new beta-0.90.0
[luatex.git] / source / libs / gmp / gmp-src / mpn / sparc64 / ultrasparct3 / mul_1.asm
blobaf05d627bcfd804c0e7e5e9ca86de28451e6d1f0
1 dnl SPARC v9 mpn_mul_1 for T3/T4/T5.
3 dnl Contributed to the GNU project by David Miller and Torbjörn Granlund.
5 dnl Copyright 2013 Free Software Foundation, Inc.
7 dnl This file is part of the GNU MP Library.
8 dnl
9 dnl The GNU MP Library is free software; you can redistribute it and/or modify
10 dnl it under the terms of either:
11 dnl
12 dnl * the GNU Lesser General Public License as published by the Free
13 dnl Software Foundation; either version 3 of the License, or (at your
14 dnl option) any later version.
15 dnl
16 dnl or
17 dnl
18 dnl * the GNU General Public License as published by the Free Software
19 dnl Foundation; either version 2 of the License, or (at your option) any
20 dnl later version.
21 dnl
22 dnl or both in parallel, as here.
23 dnl
24 dnl The GNU MP Library is distributed in the hope that it will be useful, but
25 dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
26 dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
27 dnl for more details.
28 dnl
29 dnl You should have received copies of the GNU General Public License and the
30 dnl GNU Lesser General Public License along with the GNU MP Library. If not,
31 dnl see https://www.gnu.org/licenses/.
33 include(`../config.m4')
35 C cycles/limb
36 C UltraSPARC T3: 23
37 C UltraSPARC T4: 3
39 C INPUT PARAMETERS
40 define(`rp', `%i0')
41 define(`up', `%i1')
42 define(`n', `%i2')
43 define(`v0', `%i3')
45 ASM_START()
46 REGISTER(%g2,#scratch)
47 REGISTER(%g3,#scratch)
48 PROLOGUE(mpn_mul_1)
49 save %sp, -176, %sp
51 and n, 3, %g5
52 add n, -4, n
53 brz %g5, L(b0)
54 cmp %g5, 2
55 bcs %xcc, L(b1)
56 nop
57 be %xcc, L(b2)
58 nop
60 L(b3): addcc %g0, %g0, %i5
61 ldx [up+0], %l0
62 ldx [up+8], %l1
63 ldx [up+16], %l2
64 mulx %l0, v0, %o0
65 umulxhi(%l0, v0, %o1)
66 brgz n, L(gt3)
67 add rp, -8, rp
68 mulx %l1, v0, %o2
69 umulxhi(%l1, v0, %o3)
70 b L(wd3)
71 nop
72 L(gt3): ldx [up+24], %l3
73 mulx %l1, v0, %o2
74 umulxhi(%l1, v0, %o3)
75 add up, 24, up
76 b L(lo3)
77 add n, -3, n
79 L(b2): addcc %g0, %g0, %o1
80 ldx [up+0], %l1
81 ldx [up+8], %l2
82 brgz n, L(gt2)
83 add rp, -16, rp
84 mulx %l1, v0, %o2
85 umulxhi(%l1, v0, %o3)
86 mulx %l2, v0, %o4
87 umulxhi(%l2, v0, %o5)
88 b L(wd2)
89 nop
90 L(gt2): ldx [up+16], %l3
91 mulx %l1, v0, %o2
92 umulxhi(%l1, v0, %o3)
93 ldx [up+24], %l0
94 mulx %l2, v0, %o4
95 umulxhi(%l2, v0, %o5)
96 add up, 16, up
97 b L(lo2)
98 add n, -2, n
100 L(b1): addcc %g0, %g0, %o3
101 ldx [up+0], %l2
102 brgz n, L(gt1)
104 mulx %l2, v0, %o4
105 stx %o4, [rp+0]
106 umulxhi(%l2, v0, %i0)
108 restore
109 L(gt1): ldx [up+8], %l3
110 ldx [up+16], %l0
111 mulx %l2, v0, %o4
112 umulxhi(%l2, v0, %o5)
113 ldx [up+24], %l1
114 mulx %l3, v0, %i4
115 umulxhi(%l3, v0, %i5)
116 add rp, -24, rp
117 add up, 8, up
118 b L(lo1)
119 add n, -1, n
121 L(b0): addcc %g0, %g0, %o5
122 ldx [up+0], %l3
123 ldx [up+8], %l0
124 ldx [up+16], %l1
125 mulx %l3, v0, %i4
126 umulxhi(%l3, v0, %i5)
127 ldx [up+24], %l2
128 mulx %l0, v0, %o0
129 umulxhi(%l0, v0, %o1)
130 b L(lo0)
133 ALIGN(16)
134 L(top): ldx [up+0], %l3 C 0
135 addxccc(%i4, %o5, %i4) C 0
136 mulx %l1, v0, %o2 C 1
137 stx %i4, [rp+0] C 1
138 umulxhi(%l1, v0, %o3) C 2
139 L(lo3): ldx [up+8], %l0 C 2
140 addxccc(%o0, %i5, %o0) C 3
141 mulx %l2, v0, %o4 C 3
142 stx %o0, [rp+8] C 4
143 umulxhi(%l2, v0, %o5) C 4
144 L(lo2): ldx [up+16], %l1 C 5
145 addxccc(%o2, %o1, %o2) C 5
146 mulx %l3, v0, %i4 C 6
147 stx %o2, [rp+16] C 6
148 umulxhi(%l3, v0, %i5) C 7
149 L(lo1): ldx [up+24], %l2 C 7
150 addxccc(%o4, %o3, %o4) C 8
151 mulx %l0, v0, %o0 C 8
152 stx %o4, [rp+24] C 9
153 umulxhi(%l0, v0, %o1) C 9
154 add rp, 32, rp C 10
155 L(lo0): add up, 32, up C 10
156 brgz n, L(top) C 11
157 add n, -4, n C 11
159 L(end): addxccc(%i4, %o5, %i4)
160 mulx %l1, v0, %o2
161 stx %i4, [rp+0]
162 umulxhi(%l1, v0, %o3)
163 addxccc(%o0, %i5, %o0)
164 L(wd3): mulx %l2, v0, %o4
165 stx %o0, [rp+8]
166 umulxhi(%l2, v0, %o5)
167 addxccc(%o2, %o1, %o2)
168 L(wd2): stx %o2, [rp+16]
169 addxccc(%o4, %o3, %o4)
170 stx %o4, [rp+24]
171 addxc( %g0, %o5, %i0)
173 restore
174 EPILOGUE()