From 11da511c784eca003deb90c23570f0873954e0de Mon Sep 17 00:00:00 2001 From: Duncan Wilkie Date: Sat, 18 Nov 2023 06:11:09 -0600 Subject: Initial commit. --- gmp-6.3.0/mpn/x86_64/nano/copyd.asm | 37 +++++ gmp-6.3.0/mpn/x86_64/nano/copyi.asm | 37 +++++ gmp-6.3.0/mpn/x86_64/nano/dive_1.asm | 166 ++++++++++++++++++++++ gmp-6.3.0/mpn/x86_64/nano/gcd_11.asm | 37 +++++ gmp-6.3.0/mpn/x86_64/nano/gmp-mparam.h | 243 +++++++++++++++++++++++++++++++++ gmp-6.3.0/mpn/x86_64/nano/popcount.asm | 35 +++++ 6 files changed, 555 insertions(+) create mode 100644 gmp-6.3.0/mpn/x86_64/nano/copyd.asm create mode 100644 gmp-6.3.0/mpn/x86_64/nano/copyi.asm create mode 100644 gmp-6.3.0/mpn/x86_64/nano/dive_1.asm create mode 100644 gmp-6.3.0/mpn/x86_64/nano/gcd_11.asm create mode 100644 gmp-6.3.0/mpn/x86_64/nano/gmp-mparam.h create mode 100644 gmp-6.3.0/mpn/x86_64/nano/popcount.asm (limited to 'gmp-6.3.0/mpn/x86_64/nano') diff --git a/gmp-6.3.0/mpn/x86_64/nano/copyd.asm b/gmp-6.3.0/mpn/x86_64/nano/copyd.asm new file mode 100644 index 0000000..f0dc54a --- /dev/null +++ b/gmp-6.3.0/mpn/x86_64/nano/copyd.asm @@ -0,0 +1,37 @@ +dnl X86-64 mpn_copyd optimised for Intel Sandy Bridge. + +dnl Copyright 2012 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. +dnl +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of either: +dnl +dnl * the GNU Lesser General Public License as published by the Free +dnl Software Foundation; either version 3 of the License, or (at your +dnl option) any later version. +dnl +dnl or +dnl +dnl * the GNU General Public License as published by the Free Software +dnl Foundation; either version 2 of the License, or (at your option) any +dnl later version. +dnl +dnl or both in parallel, as here. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, but +dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +dnl for more details. +dnl +dnl You should have received copies of the GNU General Public License and the +dnl GNU Lesser General Public License along with the GNU MP Library. If not, +dnl see https://www.gnu.org/licenses/. + +include(`../config.m4') + +ABI_SUPPORT(DOS64) +ABI_SUPPORT(STD64) + +MULFUNC_PROLOGUE(mpn_copyd) +include_mpn(`x86_64/fastsse/copyd-palignr.asm') diff --git a/gmp-6.3.0/mpn/x86_64/nano/copyi.asm b/gmp-6.3.0/mpn/x86_64/nano/copyi.asm new file mode 100644 index 0000000..9c26e00 --- /dev/null +++ b/gmp-6.3.0/mpn/x86_64/nano/copyi.asm @@ -0,0 +1,37 @@ +dnl X86-64 mpn_copyi optimised for Intel Sandy Bridge. + +dnl Copyright 2012 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. +dnl +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of either: +dnl +dnl * the GNU Lesser General Public License as published by the Free +dnl Software Foundation; either version 3 of the License, or (at your +dnl option) any later version. +dnl +dnl or +dnl +dnl * the GNU General Public License as published by the Free Software +dnl Foundation; either version 2 of the License, or (at your option) any +dnl later version. +dnl +dnl or both in parallel, as here. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, but +dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +dnl for more details. +dnl +dnl You should have received copies of the GNU General Public License and the +dnl GNU Lesser General Public License along with the GNU MP Library. If not, +dnl see https://www.gnu.org/licenses/. + +include(`../config.m4') + +ABI_SUPPORT(DOS64) +ABI_SUPPORT(STD64) + +MULFUNC_PROLOGUE(mpn_copyi) +include_mpn(`x86_64/fastsse/copyi-palignr.asm') diff --git a/gmp-6.3.0/mpn/x86_64/nano/dive_1.asm b/gmp-6.3.0/mpn/x86_64/nano/dive_1.asm new file mode 100644 index 0000000..e9a0763 --- /dev/null +++ b/gmp-6.3.0/mpn/x86_64/nano/dive_1.asm @@ -0,0 +1,166 @@ +dnl AMD64 mpn_divexact_1 -- mpn by limb exact division. + +dnl Copyright 2001, 2002, 2004-2006, 2010-2012 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. +dnl +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of either: +dnl +dnl * the GNU Lesser General Public License as published by the Free +dnl Software Foundation; either version 3 of the License, or (at your +dnl option) any later version. +dnl +dnl or +dnl +dnl * the GNU General Public License as published by the Free Software +dnl Foundation; either version 2 of the License, or (at your option) any +dnl later version. +dnl +dnl or both in parallel, as here. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, but +dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +dnl for more details. +dnl +dnl You should have received copies of the GNU General Public License and the +dnl GNU Lesser General Public License along with the GNU MP Library. If not, +dnl see https://www.gnu.org/licenses/. + +include(`../config.m4') + + +C cycles/limb +C norm unorm +C AMD K8,K9 11 11 +C AMD K10 11 11 +C Intel P4 ? +C Intel core2 13.5 13.25 +C Intel corei 14.25 +C Intel atom 34 36 +C VIA nano 19.25 19.25 + + +C INPUT PARAMETERS +C rp rdi +C up rsi +C n rdx +C divisor rcx + +ABI_SUPPORT(DOS64) +ABI_SUPPORT(STD64) + +ASM_START() + TEXT + ALIGN(16) +PROLOGUE(mpn_divexact_1) + FUNC_ENTRY(4) + push %rbx + + mov %rcx, %rax + xor R32(%rcx), R32(%rcx) C shift count + mov %rdx, %r8 + + bt $0, R32(%rax) + jc L(odd) C skip bsfq unless divisor is even + bsf %rax, %rcx + shr R8(%rcx), %rax +L(odd): mov %rax, %rbx + shr R32(%rax) + and $127, R32(%rax) C d/2, 7 bits + + LEA( binvert_limb_table, %rdx) + + movzbl (%rdx,%rax), R32(%rax) C inv 8 bits + + mov %rbx, %r11 C d without twos + + lea (%rax,%rax), R32(%rdx) C 2*inv + imul R32(%rax), R32(%rax) C inv*inv + imul R32(%rbx), R32(%rax) C inv*inv*d + sub R32(%rax), R32(%rdx) C inv = 2*inv - inv*inv*d, 16 bits + + lea (%rdx,%rdx), R32(%rax) C 2*inv + imul R32(%rdx), R32(%rdx) C inv*inv + imul R32(%rbx), R32(%rdx) C inv*inv*d + sub R32(%rdx), R32(%rax) C inv = 2*inv - inv*inv*d, 32 bits + + lea (%rax,%rax), %r10 C 2*inv + imul %rax, %rax C inv*inv + imul %rbx, %rax C inv*inv*d + sub %rax, %r10 C inv = 2*inv - inv*inv*d, 64 bits + + lea (%rsi,%r8,8), %rsi C up end + lea -8(%rdi,%r8,8), %rdi C rp end + neg %r8 C -n + + mov (%rsi,%r8,8), %rax C up[0] + + inc %r8 + jz L(one) + + test R32(%rcx), R32(%rcx) + jnz L(unorm) C branch if count != 0 + xor R32(%rbx), R32(%rbx) + jmp L(nent) + + ALIGN(8) +L(ntop):mul %r11 C carry limb in rdx 0 10 + mov -8(%rsi,%r8,8), %rax C + sub %rbx, %rax C apply carry bit + setc %bl C + sub %rdx, %rax C apply carry limb 5 + adc $0, %rbx C 6 +L(nent):imul %r10, %rax C 6 + mov %rax, (%rdi,%r8,8) C + inc %r8 C + jnz L(ntop) + + mov -8(%rsi), %r9 C up high limb + jmp L(com) + +L(unorm): + mov (%rsi,%r8,8), %r9 C up[1] + shr R8(%rcx), %rax C + neg R32(%rcx) + shl R8(%rcx), %r9 C + neg R32(%rcx) + or %r9, %rax + xor R32(%rbx), R32(%rbx) + jmp L(uent) + + ALIGN(8) +L(utop):mul %r11 C carry limb in rdx 0 10 + mov (%rsi,%r8,8), %rax C + shl R8(%rcx), %rax C + neg R32(%rcx) + or %r9, %rax + sub %rbx, %rax C apply carry bit + setc %bl C + sub %rdx, %rax C apply carry limb 5 + adc $0, %rbx C 6 +L(uent):imul %r10, %rax C 6 + mov (%rsi,%r8,8), %r9 C + shr R8(%rcx), %r9 C + neg R32(%rcx) + mov %rax, (%rdi,%r8,8) C + inc %r8 C + jnz L(utop) + +L(com): mul %r11 C carry limb in rdx + sub %rbx, %r9 C apply carry bit + sub %rdx, %r9 C apply carry limb + imul %r10, %r9 + mov %r9, (%rdi) + pop %rbx + FUNC_EXIT() + ret + +L(one): shr R8(%rcx), %rax + imul %r10, %rax + mov %rax, (%rdi) + pop %rbx + FUNC_EXIT() + ret +EPILOGUE() diff --git a/gmp-6.3.0/mpn/x86_64/nano/gcd_11.asm b/gmp-6.3.0/mpn/x86_64/nano/gcd_11.asm new file mode 100644 index 0000000..4723093 --- /dev/null +++ b/gmp-6.3.0/mpn/x86_64/nano/gcd_11.asm @@ -0,0 +1,37 @@ +dnl AMD64 mpn_gcd_11. + +dnl Copyright 2012 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. +dnl +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of either: +dnl +dnl * the GNU Lesser General Public License as published by the Free +dnl Software Foundation; either version 3 of the License, or (at your +dnl option) any later version. +dnl +dnl or +dnl +dnl * the GNU General Public License as published by the Free Software +dnl Foundation; either version 2 of the License, or (at your option) any +dnl later version. +dnl +dnl or both in parallel, as here. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, but +dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +dnl for more details. +dnl +dnl You should have received copies of the GNU General Public License and the +dnl GNU Lesser General Public License along with the GNU MP Library. If not, +dnl see https://www.gnu.org/licenses/. + +include(`../config.m4') + +ABI_SUPPORT(DOS64) +ABI_SUPPORT(STD64) + +MULFUNC_PROLOGUE(mpn_gcd_11) +include_mpn(`x86_64/core2/gcd_11.asm') diff --git a/gmp-6.3.0/mpn/x86_64/nano/gmp-mparam.h b/gmp-6.3.0/mpn/x86_64/nano/gmp-mparam.h new file mode 100644 index 0000000..fde69db --- /dev/null +++ b/gmp-6.3.0/mpn/x86_64/nano/gmp-mparam.h @@ -0,0 +1,243 @@ +/* VIA Nano gmp-mparam.h -- Compiler/machine parameter header file. + +Copyright 1991, 1993, 1994, 2000-2010, 2012, 2014 Free Software Foundation, +Inc. + +This file is part of the GNU MP Library. + +The GNU MP Library is free software; you can redistribute it and/or modify +it under the terms of either: + + * the GNU Lesser General Public License as published by the Free + Software Foundation; either version 3 of the License, or (at your + option) any later version. + +or + + * the GNU General Public License as published by the Free Software + Foundation; either version 2 of the License, or (at your option) any + later version. + +or both in parallel, as here. + +The GNU MP Library 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 General Public License +for more details. + +You should have received copies of the GNU General Public License and the +GNU Lesser General Public License along with the GNU MP Library. If not, +see https://www.gnu.org/licenses/. */ + +#define GMP_LIMB_BITS 64 +#define GMP_LIMB_BYTES 8 + +#define SHLD_SLOW 1 +#define SHRD_SLOW 1 + +/* 1600 MHz Nano 2xxx */ +/* FFT tuning limit = 25000000 */ +/* Generated by tuneup.c, 2014-03-12, gcc 4.2 */ + +#define MOD_1_NORM_THRESHOLD 0 /* always */ +#define MOD_1_UNNORM_THRESHOLD 0 /* always */ +#define MOD_1N_TO_MOD_1_1_THRESHOLD 4 +#define MOD_1U_TO_MOD_1_1_THRESHOLD 2 +#define MOD_1_1_TO_MOD_1_2_THRESHOLD 18 +#define MOD_1_2_TO_MOD_1_4_THRESHOLD 20 +#define PREINV_MOD_1_TO_MOD_1_THRESHOLD 8 +#define USE_PREINV_DIVREM_1 1 /* native */ +#define DIV_QR_1_NORM_THRESHOLD 1 +#define DIV_QR_1_UNNORM_THRESHOLD MP_SIZE_T_MAX /* never */ +#define DIV_QR_2_PI2_THRESHOLD MP_SIZE_T_MAX /* never */ +#define DIVEXACT_1_THRESHOLD 0 /* always (native) */ +#define BMOD_1_TO_MOD_1_THRESHOLD 22 + +#define MUL_TOOM22_THRESHOLD 27 +#define MUL_TOOM33_THRESHOLD 38 +#define MUL_TOOM44_THRESHOLD 324 +#define MUL_TOOM6H_THRESHOLD 450 +#define MUL_TOOM8H_THRESHOLD 632 + +#define MUL_TOOM32_TO_TOOM43_THRESHOLD 73 +#define MUL_TOOM32_TO_TOOM53_THRESHOLD 207 +#define MUL_TOOM42_TO_TOOM53_THRESHOLD 211 +#define MUL_TOOM42_TO_TOOM63_THRESHOLD 219 +#define MUL_TOOM43_TO_TOOM54_THRESHOLD 315 + +#define SQR_BASECASE_THRESHOLD 10 +#define SQR_TOOM2_THRESHOLD 52 +#define SQR_TOOM3_THRESHOLD 73 +#define SQR_TOOM4_THRESHOLD 387 +#define SQR_TOOM6_THRESHOLD 662 +#define SQR_TOOM8_THRESHOLD 781 + +#define MULMID_TOOM42_THRESHOLD 32 + +#define MULMOD_BNM1_THRESHOLD 14 +#define SQRMOD_BNM1_THRESHOLD 15 + +#define MUL_FFT_MODF_THRESHOLD 376 /* k = 5 */ +#define MUL_FFT_TABLE3 \ + { { 376, 5}, { 19, 6}, { 10, 5}, { 21, 6}, \ + { 23, 7}, { 12, 6}, { 25, 7}, { 21, 8}, \ + { 11, 7}, { 24, 8}, { 13, 7}, { 27, 8}, \ + { 15, 7}, { 31, 8}, { 19, 7}, { 39, 8}, \ + { 21, 9}, { 11, 8}, { 27, 9}, { 15, 8}, \ + { 35, 9}, { 19, 8}, { 41, 9}, { 23, 8}, \ + { 49, 9}, { 27,10}, { 15, 9}, { 43,10}, \ + { 23, 9}, { 55,11}, { 15,10}, { 31, 9}, \ + { 67,10}, { 39, 9}, { 83,10}, { 47, 9}, \ + { 95,10}, { 79,11}, { 47,10}, { 103,12}, \ + { 31,11}, { 63,10}, { 143,11}, { 79,10}, \ + { 159, 9}, { 319,10}, { 175,11}, { 95, 9}, \ + { 383, 8}, { 767,10}, { 207,11}, { 111,12}, \ + { 63,11}, { 127,10}, { 255,11}, { 143, 9}, \ + { 575, 8}, { 1151,10}, { 303,11}, { 159,10}, \ + { 319, 9}, { 639, 8}, { 1279,10}, { 335,12}, \ + { 95,11}, { 191,10}, { 383, 9}, { 767,11}, \ + { 207,10}, { 415, 9}, { 831, 8}, { 1663,10}, \ + { 447,13}, { 63,12}, { 127,11}, { 255,10}, \ + { 511, 9}, { 1023,11}, { 271,10}, { 543, 9}, \ + { 1087,10}, { 575, 9}, { 1215,12}, { 159,11}, \ + { 319,10}, { 639, 9}, { 1279,11}, { 335,10}, \ + { 671, 9}, { 1343,11}, { 351,10}, { 703, 9}, \ + { 1407,12}, { 191,11}, { 383,10}, { 767, 9}, \ + { 1535,10}, { 831, 9}, { 1663,12}, { 223,11}, \ + { 447,10}, { 895,13}, { 127,12}, { 255,11}, \ + { 511,10}, { 1023,11}, { 543,10}, { 1087,12}, \ + { 287,11}, { 575,10}, { 1151,11}, { 607,10}, \ + { 1215,12}, { 319,11}, { 639,10}, { 1279,11}, \ + { 671,10}, { 1343,12}, { 351,11}, { 703,10}, \ + { 1407,13}, { 191,12}, { 383,11}, { 767,10}, \ + { 1535,12}, { 415,11}, { 831,10}, { 1663,12}, \ + { 447,11}, { 895,10}, { 1791,14}, { 127,13}, \ + { 255,12}, { 511,11}, { 1023,12}, { 543,11}, \ + { 1087,12}, { 575,11}, { 1151,12}, { 607,11}, \ + { 1215,13}, { 319,12}, { 639,11}, { 1279,12}, \ + { 671,11}, { 1343,12}, { 703,11}, { 1407,13}, \ + { 383,12}, { 767,11}, { 1535,12}, { 831,11}, \ + { 1663,13}, { 447,12}, { 895,11}, { 1791,13}, \ + { 511,12}, { 1023,11}, { 2047,12}, { 1087,13}, \ + { 575,12}, { 1151,11}, { 2303,12}, { 1215,13}, \ + { 639,12}, { 1279,11}, { 2559,12}, { 1343,13}, \ + { 703,12}, { 1407,14}, { 383,13}, { 767,12}, \ + { 1535,13}, { 831,12}, { 1663,13}, { 895,12}, \ + { 1791,13}, { 959,14}, { 511,13}, { 1023,12}, \ + { 2047,13}, { 1087,12}, { 2175,13}, { 1151,12}, \ + { 2303,13}, { 1215,14}, { 639,13}, { 1279,12}, \ + { 2559,13}, { 1407,12}, { 2815,13}, { 1471,14}, \ + { 767,13}, { 1535,12}, { 3071,13}, { 1663,14}, \ + { 895,13}, { 1791,12}, { 3583,13}, { 1919,15}, \ + { 511,14}, { 1023,13}, { 2047,12}, { 4095,13}, \ + { 2175,14}, { 1151,13}, { 2303,12}, { 4607,13}, \ + { 2431,14}, { 1279,13}, { 2559,12}, { 5119,14}, \ + { 1407,13}, { 2815,12}, { 5631,15}, { 32768,16}, \ + { 65536,17}, { 131072,18}, { 262144,19}, { 524288,20}, \ + {1048576,21}, {2097152,22}, {4194304,23}, {8388608,24} } +#define MUL_FFT_TABLE3_SIZE 224 +#define MUL_FFT_THRESHOLD 3520 + +#define SQR_FFT_MODF_THRESHOLD 340 /* k = 5 */ +#define SQR_FFT_TABLE3 \ + { { 340, 5}, { 19, 6}, { 10, 5}, { 21, 6}, \ + { 21, 7}, { 11, 6}, { 23, 7}, { 21, 8}, \ + { 11, 7}, { 24, 8}, { 13, 7}, { 27, 8}, \ + { 15, 7}, { 31, 8}, { 21, 9}, { 11, 8}, \ + { 27, 9}, { 15, 8}, { 33, 9}, { 19, 8}, \ + { 41, 9}, { 23, 8}, { 47, 9}, { 27,10}, \ + { 15, 9}, { 39,10}, { 23, 9}, { 51,11}, \ + { 15,10}, { 31, 9}, { 67,10}, { 39, 9}, \ + { 79,10}, { 47, 9}, { 95,10}, { 55,11}, \ + { 31,10}, { 63, 9}, { 127,10}, { 71, 9}, \ + { 143,10}, { 79,11}, { 47,10}, { 95, 9}, \ + { 191,10}, { 103,12}, { 31,11}, { 63,10}, \ + { 127, 9}, { 255,10}, { 135, 7}, { 1087, 9}, \ + { 287,11}, { 79, 9}, { 319, 8}, { 639,10}, \ + { 167,11}, { 95,10}, { 191, 9}, { 383, 8}, \ + { 767,11}, { 111,12}, { 63,11}, { 127,10}, \ + { 255, 9}, { 511, 8}, { 1023,10}, { 271, 9}, \ + { 543, 8}, { 1087,11}, { 143, 9}, { 575, 8}, \ + { 1151,10}, { 303, 9}, { 639, 8}, { 1279,10}, \ + { 335, 9}, { 671,10}, { 351, 9}, { 703,12}, \ + { 95,11}, { 191,10}, { 383, 9}, { 767,11}, \ + { 207,10}, { 415, 9}, { 831,13}, { 63,12}, \ + { 127,11}, { 255,10}, { 511, 9}, { 1023,11}, \ + { 271,10}, { 543, 9}, { 1087,10}, { 575, 9}, \ + { 1151,11}, { 303,10}, { 607, 9}, { 1215,12}, \ + { 159,11}, { 319,10}, { 639, 9}, { 1279,10}, \ + { 671, 9}, { 1343,11}, { 351,10}, { 703, 9}, \ + { 1407,12}, { 191,11}, { 383,10}, { 767, 9}, \ + { 1535,11}, { 415,10}, { 831, 9}, { 1663,12}, \ + { 223,11}, { 447,10}, { 959,13}, { 127,12}, \ + { 255,11}, { 511,10}, { 1023,11}, { 543,10}, \ + { 1087,11}, { 575,10}, { 1215,12}, { 319,11}, \ + { 639,10}, { 1279,11}, { 671,10}, { 1343,12}, \ + { 351,11}, { 703,10}, { 1407,13}, { 191,12}, \ + { 383,11}, { 767,10}, { 1535,12}, { 415,11}, \ + { 831,10}, { 1663,12}, { 447,11}, { 895,10}, \ + { 1791,12}, { 479,11}, { 959,14}, { 127,12}, \ + { 511,11}, { 1023,12}, { 543,11}, { 1087,12}, \ + { 575,11}, { 1151,12}, { 607,11}, { 1215,13}, \ + { 319,12}, { 639,11}, { 1279,12}, { 671,11}, \ + { 1343,12}, { 703,11}, { 1407,13}, { 383,12}, \ + { 767,11}, { 1535,12}, { 831,11}, { 1663,13}, \ + { 447,12}, { 895,11}, { 1791,12}, { 959,13}, \ + { 511,12}, { 1023,11}, { 2047,12}, { 1087,13}, \ + { 575,12}, { 1215,13}, { 639,12}, { 1343,13}, \ + { 703,12}, { 1407,11}, { 2815,13}, { 767,12}, \ + { 1535,13}, { 831,12}, { 1663,13}, { 895,12}, \ + { 1791,13}, { 959,14}, { 511,13}, { 1023,12}, \ + { 2047,13}, { 1087,12}, { 2175,13}, { 1215,14}, \ + { 639,13}, { 1279,12}, { 2559,13}, { 1407,12}, \ + { 2815,14}, { 767,13}, { 1535,12}, { 3071,13}, \ + { 1663,14}, { 895,13}, { 1791,12}, { 3583,13}, \ + { 1919,15}, { 511,14}, { 1023,13}, { 2047,12}, \ + { 4095,13}, { 2175,14}, { 1151,13}, { 2303,12}, \ + { 4607,14}, { 1279,13}, { 2559,14}, { 1407,13}, \ + { 2815,15}, { 32768,16}, { 65536,17}, { 131072,18}, \ + { 262144,19}, { 524288,20}, {1048576,21}, {2097152,22}, \ + {4194304,23}, {8388608,24} } +#define SQR_FFT_TABLE3_SIZE 230 +#define SQR_FFT_THRESHOLD 2496 + +#define MULLO_BASECASE_THRESHOLD 13 +#define MULLO_DC_THRESHOLD 38 +#define MULLO_MUL_N_THRESHOLD 6633 + +#define DC_DIV_QR_THRESHOLD 56 +#define DC_DIVAPPR_Q_THRESHOLD 173 +#define DC_BDIV_QR_THRESHOLD 55 +#define DC_BDIV_Q_THRESHOLD 96 + +#define INV_MULMOD_BNM1_THRESHOLD 54 +#define INV_NEWTON_THRESHOLD 202 +#define INV_APPR_THRESHOLD 166 + +#define BINV_NEWTON_THRESHOLD 246 +#define REDC_1_TO_REDC_2_THRESHOLD 7 +#define REDC_2_TO_REDC_N_THRESHOLD 85 + +#define MU_DIV_QR_THRESHOLD 1499 +#define MU_DIVAPPR_Q_THRESHOLD 1652 +#define MUPI_DIV_QR_THRESHOLD 83 +#define MU_BDIV_QR_THRESHOLD 1210 +#define MU_BDIV_Q_THRESHOLD 1499 + +#define POWM_SEC_TABLE 1,28,129,642,2387 + +#define MATRIX22_STRASSEN_THRESHOLD 15 +#define HGCD_THRESHOLD 127 +#define HGCD_APPR_THRESHOLD 214 +#define HGCD_REDUCE_THRESHOLD 2479 +#define GCD_DC_THRESHOLD 487 +#define GCDEXT_DC_THRESHOLD 505 +#define JACOBI_BASE_METHOD 4 + +#define GET_STR_DC_THRESHOLD 12 +#define GET_STR_PRECOMPUTE_THRESHOLD 24 +#define SET_STR_DC_THRESHOLD 802 +#define SET_STR_PRECOMPUTE_THRESHOLD 2042 + +#define FAC_DSC_THRESHOLD 1737 +#define FAC_ODD_THRESHOLD 44 diff --git a/gmp-6.3.0/mpn/x86_64/nano/popcount.asm b/gmp-6.3.0/mpn/x86_64/nano/popcount.asm new file mode 100644 index 0000000..fb14dd3 --- /dev/null +++ b/gmp-6.3.0/mpn/x86_64/nano/popcount.asm @@ -0,0 +1,35 @@ +dnl x86-64 mpn_popcount. + +dnl Copyright 2007, 2011 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. +dnl +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of either: +dnl +dnl * the GNU Lesser General Public License as published by the Free +dnl Software Foundation; either version 3 of the License, or (at your +dnl option) any later version. +dnl +dnl or +dnl +dnl * the GNU General Public License as published by the Free Software +dnl Foundation; either version 2 of the License, or (at your option) any +dnl later version. +dnl +dnl or both in parallel, as here. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, but +dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +dnl for more details. +dnl +dnl You should have received copies of the GNU General Public License and the +dnl GNU Lesser General Public License along with the GNU MP Library. If not, +dnl see https://www.gnu.org/licenses/. + + +include(`../config.m4') + +MULFUNC_PROLOGUE(mpn_popcount) +include_mpn(`x86/pentium4/sse2/popcount.asm') -- cgit v1.2.3