copyi.asm revision 1.1.1.3 1 dnl AMD64 mpn_copyi -- copy limb vector, incrementing.
2
3 dnl Copyright 2003, 2005, 2007, 2011, 2012 Free Software Foundation, Inc.
4
5 dnl This file is part of the GNU MP Library.
6 dnl
7 dnl The GNU MP Library is free software; you can redistribute it and/or modify
8 dnl it under the terms of either:
9 dnl
10 dnl * the GNU Lesser General Public License as published by the Free
11 dnl Software Foundation; either version 3 of the License, or (at your
12 dnl option) any later version.
13 dnl
14 dnl or
15 dnl
16 dnl * the GNU General Public License as published by the Free Software
17 dnl Foundation; either version 2 of the License, or (at your option) any
18 dnl later version.
19 dnl
20 dnl or both in parallel, as here.
21 dnl
22 dnl The GNU MP Library is distributed in the hope that it will be useful, but
23 dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
24 dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
25 dnl for more details.
26 dnl
27 dnl You should have received copies of the GNU General Public License and the
28 dnl GNU Lesser General Public License along with the GNU MP Library. If not,
29 dnl see https://www.gnu.org/licenses/.
30
31 include(`../config.m4')
32
33 C cycles/limb
34 C AMD K8,K9 1
35 C AMD K10 1
36 C AMD bd1 1.36
37 C AMD bobcat 1.71
38 C Intel P4 2-3
39 C Intel core2 1
40 C Intel NHM 1
41 C Intel SBR 1
42 C Intel atom 2
43 C VIA nano 2
44
45
46 IFSTD(`define(`rp',`%rdi')')
47 IFSTD(`define(`up',`%rsi')')
48 IFSTD(`define(`n', `%rdx')')
49
50 IFDOS(`define(`rp',`%rcx')')
51 IFDOS(`define(`up',`%rdx')')
52 IFDOS(`define(`n', `%r8')')
53
54 ABI_SUPPORT(DOS64)
55 ABI_SUPPORT(STD64)
56
57 ASM_START()
58 TEXT
59 ALIGN(64)
60 .byte 0,0,0,0,0,0
61 PROLOGUE(mpn_copyi)
62 lea -8(rp), rp
63 sub $4, n
64 jc L(end)
65
66 L(top): mov (up), %rax
67 mov 8(up), %r9
68 lea 32(rp), rp
69 mov 16(up), %r10
70 mov 24(up), %r11
71 lea 32(up), up
72 mov %rax, -24(rp)
73 mov %r9, -16(rp)
74 sub $4, n
75 mov %r10, -8(rp)
76 mov %r11, (rp)
77 jnc L(top)
78
79 L(end): shr R32(n)
80 jnc 1f
81 mov (up), %rax
82 mov %rax, 8(rp)
83 lea 8(rp), rp
84 lea 8(up), up
85 1: shr R32(n)
86 jnc 1f
87 mov (up), %rax
88 mov 8(up), %r9
89 mov %rax, 8(rp)
90 mov %r9, 16(rp)
91 1: ret
92 EPILOGUE()
93