diff options
Diffstat (limited to 'lib/rbcodec/codecs/demac/libdemac/vector_math32_armv4.h')
-rw-r--r-- | lib/rbcodec/codecs/demac/libdemac/vector_math32_armv4.h | 201 |
1 files changed, 201 insertions, 0 deletions
diff --git a/lib/rbcodec/codecs/demac/libdemac/vector_math32_armv4.h b/lib/rbcodec/codecs/demac/libdemac/vector_math32_armv4.h new file mode 100644 index 0000000000..d6bb9b0d9c --- /dev/null +++ b/lib/rbcodec/codecs/demac/libdemac/vector_math32_armv4.h | |||
@@ -0,0 +1,201 @@ | |||
1 | /* | ||
2 | |||
3 | libdemac - A Monkey's Audio decoder | ||
4 | |||
5 | $Id$ | ||
6 | |||
7 | Copyright (C) Dave Chapman 2007 | ||
8 | |||
9 | ARMv4 vector math copyright (C) 2008 Jens Arnold | ||
10 | |||
11 | This program is free software; you can redistribute it and/or modify | ||
12 | it under the terms of the GNU General Public License as published by | ||
13 | the Free Software Foundation; either version 2 of the License, or | ||
14 | (at your option) any later version. | ||
15 | |||
16 | This program is distributed in the hope that it will be useful, | ||
17 | but WITHOUT ANY WARRANTY; without even the implied warranty of | ||
18 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | ||
19 | GNU General Public License for more details. | ||
20 | |||
21 | You should have received a copy of the GNU General Public License | ||
22 | along with this program; if not, write to the Free Software | ||
23 | Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110, USA | ||
24 | |||
25 | */ | ||
26 | |||
27 | #define FUSED_VECTOR_MATH | ||
28 | |||
29 | #if ORDER > 32 | ||
30 | #define REPEAT_BLOCK(x) x x x x x x x x | ||
31 | #elif ORDER > 16 | ||
32 | #define REPEAT_BLOCK(x) x x x x x x x | ||
33 | #else | ||
34 | #define REPEAT_BLOCK(x) x x x | ||
35 | #endif | ||
36 | |||
37 | /* Calculate scalarproduct, then add a 2nd vector (fused for performance) */ | ||
38 | static inline int32_t vector_sp_add(int32_t* v1, int32_t* f2, int32_t* s2) | ||
39 | { | ||
40 | int res; | ||
41 | #if ORDER > 32 | ||
42 | int cnt = ORDER>>5; | ||
43 | #endif | ||
44 | |||
45 | asm volatile ( | ||
46 | #if ORDER > 32 | ||
47 | "mov %[res], #0 \n" | ||
48 | "1: \n" | ||
49 | #else | ||
50 | "ldmia %[v1], {r0-r3} \n" | ||
51 | "ldmia %[f2]!, {r4-r7} \n" | ||
52 | "mul %[res], r4, r0 \n" | ||
53 | "mla %[res], r5, r1, %[res] \n" | ||
54 | "mla %[res], r6, r2, %[res] \n" | ||
55 | "mla %[res], r7, r3, %[res] \n" | ||
56 | "ldmia %[s2]!, {r4-r7} \n" | ||
57 | "add r0, r0, r4 \n" | ||
58 | "add r1, r1, r5 \n" | ||
59 | "add r2, r2, r6 \n" | ||
60 | "add r3, r3, r7 \n" | ||
61 | "stmia %[v1]!, {r0-r3} \n" | ||
62 | #endif | ||
63 | REPEAT_BLOCK( | ||
64 | "ldmia %[v1], {r0-r3} \n" | ||
65 | "ldmia %[f2]!, {r4-r7} \n" | ||
66 | "mla %[res], r4, r0, %[res] \n" | ||
67 | "mla %[res], r5, r1, %[res] \n" | ||
68 | "mla %[res], r6, r2, %[res] \n" | ||
69 | "mla %[res], r7, r3, %[res] \n" | ||
70 | "ldmia %[s2]!, {r4-r7} \n" | ||
71 | "add r0, r0, r4 \n" | ||
72 | "add r1, r1, r5 \n" | ||
73 | "add r2, r2, r6 \n" | ||
74 | "add r3, r3, r7 \n" | ||
75 | "stmia %[v1]!, {r0-r3} \n" | ||
76 | ) | ||
77 | #if ORDER > 32 | ||
78 | "subs %[cnt], %[cnt], #1 \n" | ||
79 | "bne 1b \n" | ||
80 | #endif | ||
81 | : /* outputs */ | ||
82 | #if ORDER > 32 | ||
83 | [cnt]"+r"(cnt), | ||
84 | #endif | ||
85 | [v1] "+r"(v1), | ||
86 | [f2] "+r"(f2), | ||
87 | [s2] "+r"(s2), | ||
88 | [res]"=r"(res) | ||
89 | : /* inputs */ | ||
90 | : /* clobbers */ | ||
91 | "r0", "r1", "r2", "r3", "r4", | ||
92 | "r5", "r6", "r7", "cc", "memory" | ||
93 | ); | ||
94 | return res; | ||
95 | } | ||
96 | |||
97 | /* Calculate scalarproduct, then subtract a 2nd vector (fused for performance) */ | ||
98 | static inline int32_t vector_sp_sub(int32_t* v1, int32_t* f2, int32_t* s2) | ||
99 | { | ||
100 | int res; | ||
101 | #if ORDER > 32 | ||
102 | int cnt = ORDER>>5; | ||
103 | #endif | ||
104 | |||
105 | asm volatile ( | ||
106 | #if ORDER > 32 | ||
107 | "mov %[res], #0 \n" | ||
108 | "1: \n" | ||
109 | #else | ||
110 | "ldmia %[v1], {r0-r3} \n" | ||
111 | "ldmia %[f2]!, {r4-r7} \n" | ||
112 | "mul %[res], r4, r0 \n" | ||
113 | "mla %[res], r5, r1, %[res] \n" | ||
114 | "mla %[res], r6, r2, %[res] \n" | ||
115 | "mla %[res], r7, r3, %[res] \n" | ||
116 | "ldmia %[s2]!, {r4-r7} \n" | ||
117 | "sub r0, r0, r4 \n" | ||
118 | "sub r1, r1, r5 \n" | ||
119 | "sub r2, r2, r6 \n" | ||
120 | "sub r3, r3, r7 \n" | ||
121 | "stmia %[v1]!, {r0-r3} \n" | ||
122 | #endif | ||
123 | REPEAT_BLOCK( | ||
124 | "ldmia %[v1], {r0-r3} \n" | ||
125 | "ldmia %[f2]!, {r4-r7} \n" | ||
126 | "mla %[res], r4, r0, %[res] \n" | ||
127 | "mla %[res], r5, r1, %[res] \n" | ||
128 | "mla %[res], r6, r2, %[res] \n" | ||
129 | "mla %[res], r7, r3, %[res] \n" | ||
130 | "ldmia %[s2]!, {r4-r7} \n" | ||
131 | "sub r0, r0, r4 \n" | ||
132 | "sub r1, r1, r5 \n" | ||
133 | "sub r2, r2, r6 \n" | ||
134 | "sub r3, r3, r7 \n" | ||
135 | "stmia %[v1]!, {r0-r3} \n" | ||
136 | ) | ||
137 | #if ORDER > 32 | ||
138 | "subs %[cnt], %[cnt], #1 \n" | ||
139 | "bne 1b \n" | ||
140 | #endif | ||
141 | : /* outputs */ | ||
142 | #if ORDER > 32 | ||
143 | [cnt]"+r"(cnt), | ||
144 | #endif | ||
145 | [v1] "+r"(v1), | ||
146 | [f2] "+r"(f2), | ||
147 | [s2] "+r"(s2), | ||
148 | [res]"=r"(res) | ||
149 | : /* inputs */ | ||
150 | : /* clobbers */ | ||
151 | "r0", "r1", "r2", "r3", "r4", | ||
152 | "r5", "r6", "r7", "cc", "memory" | ||
153 | ); | ||
154 | return res; | ||
155 | } | ||
156 | |||
157 | static inline int32_t scalarproduct(int32_t* v1, int32_t* v2) | ||
158 | { | ||
159 | int res; | ||
160 | #if ORDER > 32 | ||
161 | int cnt = ORDER>>5; | ||
162 | #endif | ||
163 | |||
164 | asm volatile ( | ||
165 | #if ORDER > 32 | ||
166 | "mov %[res], #0 \n" | ||
167 | "1: \n" | ||
168 | #else | ||
169 | "ldmia %[v1]!, {r0-r3} \n" | ||
170 | "ldmia %[v2]!, {r4-r7} \n" | ||
171 | "mul %[res], r4, r0 \n" | ||
172 | "mla %[res], r5, r1, %[res] \n" | ||
173 | "mla %[res], r6, r2, %[res] \n" | ||
174 | "mla %[res], r7, r3, %[res] \n" | ||
175 | #endif | ||
176 | REPEAT_BLOCK( | ||
177 | "ldmia %[v1]!, {r0-r3} \n" | ||
178 | "ldmia %[v2]!, {r4-r7} \n" | ||
179 | "mla %[res], r4, r0, %[res] \n" | ||
180 | "mla %[res], r5, r1, %[res] \n" | ||
181 | "mla %[res], r6, r2, %[res] \n" | ||
182 | "mla %[res], r7, r3, %[res] \n" | ||
183 | ) | ||
184 | #if ORDER > 32 | ||
185 | "subs %[cnt], %[cnt], #1 \n" | ||
186 | "bne 1b \n" | ||
187 | #endif | ||
188 | : /* outputs */ | ||
189 | #if ORDER > 32 | ||
190 | [cnt]"+r"(cnt), | ||
191 | #endif | ||
192 | [v1] "+r"(v1), | ||
193 | [v2] "+r"(v2), | ||
194 | [res]"=r"(res) | ||
195 | : /* inputs */ | ||
196 | : /* clobbers */ | ||
197 | "r0", "r1", "r2", "r3", | ||
198 | "r4", "r5", "r6", "r7", "cc", "memory" | ||
199 | ); | ||
200 | return res; | ||
201 | } | ||