|
Packit |
5c3484 |
/* Cray PVP mpn_sub_n -- subtract two limb vectors and store their difference
|
|
Packit |
5c3484 |
in a third limb vector.
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
Copyright 1996, 2000, 2001 Free Software 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 |
/* This code runs at 4 cycles/limb. It may be possible to bring it down
|
|
Packit |
5c3484 |
to 3 cycles/limb. */
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
#include "gmp.h"
|
|
Packit |
5c3484 |
#include "gmp-impl.h"
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
mp_limb_t
|
|
Packit |
5c3484 |
mpn_sub_n (mp_ptr rp, mp_srcptr up, mp_srcptr vp, mp_size_t n)
|
|
Packit |
5c3484 |
{
|
|
Packit |
5c3484 |
mp_limb_t cy[n];
|
|
Packit |
5c3484 |
mp_limb_t a, b, r, s0, c0, c1;
|
|
Packit |
5c3484 |
mp_size_t i;
|
|
Packit |
5c3484 |
int more_carries;
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
/* Main subtract loop. Generate a raw output difference in rp[] and a
|
|
Packit |
5c3484 |
borrow vector in cy[]. */
|
|
Packit |
5c3484 |
#pragma _CRI ivdep
|
|
Packit |
5c3484 |
for (i = 0; i < n; i++)
|
|
Packit |
5c3484 |
{
|
|
Packit |
5c3484 |
a = up[i];
|
|
Packit |
5c3484 |
b = vp[i];
|
|
Packit |
5c3484 |
s0 = a - b; /* a = s0 + b */
|
|
Packit |
5c3484 |
rp[i] = s0;
|
|
Packit |
5c3484 |
c0 = ((s0 & b) | ((s0 | b) & ~a)) >> 63;
|
|
Packit |
5c3484 |
cy[i] = c0;
|
|
Packit |
5c3484 |
}
|
|
Packit |
5c3484 |
/* Borrow subtract loop. Subtract the borrow vector cy[] from the raw
|
|
Packit |
5c3484 |
difference rp[] and store the new difference back to rp[0]. If this
|
|
Packit |
5c3484 |
generates further borrow, set more_carries. */
|
|
Packit |
5c3484 |
more_carries = 0;
|
|
Packit |
5c3484 |
#pragma _CRI ivdep
|
|
Packit |
5c3484 |
for (i = 1; i < n; i++)
|
|
Packit |
5c3484 |
{
|
|
Packit |
5c3484 |
r = rp[i];
|
|
Packit |
5c3484 |
c0 = cy[i - 1];
|
|
Packit |
5c3484 |
s0 = r - c0; /* r = s0 + c0 */
|
|
Packit |
5c3484 |
rp[i] = s0;
|
|
Packit |
5c3484 |
c0 = (s0 & ~r) >> 63;
|
|
Packit |
5c3484 |
more_carries += c0;
|
|
Packit |
5c3484 |
}
|
|
Packit |
5c3484 |
/* If that second loop generated borrow, handle that in scalar loop. */
|
|
Packit |
5c3484 |
if (more_carries)
|
|
Packit |
5c3484 |
{
|
|
Packit |
5c3484 |
mp_limb_t cyrec = 0;
|
|
Packit |
5c3484 |
/* Look for places where rp[k] contains just ones and cy[k-1] is
|
|
Packit |
5c3484 |
non-zero. These are where we got a recurrency borrow. */
|
|
Packit |
5c3484 |
for (i = 1; i < n; i++)
|
|
Packit |
5c3484 |
{
|
|
Packit |
5c3484 |
r = rp[i];
|
|
Packit |
5c3484 |
c0 = (~r == 0 && cy[i - 1] != 0);
|
|
Packit |
5c3484 |
s0 = r - cyrec;
|
|
Packit |
5c3484 |
rp[i] = s0;
|
|
Packit |
5c3484 |
c1 = (s0 & ~r) >> 63;
|
|
Packit |
5c3484 |
cyrec = c0 | c1;
|
|
Packit |
5c3484 |
}
|
|
Packit |
5c3484 |
return cyrec | cy[n - 1];
|
|
Packit |
5c3484 |
}
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
return cy[n - 1];
|
|
Packit |
5c3484 |
}
|