aboutsummaryrefslogtreecommitdiff
path: root/vendor/gmp-6.3.0/mpn/alpha/ev6/aorsmul_1.asm
blob: 0e68e6e7adb4ca296449836fe5cc94b9bc59d0f8 (plain) (blame)
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
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
dnl  Alpha ev6 mpn_addmul_1 and mpn_submul_1.

dnl  Copyright 2000, 2003-2005, 2008 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:     3.5

C  INPUT PARAMETERS
define(`rp',	`r16')
define(`up',	`r17')
define(`n',	`r18')
define(`v0',	`r19')

dnl  This code was written in cooperation with ev6 pipeline expert Steve Root.

dnl  The stores can issue a cycle late so we have paired no-op's to 'catch'
dnl  them, so that further disturbance to the schedule is damped.

dnl  We couldn't pair the loads, because the entangled schedule of the carry's
dnl  has to happen on one side {0} of the machine.

dnl  This is a great schedule for the d_cache, a poor schedule for the b_cache.
dnl  The lockup on U0 means that any stall can't be recovered from.  Consider a
dnl  ldq in L1, say that load gets stalled because it collides with a fill from
dnl  the b_cache.  On the next cycle, this load gets priority.  If first looks
dnl  at L0, and goes there.  The instruction we intended for L0 gets to look at
dnl  L1, which is NOT where we want it.  It either stalls 1, because it can't
dnl  go in L0, or goes there, and causes a further instruction to stall.

dnl  So for b_cache, we're likely going to want to put one or more cycles back
dnl  into the code! And, of course, put in lds prefetch for the rp[] operand.
dnl  At a place where we have an mt followed by a bookkeeping, put the
dnl  bookkeeping in upper, and the prefetch into lower.

dnl  Note, the ldq's and stq's are at the end of the quadpacks.  Note, we'd
dnl  like not to have an ldq or an stq to preceded a conditional branch in a
dnl  quadpack.  The conditional branch moves the retire pointer one cycle
dnl  later.

ifdef(`OPERATION_addmul_1',`
    define(`ADDSUB',	`addq')
    define(`CMPCY',	`cmpult	$2,$1')
    define(`func',	`mpn_addmul_1')
')
ifdef(`OPERATION_submul_1',`
    define(`ADDSUB',	`subq')
    define(`CMPCY',	`cmpult	$1,$2')
    define(`func',	`mpn_submul_1')
')

MULFUNC_PROLOGUE(mpn_addmul_1 mpn_submul_1)

ASM_START()
PROLOGUE(func)
	ldq	r3,	0(up)		C
	and	r18,	7,	r20	C
	lda	r18,	-9(r18)		C
	cmpeq	r20,	1,	r21	C
	beq	r21,	$L1		C

$1mod8:	ldq	r5,	0(rp)		C
	mulq	v0,	r3,	r7	C
	umulh	v0,	r3,	r8	C
	ADDSUB	r5,	r7,	r23	C
	CMPCY(	r5,	r23),	r20	C
	addq	r8,	r20,	r0	C
	stq	r23,	0(rp)		C
	bge	r18,	$ent1		C
	ret	r31,	(r26),	1	C

$L1:	lda	r8,	0(r31)		C zero carry reg
	lda	r24,	0(r31)		C zero carry reg
	cmpeq	r20,	2,	r21	C
	bne	r21,	$2mod8		C
	cmpeq	r20,	3,	r21	C
	bne	r21,	$3mod8		C
	cmpeq	r20,	4,	r21	C
	bne	r21,	$4mod8		C
	cmpeq	r20,	5,	r21	C
	bne	r21,	$5mod8		C
	cmpeq	r20,	6,	r21	C
	bne	r21,	$6mod8		C
	cmpeq	r20,	7,	r21	C
	beq	r21,	$0mod8		C

$7mod8:	ldq	r5,	0(rp)		C
	lda	up,	8(up)		C
	mulq	v0,	r3,	r7	C
	umulh	v0,	r3,	r24	C
	ADDSUB	r5,	r7,	r23	C
	CMPCY(	r5,	r23),	r20	C
	addq	r24,	r20,	r24	C
	stq	r23,	0(rp)		C
	lda	rp,	8(rp)		C
	ldq	r3,	0(up)		C
$6mod8:	ldq	r1,	8(up)		C
	mulq	v0,	r3,	r25	C
	umulh	v0,	r3,	r3	C
	mulq	v0,	r1,	r28	C
	ldq	r0,	16(up)		C
	ldq	r4,	0(rp)		C
	umulh	v0,	r1,	r8	C
	ldq	r1,	24(up)		C
	lda	up,	48(up)		C L1 bookkeeping
	mulq	v0,	r0,	r2	C
	ldq	r5,	8(rp)		C
	lda	rp,	-32(rp)		C L1 bookkeeping
	umulh	v0,	r0,	r6	C
	ADDSUB	r4,	r25,	r25	C lo + acc
	mulq	v0,	r1,	r7	C
	br	r31,	$ent6		C

$ent1:	lda	up,	8(up)		C
	lda	rp,	8(rp)		C
	lda	r8,	0(r0)		C
	ldq	r3,	0(up)		C
$0mod8:	ldq	r1,	8(up)		C
	mulq	v0,	r3,	r2	C
	umulh	v0,	r3,	r6	C
	mulq	v0,	r1,	r7	C
	ldq	r0,	16(up)		C
	ldq	r4,	0(rp)		C
	umulh	v0,	r1,	r24	C
	ldq	r1,	24(up)		C
	mulq	v0,	r0,	r25	C
	ldq	r5,	8(rp)		C
	umulh	v0,	r0,	r3	C
	ADDSUB	r4,	r2,	r2	C lo + acc
	mulq	v0,	r1,	r28	C
	lda	rp,	-16(rp)		C
	br	r31,	$ent0		C

$3mod8:	ldq	r5,	0(rp)		C
	lda	up,	8(up)		C
	mulq	v0,	r3,	r7	C
	umulh	v0,	r3,	r8	C
	ADDSUB	r5,	r7,	r23	C
	CMPCY(	r5,	r23),	r20	C
	addq	r8,	r20,	r24	C
	stq	r23,	0(rp)		C
	lda	rp,	8(rp)		C
	ldq	r3,	0(up)		C
$2mod8:	ldq	r1,	8(up)		C
	mulq	v0,	r3,	r25	C
	umulh	v0,	r3,	r3	C
	mulq	v0,	r1,	r28	C
	ble	r18,	$n23		C
	ldq	r0,	16(up)		C
	ldq	r4,	0(rp)		C
	umulh	v0,	r1,	r8	C
	ldq	r1,	24(up)		C
	lda	up,	16(up)		C L1 bookkeeping
	mulq	v0,	r0,	r2	C
	ldq	r5,	8(rp)		C
	lda	rp,	0(rp)		C L1 bookkeeping
	umulh	v0,	r0,	r6	C
	ADDSUB	r4,	r25,	r25	C lo + acc
	mulq	v0,	r1,	r7	C
	br	r31,	$ent2		C

$5mod8:	ldq	r5,	0(rp)		C
	lda	up,	8(up)		C
	mulq	v0,	r3,	r7	C
	umulh	v0,	r3,	r24	C
	ADDSUB	r5,	r7,	r23	C
	CMPCY(	r5,	r23),	r20	C
	addq	r24,	r20,	r8	C
	stq	r23,	0(rp)		C
	lda	rp,	8(rp)		C
	ldq	r3,	0(up)		C
$4mod8:	ldq	r1,	8(up)		C
	mulq	v0,	r3,	r2	C
	umulh	v0,	r3,	r6	C
	mulq	v0,	r1,	r7	C
	ldq	r0,	16(up)		C
	ldq	r4,	0(rp)		C
	umulh	v0,	r1,	r24	C
	ldq	r1,	24(up)		C
	lda	up,	32(up)		C L1 bookkeeping
	mulq	v0,	r0,	r25	C
	ldq	r5,	8(rp)		C
	lda	rp,	16(rp)		C L1 bookkeeping
	umulh	v0,	r0,	r3	C
	ADDSUB	r4,	r2,	r2	C lo + acc
	mulq	v0,	r1,	r28	C
	CMPCY(	r4,	r2),	r20	C L0 lo add => carry
	ADDSUB	r2,	r8,	r22	C U0 hi add => answer
	ble	r18,	$Lend		C
	ALIGN(16)
$Loop:
	bis	r31,	r31,	r31	C U1 mt
	CMPCY(	r2,	r22),	r21	C L0 hi add => carry
	addq	r6,	r20,	r6	C U0 hi mul + carry
	ldq	r0,	0(up)		C

	bis	r31,	r31,	r31	C U1 mt
	ADDSUB	r5,	r7,	r7	C L0 lo + acc
	addq	r6,	r21,	r6	C U0 hi mul + carry
	ldq	r4,	0(rp)		C L1

	umulh	v0,	r1,	r8	C U1
	CMPCY(	r5,	r7),	r20	C L0 lo add => carry
	ADDSUB	r7,	r6,	r23	C U0 hi add => answer
	ldq	r1,	8(up)		C L1

	mulq	v0,	r0,	r2	C U1
	CMPCY(	r7,	r23),	r21	C L0 hi add => carry
	addq	r24,	r20,	r24	C U0 hi mul + carry
	ldq	r5,	8(rp)		C L1

	umulh	v0,	r0,	r6	C U1
	ADDSUB	r4,	r25,	r25	C U0 lo + acc
	stq	r22,	-16(rp)		C L0
	stq	r23,	-8(rp)		C L1

	bis	r31,	r31,	r31	C L0 st slosh
	mulq	v0,	r1,	r7	C U1
	bis	r31,	r31,	r31	C L1 st slosh
	addq	r24,	r21,	r24	C U0 hi mul + carry
$ent2:
	CMPCY(	r4,	r25),	r20	C L0 lo add => carry
	bis	r31,	r31,	r31	C U1 mt
	lda	r18,	-8(r18)		C L1 bookkeeping
	ADDSUB	r25,	r24,	r22	C U0 hi add => answer

	bis	r31,	r31,	r31	C U1 mt
	CMPCY(	r25,	r22),	r21	C L0 hi add => carry
	addq	r3,	r20,	r3	C U0 hi mul + carry
	ldq	r0,	16(up)		C L1

	bis	r31,	r31,	r31	C U1 mt
	ADDSUB	r5,	r28,	r28	C L0 lo + acc
	addq	r3,	r21,	r3	C U0 hi mul + carry
	ldq	r4,	16(rp)		C L1

	umulh	v0,	r1,	r24	C U1
	CMPCY(	r5,	r28),	r20	C L0 lo add => carry
	ADDSUB	r28,	r3,	r23	C U0 hi add => answer
	ldq	r1,	24(up)		C L1

	mulq	v0,	r0,	r25	C U1
	CMPCY(	r28,	r23),	r21	C L0 hi add => carry
	addq	r8,	r20,	r8	C U0 hi mul + carry
	ldq	r5,	24(rp)		C L1

	umulh	v0,	r0,	r3	C U1
	ADDSUB	r4,	r2,	r2	C U0 lo + acc
	stq	r22,	0(rp)		C L0
	stq	r23,	8(rp)		C L1

	bis	r31,	r31,	r31	C L0 st slosh
	mulq	v0,	r1,	r28	C U1
	bis	r31,	r31,	r31	C L1 st slosh
	addq	r8,	r21,	r8	C U0 hi mul + carry
$ent0:
	CMPCY(	r4,	r2),	r20	C L0 lo add => carry
	bis	r31,	r31,	r31	C U1 mt
	lda	up,	64(up)		C L1 bookkeeping
	ADDSUB	r2,	r8,	r22	C U0 hi add => answer

	bis	r31,	r31,	r31	C U1 mt
	CMPCY(	r2,	r22),	r21	C L0 hi add => carry
	addq	r6,	r20,	r6	C U0 hi mul + carry
	ldq	r0,	-32(up)		C L1

	bis	r31,	r31,	r31	C U1 mt
	ADDSUB	r5,	r7,	r7	C L0 lo + acc
	addq	r6,	r21,	r6	C U0 hi mul + carry
	ldq	r4,	32(rp)		C L1

	umulh	v0,	r1,	r8	C U1
	CMPCY(	r5,	r7),	r20	C L0 lo add => carry
	ADDSUB	r7,	r6,	r23	C U0 hi add => answer
	ldq	r1,	-24(up)		C L1

	mulq	v0,	r0,	r2	C U1
	CMPCY(	r7,	r23),	r21	C L0 hi add => carry
	addq	r24,	r20,	r24	C U0 hi mul + carry
	ldq	r5,	40(rp)		C L1

	umulh	v0,	r0,	r6	C U1
	ADDSUB	r4,	r25,	r25	C U0 lo + acc
	stq	r22,	16(rp)		C L0
	stq	r23,	24(rp)		C L1

	bis	r31,	r31,	r31	C L0 st slosh
	mulq	v0,	r1,	r7	C U1
	bis	r31,	r31,	r31	C L1 st slosh
	addq	r24,	r21,	r24	C U0 hi mul + carry
$ent6:
	CMPCY(	r4,	r25),	r20	C L0 lo add => carry
	bis	r31,	r31,	r31	C U1 mt
	lda	rp,	64(rp)		C L1 bookkeeping
	ADDSUB	r25,	r24,	r22	C U0 hi add => answer

	bis	r31,	r31,	r31	C U1 mt
	CMPCY(	r25,	r22),	r21	C L0 hi add => carry
	addq	r3,	r20,	r3	C U0 hi mul + carry
	ldq	r0,	-16(up)		C L1

	bis	r31,	r31,	r31	C U1 mt
	ADDSUB	r5,	r28,	r28	C L0 lo + acc
	addq	r3,	r21,	r3	C U0 hi mul + carry
	ldq	r4,	-16(rp)		C L1

	umulh	v0,	r1,	r24	C U1
	CMPCY(	r5,	r28),	r20	C L0 lo add => carry
	ADDSUB	r28,	r3,	r23	C U0 hi add => answer
	ldq	r1,	-8(up)		C L1

	mulq	v0,	r0,	r25	C U1
	CMPCY(	r28,	r23),	r21	C L0 hi add => carry
	addq	r8,	r20,	r8	C U0 hi mul + carry
	ldq	r5,	-8(rp)		C L1

	umulh	v0,	r0,	r3	C U1
	ADDSUB	r4,	r2,	r2	C U0 lo + acc
	stq	r22,	-32(rp)		C L0
	stq	r23,	-24(rp)		C L1

	bis	r31,	r31,	r31	C L0 st slosh
	mulq	v0,	r1,	r28	C U1
	bis	r31,	r31,	r31	C L1 st slosh
	addq	r8,	r21,	r8	C U0 hi mul + carry

	CMPCY(	r4,	r2),	r20	C L0 lo add => carry
	ADDSUB	r2,	r8,	r22	C U0 hi add => answer
	ldl	r31,	256(up)		C prefetch up[]
	bgt	r18,	$Loop		C U1 bookkeeping

$Lend:	CMPCY(	r2,	r22),	r21	C
	addq	r6,	r20,	r6	C
	ADDSUB	r5,	r7,	r7	C
	addq	r6,	r21,	r6	C
	ldq	r4,	0(rp)		C
	umulh	v0,	r1,	r8	C
	CMPCY(	r5,	r7),	r20	C
	ADDSUB	r7,	r6,	r23	C
	CMPCY(r7,	r23),	r21	C
	addq	r24,	r20,	r24	C
	ldq	r5,	8(rp)		C
	ADDSUB	r4,	r25,	r25	C
	stq	r22,	-16(rp)		C
	stq	r23,	-8(rp)		C
	addq	r24,	r21,	r24	C
	br	L(x)

	ALIGN(16)
$n23:	ldq	r4,	0(rp)		C
	ldq	r5,	8(rp)		C
	umulh	v0,	r1,	r8	C
	ADDSUB	r4,	r25,	r25	C
L(x):	CMPCY(	r4,	r25),	r20	C
	ADDSUB	r25,	r24,	r22	C
	CMPCY(	r25,	r22),	r21	C
	addq	r3,	r20,	r3	C
	ADDSUB	r5,	r28,	r28	C
	addq	r3,	r21,	r3	C
	CMPCY(	r5,	r28),	r20	C
	ADDSUB	r28,	r3,	r23	C
	CMPCY(	r28,	r23),	r21	C
	addq	r8,	r20,	r8	C
	stq	r22,	0(rp)		C
	stq	r23,	8(rp)		C
	addq	r8,	r21,	r0	C
	ret	r31,	(r26),	1	C
EPILOGUE()
ASM_END()