github.com/aergoio/aergo@v1.3.1/libtool/src/gmp-6.1.2/mpn/cray/ieee/sqr_basecase.c (about) 1 /* Cray PVP/IEEE mpn_sqr_basecase. 2 3 Copyright 2000, 2001 Free Software Foundation, Inc. 4 5 This file is part of the GNU MP Library. 6 7 The GNU MP Library is free software; you can redistribute it and/or modify 8 it under the terms of either: 9 10 * the GNU Lesser General Public License as published by the Free 11 Software Foundation; either version 3 of the License, or (at your 12 option) any later version. 13 14 or 15 16 * the GNU General Public License as published by the Free Software 17 Foundation; either version 2 of the License, or (at your option) any 18 later version. 19 20 or both in parallel, as here. 21 22 The GNU MP Library is distributed in the hope that it will be useful, but 23 WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY 24 or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License 25 for more details. 26 27 You should have received copies of the GNU General Public License and the 28 GNU Lesser General Public License along with the GNU MP Library. If not, 29 see https://www.gnu.org/licenses/. */ 30 31 /* This is just mpn_mul_basecase with trivial modifications. */ 32 33 #include <intrinsics.h> 34 #include "gmp.h" 35 #include "gmp-impl.h" 36 37 void 38 mpn_sqr_basecase (mp_ptr rp, 39 mp_srcptr up, mp_size_t un) 40 { 41 mp_limb_t cy[un + un]; 42 mp_limb_t ul; 43 mp_limb_t a, b, r, s0, s1, c0, c1; 44 mp_size_t i, j; 45 int more_carries; 46 47 for (i = 0; i < un + un; i++) 48 { 49 rp[i] = 0; 50 cy[i] = 0; 51 } 52 53 #pragma _CRI novector 54 for (j = 0; j < un; j++) 55 { 56 ul = up[j]; 57 58 a = up[0] * ul; 59 r = rp[j]; 60 s0 = a + r; 61 rp[j] = s0; 62 c0 = ((a & r) | ((a | r) & ~s0)) >> 63; 63 cy[j] += c0; 64 65 #pragma _CRI ivdep 66 for (i = 1; i < un; i++) 67 { 68 a = up[i] * ul; 69 b = _int_mult_upper (up[i - 1], ul); 70 s0 = a + b; 71 c0 = ((a & b) | ((a | b) & ~s0)) >> 63; 72 r = rp[j + i]; 73 s1 = s0 + r; 74 rp[j + i] = s1; 75 c1 = ((s0 & r) | ((s0 | r) & ~s1)) >> 63; 76 cy[j + i] += c0 + c1; 77 } 78 rp[j + un] = _int_mult_upper (up[un - 1], ul); 79 } 80 81 more_carries = 0; 82 #pragma _CRI ivdep 83 for (i = 1; i < un + un; i++) 84 { 85 r = rp[i]; 86 c0 = cy[i - 1]; 87 s0 = r + c0; 88 rp[i] = s0; 89 c0 = (r & ~s0) >> 63; 90 more_carries += c0; 91 } 92 /* If that second loop generated carry, handle that in scalar loop. */ 93 if (more_carries) 94 { 95 mp_limb_t cyrec = 0; 96 for (i = 1; i < un + un; i++) 97 { 98 r = rp[i]; 99 c0 = (r < cy[i - 1]); 100 s0 = r + cyrec; 101 rp[i] = s0; 102 c1 = (r & ~s0) >> 63; 103 cyrec = c0 | c1; 104 } 105 } 106 }