1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
|
dnl AMD K7 mpn_modexact_1_odd -- exact division style remainder.
dnl Copyright 2000-2002, 2004, 2007 Free Software Foundation, Inc.
dnl This file is part of the GNU MP Library.
dnl
dnl The GNU MP Library is free software; you can redistribute it and/or modify
dnl it under the terms of either:
dnl
dnl * the GNU Lesser General Public License as published by the Free
dnl Software Foundation; either version 3 of the License, or (at your
dnl option) any later version.
dnl
dnl or
dnl
dnl * the GNU General Public License as published by the Free Software
dnl Foundation; either version 2 of the License, or (at your option) any
dnl later version.
dnl
dnl or both in parallel, as here.
dnl
dnl The GNU MP Library is distributed in the hope that it will be useful, but
dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
dnl for more details.
dnl
dnl You should have received copies of the GNU General Public License and the
dnl GNU Lesser General Public License along with the GNU MP Library. If not,
dnl see https://www.gnu.org/licenses/.
include(`../config.m4')
C cycles/limb
C Athlon: 11.0
C Hammer: 7.0
C mp_limb_t mpn_modexact_1_odd (mp_srcptr src, mp_size_t size,
C mp_limb_t divisor);
C mp_limb_t mpn_modexact_1c_odd (mp_srcptr src, mp_size_t size,
C mp_limb_t divisor, mp_limb_t carry);
C
C With the loop running at just 11 cycles it doesn't seem worth bothering to
C check for high<divisor to save one step.
C
C Using a divl for size==1 measures slower than the modexact method, which
C is not too surprising since for the latter it's only about 24 cycles to
C calculate the modular inverse.
defframe(PARAM_CARRY, 16)
defframe(PARAM_DIVISOR,12)
defframe(PARAM_SIZE, 8)
defframe(PARAM_SRC, 4)
defframe(SAVE_EBX, -4)
defframe(SAVE_ESI, -8)
defframe(SAVE_EDI, -12)
defframe(SAVE_EBP, -16)
deflit(STACK_SPACE, 16)
TEXT
ALIGN(16)
PROLOGUE(mpn_modexact_1c_odd)
deflit(`FRAME',0)
movl PARAM_CARRY, %ecx
jmp L(start_1c)
EPILOGUE()
ALIGN(16)
PROLOGUE(mpn_modexact_1_odd)
deflit(`FRAME',0)
xorl %ecx, %ecx
L(start_1c):
movl PARAM_DIVISOR, %eax
subl $STACK_SPACE, %esp FRAME_subl_esp(STACK_SPACE)
movl %esi, SAVE_ESI
movl PARAM_DIVISOR, %esi
movl %edi, SAVE_EDI
shrl %eax C d/2
andl $127, %eax
ifdef(`PIC',`
LEA( binvert_limb_table, %edi)
movzbl (%eax,%edi), %edi C inv 8 bits
',`
movzbl binvert_limb_table(%eax), %edi C inv 8 bits
')
xorl %edx, %edx C initial extra carry
leal (%edi,%edi), %eax C 2*inv
imull %edi, %edi C inv*inv
movl %ebp, SAVE_EBP
movl PARAM_SIZE, %ebp
movl %ebx, SAVE_EBX
movl PARAM_SRC, %ebx
imull %esi, %edi C inv*inv*d
subl %edi, %eax C inv = 2*inv - inv*inv*d
leal (%eax,%eax), %edi C 2*inv
imull %eax, %eax C inv*inv
imull %esi, %eax C inv*inv*d
leal (%ebx,%ebp,4), %ebx C src end
negl %ebp C -size
subl %eax, %edi C inv = 2*inv - inv*inv*d
ASSERT(e,` C d*inv == 1 mod 2^GMP_LIMB_BITS
movl %esi, %eax
imull %edi, %eax
cmpl $1, %eax')
C The dependent chain here is
C
C cycles
C subl %edx, %eax 1
C imull %edi, %eax 4
C mull %esi 6 (high limb)
C ----
C total 11
C
C Out of order execution hides the load latency for the source data, so no
C special scheduling is required.
L(top):
C eax src limb
C ebx src end ptr
C ecx next carry bit, 0 or 1 (or initial carry param)
C edx carry limb, high of last product
C esi divisor
C edi inverse
C ebp counter, limbs, negative
movl (%ebx,%ebp,4), %eax
subl %ecx, %eax C apply carry bit
movl $0, %ecx
setc %cl C new carry bit
subl %edx, %eax C apply carry limb
adcl $0, %ecx
imull %edi, %eax
mull %esi
incl %ebp
jnz L(top)
movl SAVE_ESI, %esi
movl SAVE_EDI, %edi
leal (%ecx,%edx), %eax
movl SAVE_EBX, %ebx
movl SAVE_EBP, %ebp
addl $STACK_SPACE, %esp
ret
EPILOGUE()
ASM_END()
|
