aboutsummaryrefslogtreecommitdiff
path: root/vendor/gmp-6.3.0/mpn/alpha/ev5/diveby3.asm
blob: 3758188e0200545270919fb582ac4ea440a55712 (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
dnl  Alpha mpn_divexact_by3c -- mpn division by 3, expecting no remainder.

dnl  Copyright 2004, 2005, 2009 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:    22
C EV5:    11.5
C EV6:     6.3		Note that mpn_bdiv_dbm1c is faster

C TODO
C  * Remove the unops, they benefit just ev6, which no longer uses this file.
C  * Try prefetch for destination, using lds.
C  * Improve feed-in code, by moving initial mulq earlier; make initial load
C    to u0/u0 to save some copying.
C  * Combine u0 and u2, u1 and u3.

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

ASM_START()

DATASTART(L(LC),8)
	.quad	0xAAAAAAAAAAAAAAAB
	.quad	0x5555555555555555
	.quad	0xAAAAAAAAAAAAAAAA
DATAEND()

define(`xAAAAAAAAAAAAAAAB',	`r20')
define(`x5555555555555555',	`r21')
define(`xAAAAAAAAAAAAAAAA',	`r22')
define(`u0',	`r0')	define(`u1',	`r1')
define(`u2',	`r2')	define(`u3',	`r3')
define(`l0',	`r25')	define(`x',	`r8')
define(`q0',	`r4')	define(`q1',	`r5')
define(`p6',	`r6')	define(`p7',	`r7')
define(`t0',	`r23')	define(`t1',	`r24')
define(`cymask',`r28')


PROLOGUE(mpn_divexact_by3c,gp)

	ldq	r28, 0(up)			C load first limb early

C Put magic constants in registers
	lda	r0, L(LC)
	ldq	xAAAAAAAAAAAAAAAB, 0(r0)
	ldq	x5555555555555555, 8(r0)
	ldq	xAAAAAAAAAAAAAAAA, 16(r0)

C Compute initial l0 value
	cmpeq	cy, 1, p6
	cmpeq	cy, 2, p7
	negq	p6, p6
	and	p6, x5555555555555555, l0
	cmovne	p7, xAAAAAAAAAAAAAAAA, l0

C Feed-in depending on (n mod 4)
	and	n, 3, r8
	lda	n, -3(n)
	cmpeq	r8, 1, r4
	cmpeq	r8, 2, r5
	bne	r4, $Lb01
	bne	r5, $Lb10
	beq	r8, $Lb00

$Lb11:	ldq	u3, 8(up)
	lda	up, -24(up)
	lda	rp, -24(rp)
	mulq	r28, xAAAAAAAAAAAAAAAB, q0
	mov	r28, u2
	br	r31, $L11

$Lb00:	ldq	u2, 8(up)
	lda	up, -16(up)
	lda	rp, -16(rp)
	mulq	r28, xAAAAAAAAAAAAAAAB, q1
	mov	r28, u1
	br	r31, $L00

$Lb01:	lda	rp, -8(rp)
	mulq	r28, xAAAAAAAAAAAAAAAB, q0
	mov	r28, u0
	blt	n, $Lcj1
	ldq	u1, 8(up)
	lda	up, -8(up)
	br	r31, $L01

$Lb10:	ldq	u0, 8(up)
	mulq	r28, xAAAAAAAAAAAAAAAB, q1
	mov	r28, u3
	blt	n, $Lend

	ALIGN(16)
$Ltop:
C 0
	cmpult	u3, cy, cy			C L0
	mulq	u0, xAAAAAAAAAAAAAAAB, q0	C U1
	ldq	u1, 16(up)			C L1
	addq	q1, l0, x			C U0
C 1
	negq	cy, cymask			C L0
	unop					C U1
	unop					C L1
	cmpult	x5555555555555555, x, p6	C U0
C 2
	cmpult	xAAAAAAAAAAAAAAAA, x, p7	C U1
	unop
	unop
	negq	p6, t0				C L0
C 3
	negq	p7, t1				C L0
	and	cymask, x5555555555555555, l0	C U1
	addq	p6, cy, cy
	and	t0, x5555555555555555, t0
C 4
	and	t1, x5555555555555555, t1
	addq	p7, cy, cy
	unop
	addq	t0, l0, l0
C 5
	addq	t1, l0, l0
	unop
	stq	x, 0(rp)			C L1
	unop
$L01:
C 0
	cmpult	u0, cy, cy			C L0
	mulq	u1, xAAAAAAAAAAAAAAAB, q1	C U1
	ldq	u2, 24(up)			C L1
	addq	q0, l0, x			C U0
C 1
	negq	cy, cymask			C L0
	unop					C U1
	unop					C L1
	cmpult	x5555555555555555, x, p6	C U0
C 2
	cmpult	xAAAAAAAAAAAAAAAA, x, p7	C U1
	unop
	unop
	negq	p6, t0				C L0
C 3
	negq	p7, t1				C L0
	and	cymask, x5555555555555555, l0	C U1
	addq	p6, cy, cy
	and	t0, x5555555555555555, t0
C 4
	and	t1, x5555555555555555, t1
	addq	p7, cy, cy
	unop
	addq	t0, l0, l0
C 5
	addq	t1, l0, l0
	unop
	stq	x, 8(rp)			C L1
	unop
$L00:
C 0
	cmpult	u1, cy, cy			C L0
	mulq	u2, xAAAAAAAAAAAAAAAB, q0	C U1
	ldq	u3, 32(up)			C L1
	addq	q1, l0, x			C U0
C 1
	negq	cy, cymask			C L0
	unop					C U1
	unop					C L1
	cmpult	x5555555555555555, x, p6	C U0
C 2
	cmpult	xAAAAAAAAAAAAAAAA, x, p7	C U1
	unop
	unop
	negq	p6, t0				C L0
C 3
	negq	p7, t1				C L0
	and	cymask, x5555555555555555, l0	C U1
	addq	p6, cy, cy
	and	t0, x5555555555555555, t0
C 4
	and	t1, x5555555555555555, t1
	addq	p7, cy, cy
	unop
	addq	t0, l0, l0
C 5
	addq	t1, l0, l0
	unop
	stq	x, 16(rp)			C L1
	unop
$L11:
C 0
	cmpult	u2, cy, cy			C L0
	mulq	u3, xAAAAAAAAAAAAAAAB, q1	C U1
	ldq	u0, 40(up)			C L1
	addq	q0, l0, x			C U0
C 1
	negq	cy, cymask			C L0
	unop					C U1
	unop					C L1
	cmpult	x5555555555555555, x, p6	C U0
C 2
	cmpult	xAAAAAAAAAAAAAAAA, x, p7	C U1
	lda	n, -4(n)			C L1 bookkeeping
	unop
	negq	p6, t0				C L0
C 3
	negq	p7, t1				C L0
	and	cymask, x5555555555555555, l0	C U1
	addq	p6, cy, cy
	and	t0, x5555555555555555, t0
C 4
	and	t1, x5555555555555555, t1
	addq	p7, cy, cy
	unop
	addq	t0, l0, l0
C 5
	addq	t1, l0, l0
	unop
	stq	x, 24(rp)			C L1
	lda	up, 32(up)
C
	ldl	r31, 256(up)			C prefetch
	unop
	lda	rp, 32(rp)
	bge	n, $Ltop			C U1
C *** MAIN LOOP END ***
$Lend:

	cmpult	u3, cy, cy			C L0
	mulq	u0, xAAAAAAAAAAAAAAAB, q0	C U1
	unop
	addq	q1, l0, x			C U0
C 1
	negq	cy, cymask			C L0
	unop					C U1
	unop					C L1
	cmpult	x5555555555555555, x, p6	C U0
C 2
	cmpult	xAAAAAAAAAAAAAAAA, x, p7	C U1
	unop
	unop
	negq	p6, t0				C L0
C 3
	negq	p7, t1				C L0
	and	cymask, x5555555555555555, l0	C U1
	addq	p6, cy, cy
	and	t0, x5555555555555555, t0
C 4
	and	t1, x5555555555555555, t1
	addq	p7, cy, cy
	unop
	addq	t0, l0, l0
C 5
	addq	t1, l0, l0
	unop
	stq	x, 0(rp)			C L1
	unop
$Lcj1:
	cmpult	u0, cy, cy			C L0
	addq	q0, l0, x			C U0
	cmpult	x5555555555555555, x, p6	C U0
	cmpult	xAAAAAAAAAAAAAAAA, x, p7	C U1
	addq	p6, cy, cy
	addq	p7, cy, r0
	stq	x, 8(rp)			C L1

	ret	r31,(r26),1
EPILOGUE()
ASM_END()

C This is useful for playing with various schedules.
C Expand as: one(0)one(1)one(2)one(3)
define(`one',`
C 0
	cmpult	`$'eval(($1+3)%4), cy, cy		C L0
	mulq	`$'$1, xAAAAAAAAAAAAAAAB, `$'eval(4+$1%2) C U1
	ldq	`$'eval(($1+1)%4), eval($1*8+16)(up)	C L1
	addq	`$'eval(4+($1+1)%2), l0, x		C U0
C 1
	negq	cy, cymask				C L0
	unop						C U1
	unop						C L1
	cmpult	x5555555555555555, x, p6		C U0
C 2
	cmpult	xAAAAAAAAAAAAAAAA, x, p7		C U1
	unop
	unop
	negq	p6, t0					C L0
C 3
	negq	p7, t1					C L0
	and	cymask, x5555555555555555, l0		C U1
	addq	p6, cy, cy
	and	t0, x5555555555555555, t0
C 4
	and	t1, x5555555555555555, t1
	addq	p7, cy, cy
	unop
	addq	t0, l0, l0
C 5
	addq	t1, l0, l0
	unop
	stq	x, eval($1*8)(rp)			C L1
	unop
')