aboutsummaryrefslogtreecommitdiffstatshomepage
path: root/linden/indra/libgcrypt/libgcrypt-1.2.2/mpi/hppa1.1/mpih-mul3.S
diff options
context:
space:
mode:
Diffstat (limited to 'linden/indra/libgcrypt/libgcrypt-1.2.2/mpi/hppa1.1/mpih-mul3.S')
-rw-r--r--[-rwxr-xr-x]linden/indra/libgcrypt/libgcrypt-1.2.2/mpi/hppa1.1/mpih-mul3.S250
1 files changed, 125 insertions, 125 deletions
diff --git a/linden/indra/libgcrypt/libgcrypt-1.2.2/mpi/hppa1.1/mpih-mul3.S b/linden/indra/libgcrypt/libgcrypt-1.2.2/mpi/hppa1.1/mpih-mul3.S
index 7daf7f4..c4246e4 100755..100644
--- a/linden/indra/libgcrypt/libgcrypt-1.2.2/mpi/hppa1.1/mpih-mul3.S
+++ b/linden/indra/libgcrypt/libgcrypt-1.2.2/mpi/hppa1.1/mpih-mul3.S
@@ -1,125 +1,125 @@
1/* hppa1.1 submul_1 -- Multiply a limb vector with a limb and add 1/* hppa1.1 submul_1 -- Multiply a limb vector with a limb and add
2 * the result to a second limb vector. 2 * the result to a second limb vector.
3 * 3 *
4 * Copyright (C) 1992, 1993, 1994, 1998, 4 * Copyright (C) 1992, 1993, 1994, 1998,
5 * 2001, 2002 Free Software Foundation, Inc. 5 * 2001, 2002 Free Software Foundation, Inc.
6 * 6 *
7 * This file is part of Libgcrypt. 7 * This file is part of Libgcrypt.
8 * 8 *
9 * Libgcrypt is free software; you can redistribute it and/or modify 9 * Libgcrypt is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU Lesser General Public License as 10 * it under the terms of the GNU Lesser General Public License as
11 * published by the Free Software Foundation; either version 2.1 of 11 * published by the Free Software Foundation; either version 2.1 of
12 * the License, or (at your option) any later version. 12 * the License, or (at your option) any later version.
13 * 13 *
14 * Libgcrypt is distributed in the hope that it will be useful, 14 * Libgcrypt is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of 15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU Lesser General Public License for more details. 17 * GNU Lesser General Public License for more details.
18 * 18 *
19 * You should have received a copy of the GNU Lesser General Public 19 * You should have received a copy of the GNU Lesser General Public
20 * License along with this program; if not, write to the Free Software 20 * License along with this program; if not, write to the Free Software
21 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA 21 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA
22 * 22 *
23 * Note: This code is heavily based on the GNU MP Library. 23 * Note: This code is heavily based on the GNU MP Library.
24 * Actually it's the same code with only minor changes in the 24 * Actually it's the same code with only minor changes in the
25 * way the data is stored; this is to support the abstraction 25 * way the data is stored; this is to support the abstraction
26 * of an optional secure memory allocation which may be used 26 * of an optional secure memory allocation which may be used
27 * to avoid revealing of sensitive data due to paging etc. 27 * to avoid revealing of sensitive data due to paging etc.
28 */ 28 */
29 29
30 30
31/******************* 31/*******************
32 * mpi_limb_t 32 * mpi_limb_t
33 * _gcry_mpih_submul_1( mpi_ptr_t res_ptr, (r26) 33 * _gcry_mpih_submul_1( mpi_ptr_t res_ptr, (r26)
34 * mpi_ptr_t s1_ptr, (r25) 34 * mpi_ptr_t s1_ptr, (r25)
35 * mpi_size_t s1_size, (r24) 35 * mpi_size_t s1_size, (r24)
36 * mpi_limb_t s2_limb) (r23) 36 * mpi_limb_t s2_limb) (r23)
37 * 37 *
38 * 38 *
39 * This runs at 12 cycles/limb on a PA7000. With the used instructions, it 39 * This runs at 12 cycles/limb on a PA7000. With the used instructions, it
40 * can not become faster due to data cache contention after a store. On the 40 * can not become faster due to data cache contention after a store. On the
41 * PA7100 it runs at 11 cycles/limb, and that can not be improved either, 41 * PA7100 it runs at 11 cycles/limb, and that can not be improved either,
42 * since only the xmpyu does not need the integer pipeline, so the only 42 * since only the xmpyu does not need the integer pipeline, so the only
43 * dual-issue we will get are addc+xmpyu. Unrolling could gain a cycle/limb 43 * dual-issue we will get are addc+xmpyu. Unrolling could gain a cycle/limb
44 * on the PA7100. 44 * on the PA7100.
45 * 45 *
46 * There are some ideas described in mul1.S that applies to this code too. 46 * There are some ideas described in mul1.S that applies to this code too.
47 * 47 *
48 * It seems possible to make this run as fast as addmul_1, if we use 48 * It seems possible to make this run as fast as addmul_1, if we use
49 * sub,>>= %r29,%r19,%r22 49 * sub,>>= %r29,%r19,%r22
50 * addi 1,%r28,%r28 50 * addi 1,%r28,%r28
51 * but that requires reworking the hairy software pipeline... 51 * but that requires reworking the hairy software pipeline...
52 */ 52 */
53 53
54 54
55 .code 55 .code
56 .export _gcry_mpih_submul_1 56 .export _gcry_mpih_submul_1
57 .label _gcry_mpih_submul_1 57 .label _gcry_mpih_submul_1
58 .proc 58 .proc
59 .callinfo frame=64,no_calls 59 .callinfo frame=64,no_calls
60 .entry 60 .entry
61 61
62 ldo 64(%r30),%r30 62 ldo 64(%r30),%r30
63 fldws,ma 4(%r25),%fr5 63 fldws,ma 4(%r25),%fr5
64 stw %r23,-16(%r30) ; move s2_limb ... 64 stw %r23,-16(%r30) ; move s2_limb ...
65 addib,= -1,%r24,L$just_one_limb 65 addib,= -1,%r24,L$just_one_limb
66 fldws -16(%r30),%fr4 ; ... into fr4 66 fldws -16(%r30),%fr4 ; ... into fr4
67 add %r0,%r0,%r0 ; clear carry 67 add %r0,%r0,%r0 ; clear carry
68 xmpyu %fr4,%fr5,%fr6 68 xmpyu %fr4,%fr5,%fr6
69 fldws,ma 4(%r25),%fr7 69 fldws,ma 4(%r25),%fr7
70 fstds %fr6,-16(%r30) 70 fstds %fr6,-16(%r30)
71 xmpyu %fr4,%fr7,%fr8 71 xmpyu %fr4,%fr7,%fr8
72 ldw -12(%r30),%r19 ; least significant limb in product 72 ldw -12(%r30),%r19 ; least significant limb in product
73 ldw -16(%r30),%r28 73 ldw -16(%r30),%r28
74 74
75 fstds %fr8,-16(%r30) 75 fstds %fr8,-16(%r30)
76 addib,= -1,%r24,L$end 76 addib,= -1,%r24,L$end
77 ldw -12(%r30),%r1 77 ldw -12(%r30),%r1
78 78
79; Main loop 79; Main loop
80 .label L$loop 80 .label L$loop
81 ldws 0(%r26),%r29 81 ldws 0(%r26),%r29
82 fldws,ma 4(%r25),%fr5 82 fldws,ma 4(%r25),%fr5
83 sub %r29,%r19,%r22 83 sub %r29,%r19,%r22
84 add %r22,%r19,%r0 84 add %r22,%r19,%r0
85 stws,ma %r22,4(%r26) 85 stws,ma %r22,4(%r26)
86 addc %r28,%r1,%r19 86 addc %r28,%r1,%r19
87 xmpyu %fr4,%fr5,%fr6 87 xmpyu %fr4,%fr5,%fr6
88 ldw -16(%r30),%r28 88 ldw -16(%r30),%r28
89 fstds %fr6,-16(%r30) 89 fstds %fr6,-16(%r30)
90 addc %r0,%r28,%r28 90 addc %r0,%r28,%r28
91 addib,<> -1,%r24,L$loop 91 addib,<> -1,%r24,L$loop
92 ldw -12(%r30),%r1 92 ldw -12(%r30),%r1
93 93
94 .label L$end 94 .label L$end
95 ldw 0(%r26),%r29 95 ldw 0(%r26),%r29
96 sub %r29,%r19,%r22 96 sub %r29,%r19,%r22
97 add %r22,%r19,%r0 97 add %r22,%r19,%r0
98 stws,ma %r22,4(%r26) 98 stws,ma %r22,4(%r26)
99 addc %r28,%r1,%r19 99 addc %r28,%r1,%r19
100 ldw -16(%r30),%r28 100 ldw -16(%r30),%r28
101 ldws 0(%r26),%r29 101 ldws 0(%r26),%r29
102 addc %r0,%r28,%r28 102 addc %r0,%r28,%r28
103 sub %r29,%r19,%r22 103 sub %r29,%r19,%r22
104 add %r22,%r19,%r0 104 add %r22,%r19,%r0
105 stws,ma %r22,4(%r26) 105 stws,ma %r22,4(%r26)
106 addc %r0,%r28,%r28 106 addc %r0,%r28,%r28
107 bv 0(%r2) 107 bv 0(%r2)
108 ldo -64(%r30),%r30 108 ldo -64(%r30),%r30
109 109
110 .label L$just_one_limb 110 .label L$just_one_limb
111 xmpyu %fr4,%fr5,%fr6 111 xmpyu %fr4,%fr5,%fr6
112 ldw 0(%r26),%r29 112 ldw 0(%r26),%r29
113 fstds %fr6,-16(%r30) 113 fstds %fr6,-16(%r30)
114 ldw -12(%r30),%r1 114 ldw -12(%r30),%r1
115 ldw -16(%r30),%r28 115 ldw -16(%r30),%r28
116 sub %r29,%r1,%r22 116 sub %r29,%r1,%r22
117 add %r22,%r1,%r0 117 add %r22,%r1,%r0
118 stw %r22,0(%r26) 118 stw %r22,0(%r26)
119 addc %r0,%r28,%r28 119 addc %r0,%r28,%r28
120 bv 0(%r2) 120 bv 0(%r2)
121 ldo -64(%r30),%r30 121 ldo -64(%r30),%r30
122 122
123 .exit 123 .exit
124 .procend 124 .procend
125 125
generated by cgit v1.1 at 2025-02-15 08:54:43 +0000