1dnl ARM64 mpn_bdiv_dbm1c. 2 3dnl Copyright 2008, 2011, 2012, 2014 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 Cortex-A53 ? 35C Cortex-A57 ? 36 37define(`qp', `x0') 38define(`up', `x1') 39define(`n', `x2') 40define(`bd', `x3') 41define(`cy', `x4') 42 43ASM_START() 44 TEXT 45 ALIGN(16) 46PROLOGUE(mpn_bdiv_dbm1c) 47 ldr x5, [up], #8 48 ands x6, n, #3 49 b.eq L(fi0) 50 cmp x6, #2 51 b.cc L(fi1) 52 b.eq L(fi2) 53 54L(fi3): mul x12, x5, bd 55 umulh x13, x5, bd 56 ldr x5, [up], #8 57 b L(lo3) 58 59L(fi0): mul x10, x5, bd 60 umulh x11, x5, bd 61 ldr x5, [up], #8 62 b L(lo0) 63 64L(fi1): subs n, n, #1 65 mul x12, x5, bd 66 umulh x13, x5, bd 67 b.ls L(wd1) 68 ldr x5, [up], #8 69 b L(lo1) 70 71L(fi2): mul x10, x5, bd 72 umulh x11, x5, bd 73 ldr x5, [up], #8 74 b L(lo2) 75 76L(top): ldr x5, [up], #8 77 subs x4, x4, x10 78 str x4, [qp], #8 79 sbc x4, x4, x11 80L(lo1): mul x10, x5, bd 81 umulh x11, x5, bd 82 ldr x5, [up], #8 83 subs x4, x4, x12 84 str x4, [qp], #8 85 sbc x4, x4, x13 86L(lo0): mul x12, x5, bd 87 umulh x13, x5, bd 88 ldr x5, [up], #8 89 subs x4, x4, x10 90 str x4, [qp], #8 91 sbc x4, x4, x11 92L(lo3): mul x10, x5, bd 93 umulh x11, x5, bd 94 ldr x5, [up], #8 95 subs x4, x4, x12 96 str x4, [qp], #8 97 sbc x4, x4, x13 98L(lo2): subs n, n, #4 99 mul x12, x5, bd 100 umulh x13, x5, bd 101 b.hi L(top) 102 103L(wd2): subs x4, x4, x10 104 str x4, [qp], #8 105 sbc x4, x4, x11 106L(wd1): subs x4, x4, x12 107 str x4, [qp] 108 sbc x0, x4, x13 109 ret 110EPILOGUE() 111