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
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
|
/* Optimized memcpy implementation for CELL BE PowerPC.
Copyright (C) 2010-2015 Free Software Foundation, Inc.
This file is part of the GNU C Library.
The GNU C Library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
The GNU C Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
#include <sysdep.h>
#define PREFETCH_AHEAD 6 /* no cache lines SRC prefetching ahead */
#define ZERO_AHEAD 4 /* no cache lines DST zeroing ahead */
/* memcpy routine optimized for CELL-BE-PPC v2.0
*
* The CELL PPC core has 1 integer unit and 1 load/store unit
* CELL:
* 1st level data cache = 32K
* 2nd level data cache = 512K
* 3rd level data cache = 0K
* With 3.2 GHz clockrate the latency to 2nd level cache is >36 clocks,
* latency to memory is >400 clocks
* To improve copy performance we need to prefetch source data
* far ahead to hide this latency
* For best performance instruction forms ending in "." like "andi."
* should be avoided as the are implemented in microcode on CELL.
* The below code is loop unrolled for the CELL cache line of 128 bytes
*/
.align 7
EALIGN (memcpy, 5, 0)
CALL_MCOUNT
dcbt 0,r4 /* Prefetch ONE SRC cacheline */
cmplwi cr1,r5,16 /* is size < 16 ? */
mr r6,r3
blt+ cr1,.Lshortcopy
.Lbigcopy:
neg r8,r3 /* LS 3 bits = # bytes to 8-byte dest bdry */
clrlwi r8,r8,32-4 /* align to 16byte boundary */
sub r7,r4,r3
cmplwi cr0,r8,0
beq+ .Ldst_aligned
.Ldst_unaligned:
mtcrf 0x01,r8 /* put #bytes to boundary into cr7 */
subf r5,r8,r5
bf cr7*4+3,1f
lbzx r0,r7,r6 /* copy 1 byte */
stb r0,0(r6)
addi r6,r6,1
1: bf cr7*4+2,2f
lhzx r0,r7,r6 /* copy 2 byte */
sth r0,0(r6)
addi r6,r6,2
2: bf cr7*4+1,4f
lwzx r0,r7,r6 /* copy 4 byte */
stw r0,0(r6)
addi r6,r6,4
4: bf cr7*4+0,8f
lfdx fp9,r7,r6 /* copy 8 byte */
stfd fp9,0(r6)
addi r6,r6,8
8:
add r4,r7,r6
.Ldst_aligned:
cmpwi cr5,r5,128-1
neg r7,r6
addi r6,r6,-8 /* prepare for stfdu */
addi r4,r4,-8 /* prepare for lfdu */
clrlwi r7,r7,32-7 /* align to cacheline boundary */
ble+ cr5,.Llessthancacheline
cmplwi cr6,r7,0
subf r5,r7,r5
srwi r7,r7,4 /* divide size by 16 */
srwi r10,r5,7 /* number of cache lines to copy */
cmplwi r10,0
li r11,0 /* number cachelines to copy with prefetch */
beq .Lnocacheprefetch
cmplwi r10,PREFETCH_AHEAD
li r12,128+8 /* prefetch distance */
ble .Llessthanmaxprefetch
subi r11,r10,PREFETCH_AHEAD
li r10,PREFETCH_AHEAD
.Llessthanmaxprefetch:
mtctr r10
.LprefetchSRC:
dcbt r12,r4
addi r12,r12,128
bdnz .LprefetchSRC
.Lnocacheprefetch:
mtctr r7
cmplwi cr1,r5,128
clrlwi r5,r5,32-7
beq cr6,.Lcachelinealigned
.Laligntocacheline:
lfd fp9,0x08(r4)
lfdu fp10,0x10(r4)
stfd fp9,0x08(r6)
stfdu fp10,0x10(r6)
bdnz .Laligntocacheline
.Lcachelinealigned: /* copy while cache lines */
blt- cr1,.Llessthancacheline /* size <128 */
.Louterloop:
cmpwi r11,0
mtctr r11
beq- .Lendloop
li r11,128*ZERO_AHEAD +8 /* DCBZ dist */
.align 4
/* Copy whole cachelines, optimized by prefetching SRC cacheline */
.Lloop: /* Copy aligned body */
dcbt r12,r4 /* PREFETCH SOURCE some cache lines ahead */
lfd fp9, 0x08(r4)
dcbz r11,r6
lfd fp10, 0x10(r4) /* 4 register stride copy is optimal */
lfd fp11, 0x18(r4) /* to hide 1st level cache latency. */
lfd fp12, 0x20(r4)
stfd fp9, 0x08(r6)
stfd fp10, 0x10(r6)
stfd fp11, 0x18(r6)
stfd fp12, 0x20(r6)
lfd fp9, 0x28(r4)
lfd fp10, 0x30(r4)
lfd fp11, 0x38(r4)
lfd fp12, 0x40(r4)
stfd fp9, 0x28(r6)
stfd fp10, 0x30(r6)
stfd fp11, 0x38(r6)
stfd fp12, 0x40(r6)
lfd fp9, 0x48(r4)
lfd fp10, 0x50(r4)
lfd fp11, 0x58(r4)
lfd fp12, 0x60(r4)
stfd fp9, 0x48(r6)
stfd fp10, 0x50(r6)
stfd fp11, 0x58(r6)
stfd fp12, 0x60(r6)
lfd fp9, 0x68(r4)
lfd fp10, 0x70(r4)
lfd fp11, 0x78(r4)
lfdu fp12, 0x80(r4)
stfd fp9, 0x68(r6)
stfd fp10, 0x70(r6)
stfd fp11, 0x78(r6)
stfdu fp12, 0x80(r6)
bdnz .Lloop
.Lendloop:
cmpwi r10,0
slwi r10,r10,2 /* adjust from 128 to 32 byte stride */
beq- .Lendloop2
mtctr r10
.Lloop2: /* Copy aligned body */
lfd fp9, 0x08(r4)
lfd fp10, 0x10(r4)
lfd fp11, 0x18(r4)
lfdu fp12, 0x20(r4)
stfd fp9, 0x08(r6)
stfd fp10, 0x10(r6)
stfd fp11, 0x18(r6)
stfdu fp12, 0x20(r6)
bdnz .Lloop2
.Lendloop2:
.Llessthancacheline: /* less than cache to do ? */
cmplwi cr0,r5,16
srwi r7,r5,4 /* divide size by 16 */
blt- .Ldo_lt16
mtctr r7
.Lcopy_remaining:
lfd fp9,0x08(r4)
lfdu fp10,0x10(r4)
stfd fp9,0x08(r6)
stfdu fp10,0x10(r6)
bdnz .Lcopy_remaining
.Ldo_lt16: /* less than 16 ? */
cmplwi cr0,r5,0 /* copy remaining bytes (0-15) */
beqlr+ /* no rest to copy */
addi r4,r4,8
addi r6,r6,8
.Lshortcopy: /* SIMPLE COPY to handle size =< 15 bytes */
mtcrf 0x01,r5
sub r7,r4,r6
bf- cr7*4+0,8f
lfdx fp9,r7,r6 /* copy 8 byte */
stfd fp9,0(r6)
addi r6,r6,8
8:
bf cr7*4+1,4f
lwzx r0,r7,r6 /* copy 4 byte */
stw r0,0(r6)
addi r6,r6,4
4:
bf cr7*4+2,2f
lhzx r0,r7,r6 /* copy 2 byte */
sth r0,0(r6)
addi r6,r6,2
2:
bf cr7*4+3,1f
lbzx r0,r7,r6 /* copy 1 byte */
stb r0,0(r6)
1: blr
END (memcpy)
libc_hidden_builtin_def (memcpy)
|
