a, al, s, sl);
- if (MPX_UCMP(d, dl, !=, c, cl)) {
+ if (!mpx_ueq(d, dl, c, cl)) {
fprintf(stderr, "\n*** usqr failed\n");
- dumpmp(" a", a, al);
+ dumpmp(" a", a, al);
dumpmp("expected", c, cl);
dumpmp(" result", d, dl);
ok = 0;
}
- free(a); free(c); free(d); free(s);
+ xfree(a); xfree(c); xfree(d); xfree(s);
return (ok);
}