|
Packit |
5c3484 |
/* mpn_sqr -- square natural numbers.
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
Copyright 1991, 1993, 1994, 1996-2003, 2005, 2008, 2009 Free Software
|
|
Packit |
5c3484 |
Foundation, Inc.
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
This file is part of the GNU MP Library.
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
The GNU MP Library is free software; you can redistribute it and/or modify
|
|
Packit |
5c3484 |
it under the terms of either:
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
* the GNU Lesser General Public License as published by the Free
|
|
Packit |
5c3484 |
Software Foundation; either version 3 of the License, or (at your
|
|
Packit |
5c3484 |
option) any later version.
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
or
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
* the GNU General Public License as published by the Free Software
|
|
Packit |
5c3484 |
Foundation; either version 2 of the License, or (at your option) any
|
|
Packit |
5c3484 |
later version.
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
or both in parallel, as here.
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
The GNU MP Library is distributed in the hope that it will be useful, but
|
|
Packit |
5c3484 |
WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
|
|
Packit |
5c3484 |
or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
|
|
Packit |
5c3484 |
for more details.
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
You should have received copies of the GNU General Public License and the
|
|
Packit |
5c3484 |
GNU Lesser General Public License along with the GNU MP Library. If not,
|
|
Packit |
5c3484 |
see https://www.gnu.org/licenses/. */
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
#include "gmp.h"
|
|
Packit |
5c3484 |
#include "gmp-impl.h"
|
|
Packit |
5c3484 |
#include "longlong.h"
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
void
|
|
Packit |
5c3484 |
mpn_sqr (mp_ptr p, mp_srcptr a, mp_size_t n)
|
|
Packit |
5c3484 |
{
|
|
Packit |
5c3484 |
ASSERT (n >= 1);
|
|
Packit |
5c3484 |
ASSERT (! MPN_OVERLAP_P (p, 2 * n, a, n));
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
if (BELOW_THRESHOLD (n, SQR_BASECASE_THRESHOLD))
|
|
Packit |
5c3484 |
{ /* mul_basecase is faster than sqr_basecase on small sizes sometimes */
|
|
Packit |
5c3484 |
mpn_mul_basecase (p, a, n, a, n);
|
|
Packit |
5c3484 |
}
|
|
Packit |
5c3484 |
else if (BELOW_THRESHOLD (n, SQR_TOOM2_THRESHOLD))
|
|
Packit |
5c3484 |
{
|
|
Packit |
5c3484 |
mpn_sqr_basecase (p, a, n);
|
|
Packit |
5c3484 |
}
|
|
Packit |
5c3484 |
else if (BELOW_THRESHOLD (n, SQR_TOOM3_THRESHOLD))
|
|
Packit |
5c3484 |
{
|
|
Packit |
5c3484 |
/* Allocate workspace of fixed size on stack: fast! */
|
|
Packit |
5c3484 |
mp_limb_t ws[mpn_toom2_sqr_itch (SQR_TOOM3_THRESHOLD_LIMIT-1)];
|
|
Packit |
5c3484 |
ASSERT (SQR_TOOM3_THRESHOLD <= SQR_TOOM3_THRESHOLD_LIMIT);
|
|
Packit |
5c3484 |
mpn_toom2_sqr (p, a, n, ws);
|
|
Packit |
5c3484 |
}
|
|
Packit |
5c3484 |
else if (BELOW_THRESHOLD (n, SQR_TOOM4_THRESHOLD))
|
|
Packit |
5c3484 |
{
|
|
Packit |
5c3484 |
mp_ptr ws;
|
|
Packit |
5c3484 |
TMP_SDECL;
|
|
Packit |
5c3484 |
TMP_SMARK;
|
|
Packit |
5c3484 |
ws = TMP_SALLOC_LIMBS (mpn_toom3_sqr_itch (n));
|
|
Packit |
5c3484 |
mpn_toom3_sqr (p, a, n, ws);
|
|
Packit |
5c3484 |
TMP_SFREE;
|
|
Packit |
5c3484 |
}
|
|
Packit |
5c3484 |
else if (BELOW_THRESHOLD (n, SQR_TOOM6_THRESHOLD))
|
|
Packit |
5c3484 |
{
|
|
Packit |
5c3484 |
mp_ptr ws;
|
|
Packit |
5c3484 |
TMP_SDECL;
|
|
Packit |
5c3484 |
TMP_SMARK;
|
|
Packit |
5c3484 |
ws = TMP_SALLOC_LIMBS (mpn_toom4_sqr_itch (n));
|
|
Packit |
5c3484 |
mpn_toom4_sqr (p, a, n, ws);
|
|
Packit |
5c3484 |
TMP_SFREE;
|
|
Packit |
5c3484 |
}
|
|
Packit |
5c3484 |
else if (BELOW_THRESHOLD (n, SQR_TOOM8_THRESHOLD))
|
|
Packit |
5c3484 |
{
|
|
Packit |
5c3484 |
mp_ptr ws;
|
|
Packit |
5c3484 |
TMP_SDECL;
|
|
Packit |
5c3484 |
TMP_SMARK;
|
|
Packit |
5c3484 |
ws = TMP_SALLOC_LIMBS (mpn_toom6_sqr_itch (n));
|
|
Packit |
5c3484 |
mpn_toom6_sqr (p, a, n, ws);
|
|
Packit |
5c3484 |
TMP_SFREE;
|
|
Packit |
5c3484 |
}
|
|
Packit |
5c3484 |
else if (BELOW_THRESHOLD (n, SQR_FFT_THRESHOLD))
|
|
Packit |
5c3484 |
{
|
|
Packit |
5c3484 |
mp_ptr ws;
|
|
Packit |
5c3484 |
TMP_DECL;
|
|
Packit |
5c3484 |
TMP_MARK;
|
|
Packit |
5c3484 |
ws = TMP_ALLOC_LIMBS (mpn_toom8_sqr_itch (n));
|
|
Packit |
5c3484 |
mpn_toom8_sqr (p, a, n, ws);
|
|
Packit |
5c3484 |
TMP_FREE;
|
|
Packit |
5c3484 |
}
|
|
Packit |
5c3484 |
else
|
|
Packit |
5c3484 |
{
|
|
Packit |
5c3484 |
/* The current FFT code allocates its own space. That should probably
|
|
Packit |
5c3484 |
change. */
|
|
Packit |
5c3484 |
mpn_fft_mul (p, a, n, a, n);
|
|
Packit |
5c3484 |
}
|
|
Packit |
5c3484 |
}
|