1dnl x86_64 mpn_bdiv_dbm1. 2 3dnl Copyright 2008, 2011, 2012 Free Software Foundation, Inc. 4 5dnl This file is part of the GNU MP Library. 6dnl 7dnl The GNU MP Library is free software; you can redistribute it and/or modify 8dnl it under the terms of either: 9dnl 10dnl * the GNU Lesser General Public License as published by the Free 11dnl Software Foundation; either version 3 of the License, or (at your 12dnl option) any later version. 13dnl 14dnl or 15dnl 16dnl * the GNU General Public License as published by the Free Software 17dnl Foundation; either version 2 of the License, or (at your option) any 18dnl later version. 19dnl 20dnl or both in parallel, as here. 21dnl 22dnl The GNU MP Library is distributed in the hope that it will be useful, but 23dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY 24dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License 25dnl for more details. 26dnl 27dnl You should have received copies of the GNU General Public License and the 28dnl GNU Lesser General Public License along with the GNU MP Library. If not, 29dnl see https://www.gnu.org/licenses/. 30 31include(`../config.m4') 32 33C cycles/limb 34C AMD K8,K9 2.25 35C AMD K10 2.25 36C Intel P4 12.5 37C Intel core2 4 38C Intel NHM 3.75 39C Intel SBR 3.6 40C Intel atom 20 41C VIA nano 4 42 43C TODO 44C * Optimise feed-in code. 45 46C INPUT PARAMETERS 47define(`qp', `%rdi') 48define(`up', `%rsi') 49define(`n_param', `%rdx') 50define(`bd', `%rcx') 51define(`cy', `%r8') 52 53define(`n', `%r9') 54 55ABI_SUPPORT(DOS64) 56ABI_SUPPORT(STD64) 57 58ASM_START() 59 TEXT 60 ALIGN(16) 61PROLOGUE(mpn_bdiv_dbm1c) 62 FUNC_ENTRY(4) 63IFDOS(` mov 56(%rsp), %r8 ') 64 mov (up), %rax 65 mov n_param, n 66 mov R32(n_param), R32(%r11) 67 mul bd 68 lea (up,n,8), up 69 lea (qp,n,8), qp 70 neg n 71 and $3, R32(%r11) 72 jz L(lo0) 73 lea -4(n,%r11), n 74 cmp $2, R32(%r11) 75 jc L(lo1) 76 jz L(lo2) 77 jmp L(lo3) 78 79 ALIGN(16) 80L(top): mov (up,n,8), %rax 81 mul bd 82L(lo0): sub %rax, %r8 83 mov %r8, (qp,n,8) 84 sbb %rdx, %r8 85 mov 8(up,n,8), %rax 86 mul bd 87L(lo3): sub %rax, %r8 88 mov %r8, 8(qp,n,8) 89 sbb %rdx, %r8 90 mov 16(up,n,8), %rax 91 mul bd 92L(lo2): sub %rax, %r8 93 mov %r8, 16(qp,n,8) 94 sbb %rdx, %r8 95 mov 24(up,n,8), %rax 96 mul bd 97L(lo1): sub %rax, %r8 98 mov %r8, 24(qp,n,8) 99 sbb %rdx, %r8 100 add $4, n 101 jnz L(top) 102 103 mov %r8, %rax 104 FUNC_EXIT() 105 ret 106EPILOGUE() 107