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
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
|
dnl Alpha ev6 nails mpn_mul_1.
dnl Copyright 2002, 2005, 2006 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 EV4: 42
C EV5: 18
C EV6: 3.25
C TODO
C * Reroll loop for 3.0 c/l with current 4-way unrolling.
C * The loop is overscheduled wrt loads and wrt multiplies, in particular
C umulh.
C * Use FP loop count and multiple exit points, that would simplify feed-in lp0
C and would work since the loop structure is really regular.
C INPUT PARAMETERS
define(`rp',`r16')
define(`up',`r17')
define(`n', `r18')
define(`vl0',`r19')
define(`numb_mask',`r6')
define(`m0a',`r0')
define(`m0b',`r1')
define(`m1a',`r2')
define(`m1b',`r3')
define(`m2a',`r20')
define(`m2b',`r21')
define(`m3a',`r22')
define(`m3b',`r23')
define(`acc0',`r25')
define(`acc1',`r27')
define(`ul0',`r4')
define(`ul1',`r5')
define(`ul2',`r4')
define(`ul3',`r5')
define(`rl0',`r24')
define(`rl1',`r24')
define(`rl2',`r24')
define(`rl3',`r24')
define(`t0',`r7')
define(`t1',`r8')
define(`NAIL_BITS',`GMP_NAIL_BITS')
define(`NUMB_BITS',`GMP_NUMB_BITS')
dnl This declaration is munged by configure
NAILS_SUPPORT(1-63)
ASM_START()
PROLOGUE(mpn_mul_1)
sll vl0, NAIL_BITS, vl0
lda numb_mask, -1(r31)
srl numb_mask, NAIL_BITS, numb_mask
and n, 3, r25
cmpeq r25, 1, r21
bne r21, L(1m4)
cmpeq r25, 2, r21
bne r21, L(2m4)
beq r25, L(0m4)
L(3m4): ldq ul3, 0(up)
lda n, -4(n)
ldq ul0, 8(up)
mulq vl0, ul3, m3a
umulh vl0, ul3, m3b
ldq ul1, 16(up)
lda up, 24(up)
lda rp, -8(rp)
mulq vl0, ul0, m0a
umulh vl0, ul0, m0b
bge n, L(ge3)
mulq vl0, ul1, m1a
umulh vl0, ul1, m1b
srl m3a,NAIL_BITS, t0
addq t0, r31, acc1
srl m0a,NAIL_BITS, t0
addq t0, m3b, acc0
srl acc1,NUMB_BITS, t1
br r31, L(ta3)
L(ge3): ldq ul2, 0(up)
mulq vl0, ul1, m1a
umulh vl0, ul1, m1b
srl m3a,NAIL_BITS, t0
ldq ul3, 8(up)
lda n, -4(n)
mulq vl0, ul2, m2a
addq t0, r31, acc1
umulh vl0, ul2, m2b
srl m0a,NAIL_BITS, t0
ldq ul0, 16(up)
mulq vl0, ul3, m3a
addq t0, m3b, acc0
srl acc1,NUMB_BITS, t1
br r31, L(el3)
L(0m4): lda n, -8(n)
ldq ul2, 0(up)
ldq ul3, 8(up)
mulq vl0, ul2, m2a
umulh vl0, ul2, m2b
ldq ul0, 16(up)
mulq vl0, ul3, m3a
umulh vl0, ul3, m3b
ldq ul1, 24(up)
lda up, 32(up)
mulq vl0, ul0, m0a
umulh vl0, ul0, m0b
bge n, L(ge4)
srl m2a,NAIL_BITS, t0
mulq vl0, ul1, m1a
addq t0, r31, acc0
umulh vl0, ul1, m1b
srl m3a,NAIL_BITS, t0
addq t0, m2b, acc1
srl acc0,NUMB_BITS, t1
br r31, L(ta4)
L(ge4): srl m2a,NAIL_BITS, t0
ldq ul2, 0(up)
mulq vl0, ul1, m1a
addq t0, r31, acc0
umulh vl0, ul1, m1b
srl m3a,NAIL_BITS, t0
ldq ul3, 8(up)
lda n, -4(n)
mulq vl0, ul2, m2a
addq t0, m2b, acc1
srl acc0,NUMB_BITS, t1
br r31, L(el0)
L(2m4): lda n, -4(n)
ldq ul0, 0(up)
ldq ul1, 8(up)
lda up, 16(up)
lda rp, -16(rp)
mulq vl0, ul0, m0a
umulh vl0, ul0, m0b
bge n, L(ge2)
mulq vl0, ul1, m1a
umulh vl0, ul1, m1b
srl m0a,NAIL_BITS, t0
addq t0, r31, acc0
srl m1a,NAIL_BITS, t0
addq t0, m0b, acc1
srl acc0,NUMB_BITS, t1
br r31, L(ta2)
L(ge2): ldq ul2, 0(up)
mulq vl0, ul1, m1a
umulh vl0, ul1, m1b
ldq ul3, 8(up)
lda n, -4(n)
mulq vl0, ul2, m2a
umulh vl0, ul2, m2b
srl m0a,NAIL_BITS, t0
ldq ul0, 16(up)
mulq vl0, ul3, m3a
addq t0, r31, acc0
umulh vl0, ul3, m3b
srl m1a,NAIL_BITS, t0
ldq ul1, 24(up)
lda up, 32(up)
lda rp, 32(rp)
mulq vl0, ul0, m0a
addq t0, m0b, acc1
srl acc0,NUMB_BITS, t1
bge n, L(el2)
br r31, L(ta6)
L(1m4): lda n, -4(n)
ldq ul1, 0(up)
lda up, 8(up)
lda rp, -24(rp)
bge n, L(ge1)
mulq vl0, ul1, m1a
umulh vl0, ul1, m1b
srl m1a,NAIL_BITS, t0
addq t0, r31, acc1
and acc1,numb_mask, r28
srl acc1,NUMB_BITS, t1
stq r28, 24(rp)
addq t1, m1b, r0
ret r31, (r26), 1
L(ge1): ldq ul2, 0(up)
mulq vl0, ul1, m1a
umulh vl0, ul1, m1b
ldq ul3, 8(up)
lda n, -4(n)
mulq vl0, ul2, m2a
umulh vl0, ul2, m2b
ldq ul0, 16(up)
mulq vl0, ul3, m3a
umulh vl0, ul3, m3b
srl m1a,NAIL_BITS, t0
ldq ul1, 24(up)
lda up, 32(up)
lda rp, 32(rp)
mulq vl0, ul0, m0a
addq t0, r31, acc1
umulh vl0, ul0, m0b
srl m2a,NAIL_BITS, t0
mulq vl0, ul1, m1a
addq t0, m1b, acc0
srl acc1,NUMB_BITS, t1
blt n, L(ta5)
L(ge5): ldq ul2, 0(up)
br r31, L(el1)
ALIGN(16)
L(top): mulq vl0, ul0, m0a C U1
addq t0, m0b, acc1 C L0
srl acc0,NUMB_BITS, t1 C U0
stq r28, -24(rp) C L1
C
L(el2): umulh vl0, ul0, m0b C U1
and acc0,numb_mask, r28 C L0
unop C U0
unop C L1
C
unop C U1
addq t1, acc1, acc1 C L0
srl m2a,NAIL_BITS, t0 C U0
ldq ul2, 0(up) C L1
C
mulq vl0, ul1, m1a C U1
addq t0, m1b, acc0 C L0
srl acc1,NUMB_BITS, t1 C U0
stq r28, -16(rp) C L1
C
L(el1): umulh vl0, ul1, m1b C U1
and acc1,numb_mask, r28 C L0
unop C U0
lda n, -4(n) C L1
C
unop C U1
addq t1, acc0, acc0 C L0
srl m3a,NAIL_BITS, t0 C U0
ldq ul3, 8(up) C L1
C
mulq vl0, ul2, m2a C U1
addq t0, m2b, acc1 C L0
srl acc0,NUMB_BITS, t1 C U0
stq r28, -8(rp) C L1
C
L(el0): umulh vl0, ul2, m2b C U1
and acc0,numb_mask, r28 C L0
unop C U0
unop C L1
C
unop C U1
addq t1, acc1, acc1 C L0
srl m0a,NAIL_BITS, t0 C U0
ldq ul0, 16(up) C L1
C
mulq vl0, ul3, m3a C U1
addq t0, m3b, acc0 C L0
srl acc1,NUMB_BITS, t1 C U0
stq r28, 0(rp) C L1
C
L(el3): umulh vl0, ul3, m3b C U1
and acc1,numb_mask, r28 C L0
unop C U0
unop C L1
C
unop C U1
addq t1, acc0, acc0 C L0
srl m1a,NAIL_BITS, t0 C U0
ldq ul1, 24(up) C L1
C
lda up, 32(up) C L0
unop C U1
lda rp, 32(rp) C L1
bge n, L(top) C U0
L(end): mulq vl0, ul0, m0a
addq t0, m0b, acc1
srl acc0,NUMB_BITS, t1
stq r28, -24(rp)
L(ta6): umulh vl0, ul0, m0b
and acc0,numb_mask, r28
addq t1, acc1, acc1
srl m2a,NAIL_BITS, t0
mulq vl0, ul1, m1a
addq t0, m1b, acc0
srl acc1,NUMB_BITS, t1
stq r28, -16(rp)
L(ta5): umulh vl0, ul1, m1b
and acc1,numb_mask, r28
addq t1, acc0, acc0
srl m3a,NAIL_BITS, t0
addq t0, m2b, acc1
srl acc0,NUMB_BITS, t1
stq r28, -8(rp)
ALIGN(16)
L(ta4): and acc0,numb_mask, r28
addq t1, acc1, acc1
srl m0a,NAIL_BITS, t0
addq t0, m3b, acc0
srl acc1,NUMB_BITS, t1
stq r28, 0(rp)
unop
ALIGN(16)
L(ta3): and acc1,numb_mask, r28
addq t1, acc0, acc0
srl m1a,NAIL_BITS, t0
addq t0, m0b, acc1
srl acc0,NUMB_BITS, t1
stq r28, 8(rp)
unop
ALIGN(16)
L(ta2): and acc0,numb_mask, r28
addq t1, acc1, acc1
srl acc1,NUMB_BITS, t1
stq r28, 16(rp)
and acc1,numb_mask, r28
addq t1, m1b, r0
stq r28, 24(rp)
ret r31, (r26), 1
EPILOGUE()
ASM_END()
|
