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
|
dnl AMD K7 mpn_sublsh1_n_ip1 -- rp[] = rp[] - (up[] << 1)
dnl Copyright 2011 Free Software Foundation, Inc.
dnl Contributed to the GNU project by Torbjorn Granlund and Marco Bodrato.
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 This is an attempt at a sublsh1_n for x86-32, not relying on sse2 insns. The
C innerloop is 2*3-way unrolled, which is best we can do with the available
C registers. It seems tricky to use the same structure for rsblsh1_n, since we
C cannot feed carry between operations there.
C cycles/limb
C P5
C P6 model 0-8,10-12
C P6 model 9 (Banias)
C P6 model 13 (Dothan)
C P4 model 0 (Willamette)
C P4 model 1 (?)
C P4 model 2 (Northwood)
C P4 model 3 (Prescott)
C P4 model 4 (Nocona)
C Intel Atom 6.75
C AMD K6
C AMD K7
C AMD K8
C This is a basic sublsh1_n for k7, atom, and perhaps some other x86-32
C processors. It uses 2*4-way unrolling, for good reasons.
C
C Breaking carry recurrency might be a good idea. We would then need separate
C registers for the shift carry and add/subtract carry, which in turn would
C force us to 2*2-way unrolling.
defframe(PARAM_SIZE, 12)
defframe(PARAM_SRC, 8)
defframe(PARAM_DST, 4)
dnl re-use parameter space
define(VAR_COUNT,`PARAM_SIZE')
define(SAVE_EBX,`PARAM_SRC')
define(SAVE_EBP,`PARAM_DST')
ASM_START()
TEXT
ALIGN(8)
PROLOGUE(mpn_sublsh1_n_ip1)
deflit(`FRAME',0)
define(`rp', `%edi')
define(`up', `%esi')
mov PARAM_SIZE, %eax C size
push up FRAME_pushl()
push rp FRAME_pushl()
xor %edx, %edx
mov PARAM_SRC, up
mov PARAM_DST, rp
mov %ebx, SAVE_EBX
mov %eax, %ebx
shr $3, %eax
not %eax C count = -(size\8)-i
and $7, %ebx C size % 8
jz L(exact)
L(oop):
ifdef(`CPU_P6',`
shr %edx ') C restore 2nd saved carry bit
mov (up), %ecx
adc %ecx, %ecx
rcr %edx C restore 1st saved carry bit
lea 4(up), up
sbb %ecx, (rp)
lea 4(rp), rp
adc %edx, %edx C save a carry bit in edx
ifdef(`CPU_P6',`
adc %edx, %edx ') C save another carry bit in edx
dec %ebx
jnz L(oop)
L(exact):
inc %eax
jz L(end)
mov %eax, VAR_COUNT
mov %ebp, SAVE_EBP
ALIGN(16)
L(top):
ifdef(`CPU_P6',`
shr %edx ') C restore 2nd saved carry bit
mov (up), %eax
adc %eax, %eax
mov 4(up), %ebx
adc %ebx, %ebx
mov 8(up), %ecx
adc %ecx, %ecx
mov 12(up), %ebp
adc %ebp, %ebp
rcr %edx C restore 1st saved carry bit
sbb %eax, (rp)
sbb %ebx, 4(rp)
sbb %ecx, 8(rp)
sbb %ebp, 12(rp)
mov 16(up), %eax
adc %eax, %eax
mov 20(up), %ebx
adc %ebx, %ebx
mov 24(up), %ecx
adc %ecx, %ecx
mov 28(up), %ebp
adc %ebp, %ebp
lea 32(up), up
adc %edx, %edx C save a carry bit in edx
sbb %eax, 16(rp)
sbb %ebx, 20(rp)
sbb %ecx, 24(rp)
sbb %ebp, 28(rp)
ifdef(`CPU_P6',`
adc %edx, %edx ') C save another carry bit in edx
incl VAR_COUNT
lea 32(rp), rp
jne L(top)
mov SAVE_EBP, %ebp
L(end):
mov SAVE_EBX, %ebx
ifdef(`CPU_P6',`
xor %eax, %eax
shr $1, %edx
adc %edx, %eax
',`
adc $0, %edx
mov %edx, %eax
')
pop rp FRAME_popl()
pop up FRAME_popl()
ret
EPILOGUE()
ASM_END()
|
