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
|
dnl SPARC v9 32-bit mpn_submul_1 -- Multiply a limb vector with a limb and
dnl subtract the result from a second limb vector.
dnl Copyright 1998, 2000, 2001, 2003 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 Algorithm: We use two floating-point multiplies per limb product, with the
C invariant v operand split into two 16-bit pieces, and the u operand split
C into 32-bit pieces. We convert the two 48-bit products and transfer them to
C the integer unit.
C cycles/limb
C UltraSPARC 1&2: 6.5
C UltraSPARC 3: ?
C Possible optimizations:
C 1. Combine 32-bit memory operations into 64-bit operations. Since we're
C memory bandwidth limited, this could save 1.5 cycles/limb.
C 2. Unroll the inner loop. Since we already use alternate temporary areas,
C it is very straightforward to unroll, using an exit branch midways.
C Unrolling would allow deeper scheduling which could improve speed for L2
C cache case.
C 3. For mpn_mul_1: Use more alternating temp areas. The std'es and ldx'es
C aren't sufficiently apart-scheduled with just two temp areas.
C 4. Specialize for particular v values. If its upper 16 bits are zero, we
C could save many operations.
C INPUT PARAMETERS
C rp i0
C up i1
C n i2
C v i3
define(`FSIZE',224)
ASM_START()
PROLOGUE(mpn_submul_1)
add %sp, -FSIZE, %sp
sethi %hi(0xffff), %g1
srl %o3, 16, %g2
or %g1, %lo(0xffff), %g1
and %o3, %g1, %g1
stx %g1, [%sp+104]
stx %g2, [%sp+112]
ldd [%sp+104], %f6
ldd [%sp+112], %f8
fxtod %f6, %f6
fxtod %f8, %f8
ld [%sp+104], %f10 C zero f10
mov 0, %g3 C cy = 0
define(`fanop', `fitod %f18, %f0') C A quasi nop running in the FA pipe
add %sp, 160, %o5 C point in scratch area
and %o5, -32, %o5 C align at 0 (mod 32) in scratch area
subcc %o2, 1, %o2
ld [%o1], %f11 C read up[i]
add %o1, 4, %o1 C up++
bne,pt %icc, .L_two_or_more
fxtod %f10, %f2
fmuld %f2, %f8, %f16
fmuld %f2, %f6, %f4
fdtox %f16, %f14
fdtox %f4, %f12
std %f14, [%o5+16]
std %f12, [%o5+24]
ldx [%o5+16], %g2 C p16
ldx [%o5+24], %g1 C p0
lduw [%o0], %g5 C read rp[i]
b .L1
add %o0, -16, %o0
.align 16
.L_two_or_more:
subcc %o2, 1, %o2
ld [%o1], %f11 C read up[i]
fmuld %f2, %f8, %f16
fmuld %f2, %f6, %f4
add %o1, 4, %o1 C up++
bne,pt %icc, .L_three_or_more
fxtod %f10, %f2
fdtox %f16, %f14
fdtox %f4, %f12
std %f14, [%o5+16]
fmuld %f2, %f8, %f16
std %f12, [%o5+24]
fmuld %f2, %f6, %f4
fdtox %f16, %f14
fdtox %f4, %f12
std %f14, [%o5+0]
std %f12, [%o5+8]
lduw [%o0], %g5 C read rp[i]
ldx [%o5+16], %g2 C p16
ldx [%o5+24], %g1 C p0
b .L2
add %o0, -12, %o0
.align 16
.L_three_or_more:
subcc %o2, 1, %o2
ld [%o1], %f11 C read up[i]
fdtox %f16, %f14
fdtox %f4, %f12
std %f14, [%o5+16]
fmuld %f2, %f8, %f16
std %f12, [%o5+24]
fmuld %f2, %f6, %f4
add %o1, 4, %o1 C up++
bne,pt %icc, .L_four_or_more
fxtod %f10, %f2
fdtox %f16, %f14
fdtox %f4, %f12
std %f14, [%o5+0]
fmuld %f2, %f8, %f16
std %f12, [%o5+8]
fmuld %f2, %f6, %f4
fdtox %f16, %f14
ldx [%o5+16], %g2 C p16
fdtox %f4, %f12
ldx [%o5+24], %g1 C p0
std %f14, [%o5+16]
std %f12, [%o5+24]
lduw [%o0], %g5 C read rp[i]
b .L3
add %o0, -8, %o0
.align 16
.L_four_or_more:
subcc %o2, 1, %o2
ld [%o1], %f11 C read up[i]
fdtox %f16, %f14
fdtox %f4, %f12
std %f14, [%o5+0]
fmuld %f2, %f8, %f16
std %f12, [%o5+8]
fmuld %f2, %f6, %f4
add %o1, 4, %o1 C up++
bne,pt %icc, .L_five_or_more
fxtod %f10, %f2
fdtox %f16, %f14
ldx [%o5+16], %g2 C p16
fdtox %f4, %f12
ldx [%o5+24], %g1 C p0
std %f14, [%o5+16]
fmuld %f2, %f8, %f16
std %f12, [%o5+24]
fmuld %f2, %f6, %f4
add %o1, 4, %o1 C up++
lduw [%o0], %g5 C read rp[i]
b .L4
add %o0, -4, %o0
.align 16
.L_five_or_more:
subcc %o2, 1, %o2
ld [%o1], %f11 C read up[i]
fdtox %f16, %f14
ldx [%o5+16], %g2 C p16
fdtox %f4, %f12
ldx [%o5+24], %g1 C p0
std %f14, [%o5+16]
fmuld %f2, %f8, %f16
std %f12, [%o5+24]
fmuld %f2, %f6, %f4
add %o1, 4, %o1 C up++
lduw [%o0], %g5 C read rp[i]
bne,pt %icc, .Loop
fxtod %f10, %f2
b,a .L5
C BEGIN MAIN LOOP
.align 16
C -- 0
.Loop: sub %g0, %g3, %g3
subcc %o2, 1, %o2
ld [%o1], %f11 C read up[i]
fdtox %f16, %f14
C -- 1
sllx %g2, 16, %g4 C (p16 << 16)
add %o0, 4, %o0 C rp++
ldx [%o5+0], %g2 C p16
fdtox %f4, %f12
C -- 2
srl %g3, 0, %g3 C zero most significant 32 bits
add %g1, %g4, %g4 C p = p0 + (p16 << 16)
ldx [%o5+8], %g1 C p0
fanop
C -- 3
nop
add %g3, %g4, %g4 C p += cy
std %f14, [%o5+0]
fmuld %f2, %f8, %f16
C -- 4
nop
sub %g5, %g4, %g4 C p += rp[i]
std %f12, [%o5+8]
fmuld %f2, %f6, %f4
C -- 5
xor %o5, 16, %o5 C alternate scratch variables
add %o1, 4, %o1 C up++
stw %g4, [%o0-4]
fanop
C -- 6
srlx %g4, 32, %g3 C new cy
lduw [%o0], %g5 C read rp[i]
bne,pt %icc, .Loop
fxtod %f10, %f2
C END MAIN LOOP
.L5: sub %g0, %g3, %g3
fdtox %f16, %f14
sllx %g2, 16, %g4 C (p16 << 16)
ldx [%o5+0], %g2 C p16
fdtox %f4, %f12
srl %g3, 0, %g3 C zero most significant 32 bits
add %g1, %g4, %g4 C p = p0 + (p16 << 16)
ldx [%o5+8], %g1 C p0
add %g4, %g3, %g4 C p += cy
std %f14, [%o5+0]
fmuld %f2, %f8, %f16
sub %g5, %g4, %g4 C p += rp[i]
std %f12, [%o5+8]
fmuld %f2, %f6, %f4
xor %o5, 16, %o5
stw %g4, [%o0+0]
srlx %g4, 32, %g3 C new cy
lduw [%o0+4], %g5 C read rp[i]
sub %g0, %g3, %g3
.L4: fdtox %f16, %f14
sllx %g2, 16, %g4 C (p16 << 16)
ldx [%o5+0], %g2 C p16
fdtox %f4, %f12
srl %g3, 0, %g3 C zero most significant 32 bits
add %g1, %g4, %g4 C p = p0 + (p16 << 16)
ldx [%o5+8], %g1 C p0
add %g3, %g4, %g4 C p += cy
std %f14, [%o5+0]
sub %g5, %g4, %g4 C p += rp[i]
std %f12, [%o5+8]
xor %o5, 16, %o5
stw %g4, [%o0+4]
srlx %g4, 32, %g3 C new cy
lduw [%o0+8], %g5 C read rp[i]
sub %g0, %g3, %g3
.L3: sllx %g2, 16, %g4 C (p16 << 16)
ldx [%o5+0], %g2 C p16
srl %g3, 0, %g3 C zero most significant 32 bits
add %g1, %g4, %g4 C p = p0 + (p16 << 16)
ldx [%o5+8], %g1 C p0
add %g3, %g4, %g4 C p += cy
sub %g5, %g4, %g4 C p += rp[i]
xor %o5, 16, %o5
stw %g4, [%o0+8]
srlx %g4, 32, %g3 C new cy
lduw [%o0+12], %g5 C read rp[i]
sub %g0, %g3, %g3
.L2: sllx %g2, 16, %g4 C (p16 << 16)
ldx [%o5+0], %g2 C p16
srl %g3, 0, %g3 C zero most significant 32 bits
add %g1, %g4, %g4 C p = p0 + (p16 << 16)
ldx [%o5+8], %g1 C p0
add %g3, %g4, %g4 C p += cy
sub %g5, %g4, %g4 C p += rp[i]
stw %g4, [%o0+12]
srlx %g4, 32, %g3 C new cy
lduw [%o0+16], %g5 C read rp[i]
sub %g0, %g3, %g3
.L1: sllx %g2, 16, %g4 C (p16 << 16)
srl %g3, 0, %g3 C zero most significant 32 bits
add %g1, %g4, %g4 C p = p0 + (p16 << 16)
add %g3, %g4, %g4 C p += cy
sub %g5, %g4, %g4 C p += rp[i]
stw %g4, [%o0+16]
srlx %g4, 32, %g3 C new cy
sub %g0, %g3, %o0
retl
sub %sp, -FSIZE, %sp
EPILOGUE(mpn_submul_1)
|
