OLD | NEW |
| (Empty) |
1 dnl PowerPC 750 mpn_lshift -- mpn left shift. | |
2 | |
3 dnl Copyright 2002, 2003 Free Software Foundation, Inc. | |
4 | |
5 dnl This file is part of the GNU MP Library. | |
6 | |
7 dnl The GNU MP Library is free software; you can redistribute it and/or modify | |
8 dnl it under the terms of the GNU Lesser General Public License as published | |
9 dnl by the Free Software Foundation; either version 3 of the License, or (at | |
10 dnl your option) any later version. | |
11 | |
12 dnl The GNU MP Library is distributed in the hope that it will be useful, but | |
13 dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY | |
14 dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public | |
15 dnl License for more details. | |
16 | |
17 dnl You should have received a copy of the GNU Lesser General Public License | |
18 dnl along with the GNU MP Library. If not, see http://www.gnu.org/licenses/. | |
19 | |
20 include(`../config.m4') | |
21 | |
22 | |
23 C cycles/limb | |
24 C 750: 3.0 | |
25 C 7400: 3.0 | |
26 | |
27 | |
28 C mp_limb_t mpn_lshift (mp_ptr dst, mp_srcptr src, mp_size_t size, | |
29 C unsigned shift); | |
30 C | |
31 C This code is the same per-limb speed as mpn/powerpc32/lshift.asm, but | |
32 C smaller and saving about 30 or so cycles of overhead. | |
33 | |
34 ASM_START() | |
35 PROLOGUE(mpn_lshift) | |
36 | |
37 C r3 dst | |
38 C r4 src | |
39 C r5 size | |
40 C r6 shift | |
41 | |
42 mtctr r5 C size | |
43 slwi r5, r5, 2 C 4*size | |
44 | |
45 subfic r7, r6, 32 C 32-shift | |
46 add r4, r4, r5 C &src[size] | |
47 | |
48 add r5, r3, r5 C &dst[size] | |
49 lwz r8, -4(r4) C src[size-1] | |
50 bdz L(one) | |
51 | |
52 lwzu r9, -8(r4) C src[size-2] | |
53 | |
54 srw r3, r8, r7 C return value | |
55 slw r8, r8, r6 C src[size-1] << shift | |
56 bdz L(two) | |
57 | |
58 | |
59 L(top): | |
60 C r3 return value | |
61 C r4 src, incrementing | |
62 C r5 dst, incrementing | |
63 C r6 lshift | |
64 C r7 32-shift | |
65 C r8 src[i+1] << shift | |
66 C r9 src[i] | |
67 C r10 | |
68 | |
69 lwzu r10, -4(r4) | |
70 srw r11, r9, r7 | |
71 | |
72 or r8, r8, r11 | |
73 stwu r8, -4(r5) | |
74 | |
75 slw r8, r9, r6 | |
76 bdz L(odd) | |
77 | |
78 C r8 src[i+1] << shift | |
79 C r9 | |
80 C r10 src[i] | |
81 | |
82 lwzu r9, -4(r4) | |
83 srw r11, r10, r7 | |
84 | |
85 or r8, r8, r11 | |
86 stwu r8, -4(r5) | |
87 | |
88 slw r8, r10, r6 | |
89 bdnz L(top) | |
90 | |
91 | |
92 L(two): | |
93 C r3 return value | |
94 C r4 | |
95 C r5 &dst[2] | |
96 C r6 shift | |
97 C r7 32-shift | |
98 C r8 src[1] << shift | |
99 C r9 src[0] | |
100 C r10 | |
101 | |
102 srw r11, r9, r7 | |
103 slw r12, r9, r6 C src[0] << shift | |
104 | |
105 or r8, r8, r11 | |
106 stw r12, -8(r5) C dst[0] | |
107 | |
108 stw r8, -4(r5) C dst[1] | |
109 blr | |
110 | |
111 | |
112 L(odd): | |
113 C r3 return value | |
114 C r4 | |
115 C r5 &dst[2] | |
116 C r6 shift | |
117 C r7 32-shift | |
118 C r8 src[1] << shift | |
119 C r9 | |
120 C r10 src[0] | |
121 | |
122 srw r11, r10, r7 | |
123 slw r12, r10, r6 | |
124 | |
125 or r8, r8, r11 | |
126 stw r12, -8(r5) C dst[0] | |
127 | |
128 stw r8, -4(r5) C dst[1] | |
129 blr | |
130 | |
131 | |
132 L(one): | |
133 C r5 &dst[1] | |
134 C r6 shift | |
135 C r7 32-shift | |
136 C r8 src[0] | |
137 | |
138 srw r3, r8, r7 C return value | |
139 slw r8, r8, r6 C src[size-1] << shift | |
140 | |
141 stw r8, -4(r5) C dst[0] | |
142 blr | |
143 | |
144 EPILOGUE(mpn_lshift) | |
OLD | NEW |