| 1 |
dnl ARM64 mpn_addmul_1 and mpn_submul_1. |
| 2 |
|
| 3 |
dnl Contributed to the GNU project by Torbjörn Granlund. |
| 4 |
|
| 5 |
dnl Copyright 2020 Free Software Foundation, Inc. |
| 6 |
|
| 7 |
dnl This file is part of the GNU MP Library. |
| 8 |
dnl |
| 9 |
dnl The GNU MP Library is free software; you can redistribute it and/or modify |
| 10 |
dnl it under the terms of either: |
| 11 |
dnl |
| 12 |
dnl * the GNU Lesser General Public License as published by the Free |
| 13 |
dnl Software Foundation; either version 3 of the License, or (at your |
| 14 |
dnl option) any later version. |
| 15 |
dnl |
| 16 |
dnl or |
| 17 |
dnl |
| 18 |
dnl * the GNU General Public License as published by the Free Software |
| 19 |
dnl Foundation; either version 2 of the License, or (at your option) any |
| 20 |
dnl later version. |
| 21 |
dnl |
| 22 |
dnl or both in parallel, as here. |
| 23 |
dnl |
| 24 |
dnl The GNU MP Library is distributed in the hope that it will be useful, but |
| 25 |
dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY |
| 26 |
dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License |
| 27 |
dnl for more details. |
| 28 |
dnl |
| 29 |
dnl You should have received copies of the GNU General Public License and the |
| 30 |
dnl GNU Lesser General Public License along with the GNU MP Library. If not, |
| 31 |
dnl see https://www.gnu.org/licenses/. |
| 32 |
|
| 33 |
include(`../config.m4') |
| 34 |
|
| 35 |
C cycles/limb |
| 36 |
C Cortex-A53 |
| 37 |
C Cortex-A55 |
| 38 |
C Cortex-A57 |
| 39 |
C Cortex-A72 |
| 40 |
C Cortex-A73 |
| 41 |
C X-Gene |
| 42 |
C Apple M1 1.25 |
| 43 |
|
| 44 |
changecom(blah) |
| 45 |
|
| 46 |
define(`rp', x0) |
| 47 |
define(`up', x1) |
| 48 |
define(`n', x2) |
| 49 |
define(`v0', x3) |
| 50 |
define(`cin',x4) |
| 51 |
|
| 52 |
define(`CY',x17) |
| 53 |
|
| 54 |
ifdef(`OPERATION_addmul_1', ` |
| 55 |
define(`ADDSUB', adds) |
| 56 |
define(`ADDSUBC', adcs) |
| 57 |
define(`COND', `cc') |
| 58 |
define(`func', mpn_addmul_1)') |
| 59 |
ifdef(`OPERATION_submul_1', ` |
| 60 |
define(`ADDSUB', subs) |
| 61 |
define(`ADDSUBC', sbcs) |
| 62 |
define(`COND', `cs') |
| 63 |
define(`func', mpn_submul_1)') |
| 64 |
|
| 65 |
MULFUNC_PROLOGUE(mpn_addmul_1 mpn_submul_1 mpn_addmul_1c) |
| 66 |
|
| 67 |
ifdef(`OPERATION_addmul_1', ` |
| 68 |
PROLOGUE(mpn_addmul_1c) |
| 69 |
mov CY, cin |
| 70 |
b L(ent) |
| 71 |
EPILOGUE() |
| 72 |
') |
| 73 |
|
| 74 |
PROLOGUE(func) |
| 75 |
mov CY, #0 C W0 |
| 76 |
L(ent): lsr x16, n, #2 |
| 77 |
tbz n, #0, L(bx0) |
| 78 |
|
| 79 |
L(bx1): ldr x4, [up], #8 |
| 80 |
mul x8, x4, v0 |
| 81 |
umulh x4, x4, v0 |
| 82 |
tbz n, #1, L(b01) |
| 83 |
|
| 84 |
L(b11): ldp x5,x6, [up], #16 |
| 85 |
ldp x12,x13, [rp] |
| 86 |
ldr x14, [rp,#16] |
| 87 |
mul x9, x5, v0 |
| 88 |
umulh x5, x5, v0 |
| 89 |
mul x10, x6, v0 |
| 90 |
umulh x6, x6, v0 |
| 91 |
ADDSUB x8, x12, x8 |
| 92 |
ADDSUBC x4, x13, x4 |
| 93 |
ADDSUBC x5, x14, x5 |
| 94 |
csinc x6, x6, x6, COND |
| 95 |
ADDSUB x8, x8, CY |
| 96 |
ADDSUBC x4, x4, x9 |
| 97 |
ADDSUBC x5, x5, x10 |
| 98 |
csinc CY, x6, x6, COND |
| 99 |
stp x8, x4, [rp], #16 |
| 100 |
str x5, [rp], #8 |
| 101 |
cbnz x16, L(top) |
| 102 |
mov x0, CY |
| 103 |
ret |
| 104 |
|
| 105 |
L(b01): ldr x12, [rp] |
| 106 |
ADDSUB x8, x12, x8 |
| 107 |
csinc x4, x4, x4, COND |
| 108 |
ADDSUB x8, x8, CY |
| 109 |
csinc CY, x4, x4, COND |
| 110 |
str x8, [rp], #8 |
| 111 |
cbnz x16, L(top) |
| 112 |
mov x0, CY |
| 113 |
ret |
| 114 |
|
| 115 |
L(bx0): ldp x4,x5, [up], #16 |
| 116 |
tbz n, #1, L(top)+4 |
| 117 |
|
| 118 |
L(b10): ldp x12,x13, [rp] |
| 119 |
mul x8, x4, v0 |
| 120 |
umulh x4, x4, v0 |
| 121 |
mul x9, x5, v0 |
| 122 |
umulh x5, x5, v0 |
| 123 |
ADDSUB x8, x12, x8 |
| 124 |
ADDSUBC x4, x13, x4 |
| 125 |
csinc x5, x5, x5, COND |
| 126 |
ADDSUB x8, x8, CY |
| 127 |
ADDSUBC x4, x4, x9 |
| 128 |
csinc CY, x5, x5, COND |
| 129 |
stp x8, x4, [rp], #16 |
| 130 |
cbz x16, L(done) |
| 131 |
|
| 132 |
L(top): ldp x4,x5, [up], #16 C W0 W1 |
| 133 |
ldp x6,x7, [up], #16 C W2 W3 |
| 134 |
ldp x12,x13, [rp] C W0 W1 |
| 135 |
ldp x14,x15, [rp,#16] C W2 W3 |
| 136 |
mul x8, x4, v0 C W0 |
| 137 |
umulh x4, x4, v0 C W1 |
| 138 |
mul x9, x5, v0 C W1 |
| 139 |
umulh x5, x5, v0 C W2 |
| 140 |
mul x10, x6, v0 C W2 |
| 141 |
umulh x6, x6, v0 C W3 |
| 142 |
mul x11, x7, v0 C W3 |
| 143 |
umulh x7, x7, v0 C W4 |
| 144 |
ADDSUB x8, x12, x8 C W0 |
| 145 |
ADDSUBC x4, x13, x4 C W1 |
| 146 |
ADDSUBC x5, x14, x5 C W2 |
| 147 |
ADDSUBC x6, x15, x6 C W3 |
| 148 |
csinc x7, x7, x7, COND C W4 |
| 149 |
ADDSUB x8, x8, CY C W0 carry-in |
| 150 |
ADDSUBC x4, x4, x9 C W1 |
| 151 |
ADDSUBC x5, x5, x10 C W2 |
| 152 |
ADDSUBC x6, x6, x11 C W2 |
| 153 |
csinc CY, x7, x7, COND C W3 carry-out |
| 154 |
stp x8, x4, [rp], #16 |
| 155 |
stp x5, x6, [rp], #16 |
| 156 |
sub x16, x16, #1 |
| 157 |
cbnz x16, L(top) |
| 158 |
|
| 159 |
L(done):mov x0, CY |
| 160 |
ret |
| 161 |
EPILOGUE() |