| 1 |
dnl Core 2 mpn_add_err1_n, mpn_sub_err1_n |
| 2 |
|
| 3 |
dnl Contributed by David Harvey. |
| 4 |
|
| 5 |
dnl Copyright 2011 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 AMD K8,K9 ? |
| 37 |
C AMD K10 ? |
| 38 |
C Intel P4 ? |
| 39 |
C Intel core2 4.14 |
| 40 |
C Intel corei ? |
| 41 |
C Intel atom ? |
| 42 |
C VIA nano ? |
| 43 |
|
| 44 |
|
| 45 |
C INPUT PARAMETERS |
| 46 |
define(`rp', `%rdi') |
| 47 |
define(`up', `%rsi') |
| 48 |
define(`vp', `%rdx') |
| 49 |
define(`ep', `%rcx') |
| 50 |
define(`yp', `%r8') |
| 51 |
define(`n', `%r9') |
| 52 |
define(`cy_param', `8(%rsp)') |
| 53 |
|
| 54 |
define(`el', `%rbx') |
| 55 |
define(`eh', `%rbp') |
| 56 |
define(`t0', `%r10') |
| 57 |
define(`t1', `%r11') |
| 58 |
define(`t2', `%r12') |
| 59 |
define(`t3', `%r13') |
| 60 |
define(`w0', `%r14') |
| 61 |
define(`w1', `%r15') |
| 62 |
|
| 63 |
ifdef(`OPERATION_add_err1_n', ` |
| 64 |
define(ADCSBB, adc) |
| 65 |
define(func, mpn_add_err1_n)') |
| 66 |
ifdef(`OPERATION_sub_err1_n', ` |
| 67 |
define(ADCSBB, sbb) |
| 68 |
define(func, mpn_sub_err1_n)') |
| 69 |
|
| 70 |
MULFUNC_PROLOGUE(mpn_add_err1_n mpn_sub_err1_n) |
| 71 |
|
| 72 |
|
| 73 |
ASM_START() |
| 74 |
TEXT |
| 75 |
ALIGN(16) |
| 76 |
PROLOGUE(func) |
| 77 |
mov cy_param, %rax |
| 78 |
|
| 79 |
push %rbx |
| 80 |
push %rbp |
| 81 |
push %r12 |
| 82 |
push %r13 |
| 83 |
push %r14 |
| 84 |
push %r15 |
| 85 |
|
| 86 |
lea (up,n,8), up |
| 87 |
lea (vp,n,8), vp |
| 88 |
lea (rp,n,8), rp |
| 89 |
|
| 90 |
mov R32(n), R32(%r10) |
| 91 |
and $3, R32(%r10) |
| 92 |
jz L(0mod4) |
| 93 |
cmp $2, R32(%r10) |
| 94 |
jc L(1mod4) |
| 95 |
jz L(2mod4) |
| 96 |
L(3mod4): |
| 97 |
xor R32(el), R32(el) |
| 98 |
xor R32(eh), R32(eh) |
| 99 |
xor R32(t0), R32(t0) |
| 100 |
xor R32(t1), R32(t1) |
| 101 |
lea -24(yp,n,8), yp |
| 102 |
neg n |
| 103 |
|
| 104 |
shr $1, %al C restore carry |
| 105 |
mov (up,n,8), w0 |
| 106 |
mov 8(up,n,8), w1 |
| 107 |
ADCSBB (vp,n,8), w0 |
| 108 |
mov w0, (rp,n,8) |
| 109 |
cmovc 16(yp), el |
| 110 |
ADCSBB 8(vp,n,8), w1 |
| 111 |
mov w1, 8(rp,n,8) |
| 112 |
cmovc 8(yp), t0 |
| 113 |
mov 16(up,n,8), w0 |
| 114 |
ADCSBB 16(vp,n,8), w0 |
| 115 |
mov w0, 16(rp,n,8) |
| 116 |
cmovc (yp), t1 |
| 117 |
setc %al C save carry |
| 118 |
add t0, el |
| 119 |
adc $0, eh |
| 120 |
add t1, el |
| 121 |
adc $0, eh |
| 122 |
|
| 123 |
add $3, n |
| 124 |
jnz L(loop) |
| 125 |
jmp L(end) |
| 126 |
|
| 127 |
ALIGN(16) |
| 128 |
L(0mod4): |
| 129 |
xor R32(el), R32(el) |
| 130 |
xor R32(eh), R32(eh) |
| 131 |
lea (yp,n,8), yp |
| 132 |
neg n |
| 133 |
jmp L(loop) |
| 134 |
|
| 135 |
ALIGN(16) |
| 136 |
L(1mod4): |
| 137 |
xor R32(el), R32(el) |
| 138 |
xor R32(eh), R32(eh) |
| 139 |
lea -8(yp,n,8), yp |
| 140 |
neg n |
| 141 |
|
| 142 |
shr $1, %al C restore carry |
| 143 |
mov (up,n,8), w0 |
| 144 |
ADCSBB (vp,n,8), w0 |
| 145 |
mov w0, (rp,n,8) |
| 146 |
cmovc (yp), el |
| 147 |
setc %al C save carry |
| 148 |
|
| 149 |
add $1, n |
| 150 |
jnz L(loop) |
| 151 |
jmp L(end) |
| 152 |
|
| 153 |
ALIGN(16) |
| 154 |
L(2mod4): |
| 155 |
xor R32(el), R32(el) |
| 156 |
xor R32(eh), R32(eh) |
| 157 |
xor R32(t0), R32(t0) |
| 158 |
lea -16(yp,n,8), yp |
| 159 |
neg n |
| 160 |
|
| 161 |
shr $1, %al C restore carry |
| 162 |
mov (up,n,8), w0 |
| 163 |
mov 8(up,n,8), w1 |
| 164 |
ADCSBB (vp,n,8), w0 |
| 165 |
mov w0, (rp,n,8) |
| 166 |
cmovc 8(yp), el |
| 167 |
ADCSBB 8(vp,n,8), w1 |
| 168 |
mov w1, 8(rp,n,8) |
| 169 |
cmovc (yp), t0 |
| 170 |
setc %al C save carry |
| 171 |
add t0, el |
| 172 |
adc $0, eh |
| 173 |
|
| 174 |
add $2, n |
| 175 |
jnz L(loop) |
| 176 |
jmp L(end) |
| 177 |
|
| 178 |
ALIGN(32) |
| 179 |
L(loop): |
| 180 |
mov (up,n,8), w0 |
| 181 |
shr $1, %al C restore carry |
| 182 |
mov -8(yp), t0 |
| 183 |
mov $0, R32(t3) |
| 184 |
ADCSBB (vp,n,8), w0 |
| 185 |
cmovnc t3, t0 |
| 186 |
mov w0, (rp,n,8) |
| 187 |
mov 8(up,n,8), w1 |
| 188 |
mov 16(up,n,8), w0 |
| 189 |
ADCSBB 8(vp,n,8), w1 |
| 190 |
mov -16(yp), t1 |
| 191 |
cmovnc t3, t1 |
| 192 |
mov -24(yp), t2 |
| 193 |
mov w1, 8(rp,n,8) |
| 194 |
ADCSBB 16(vp,n,8), w0 |
| 195 |
cmovnc t3, t2 |
| 196 |
mov 24(up,n,8), w1 |
| 197 |
ADCSBB 24(vp,n,8), w1 |
| 198 |
cmovc -32(yp), t3 |
| 199 |
setc %al C save carry |
| 200 |
add t0, el |
| 201 |
adc $0, eh |
| 202 |
add t1, el |
| 203 |
adc $0, eh |
| 204 |
add t2, el |
| 205 |
adc $0, eh |
| 206 |
lea -32(yp), yp |
| 207 |
mov w0, 16(rp,n,8) |
| 208 |
add t3, el |
| 209 |
adc $0, eh |
| 210 |
add $4, n |
| 211 |
mov w1, -8(rp,n,8) |
| 212 |
jnz L(loop) |
| 213 |
|
| 214 |
L(end): |
| 215 |
mov el, (ep) |
| 216 |
mov eh, 8(ep) |
| 217 |
|
| 218 |
pop %r15 |
| 219 |
pop %r14 |
| 220 |
pop %r13 |
| 221 |
pop %r12 |
| 222 |
pop %rbp |
| 223 |
pop %rbx |
| 224 |
ret |
| 225 |
EPILOGUE() |