Mercurial > libavcodec.hg
annotate ppc/float_altivec.c @ 8579:dc7bfff68156 libavcodec
add the necessary include for LIBAV*_VERSION_*
author | aurel |
---|---|
date | Tue, 13 Jan 2009 00:14:43 +0000 |
parents | 5a923bd4f5c6 |
children | dd2b5e52336a |
rev | line source |
---|---|
3581 | 1 /* |
2 * Copyright (c) 2006 Luca Barbato <lu_zero@gentoo.org> | |
3 * | |
3947
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3659
diff
changeset
|
4 * This file is part of FFmpeg. |
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3659
diff
changeset
|
5 * |
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3659
diff
changeset
|
6 * FFmpeg is free software; you can redistribute it and/or |
3581 | 7 * modify it under the terms of the GNU Lesser General Public |
8 * License as published by the Free Software Foundation; either | |
3947
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3659
diff
changeset
|
9 * version 2.1 of the License, or (at your option) any later version. |
3581 | 10 * |
3947
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3659
diff
changeset
|
11 * FFmpeg is distributed in the hope that it will be useful, |
3581 | 12 * but WITHOUT ANY WARRANTY; without even the implied warranty of |
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
14 * Lesser General Public License for more details. | |
15 * | |
16 * You should have received a copy of the GNU Lesser General Public | |
3947
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3659
diff
changeset
|
17 * License along with FFmpeg; if not, write to the Free Software |
3581 | 18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
19 */ | |
20 | |
6763 | 21 #include "libavcodec/dsputil.h" |
3581 | 22 |
23 #include "gcc_fixes.h" | |
24 | |
25 #include "dsputil_altivec.h" | |
8542 | 26 #include "util_altivec.h" |
3581 | 27 |
28 static void vector_fmul_altivec(float *dst, const float *src, int len) | |
29 { | |
30 int i; | |
31 vector float d0, d1, s, zero = (vector float)vec_splat_u32(0); | |
32 for(i=0; i<len-7; i+=8) { | |
33 d0 = vec_ld(0, dst+i); | |
34 s = vec_ld(0, src+i); | |
35 d1 = vec_ld(16, dst+i); | |
36 d0 = vec_madd(d0, s, zero); | |
37 d1 = vec_madd(d1, vec_ld(16,src+i), zero); | |
38 vec_st(d0, 0, dst+i); | |
39 vec_st(d1, 16, dst+i); | |
40 } | |
41 } | |
42 | |
43 static void vector_fmul_reverse_altivec(float *dst, const float *src0, | |
44 const float *src1, int len) | |
45 { | |
46 int i; | |
47 vector float d, s0, s1, h0, l0, | |
48 s2, s3, zero = (vector float)vec_splat_u32(0); | |
49 src1 += len-4; | |
50 for(i=0; i<len-7; i+=8) { | |
51 s1 = vec_ld(0, src1-i); // [a,b,c,d] | |
52 s0 = vec_ld(0, src0+i); | |
53 l0 = vec_mergel(s1, s1); // [c,c,d,d] | |
54 s3 = vec_ld(-16, src1-i); | |
55 h0 = vec_mergeh(s1, s1); // [a,a,b,b] | |
56 s2 = vec_ld(16, src0+i); | |
57 s1 = vec_mergeh(vec_mergel(l0,h0), // [d,b,d,b] | |
58 vec_mergeh(l0,h0)); // [c,a,c,a] | |
59 // [d,c,b,a] | |
60 l0 = vec_mergel(s3, s3); | |
61 d = vec_madd(s0, s1, zero); | |
62 h0 = vec_mergeh(s3, s3); | |
63 vec_st(d, 0, dst+i); | |
64 s3 = vec_mergeh(vec_mergel(l0,h0), | |
65 vec_mergeh(l0,h0)); | |
66 d = vec_madd(s2, s3, zero); | |
67 vec_st(d, 16, dst+i); | |
68 } | |
69 } | |
70 | |
71 static void vector_fmul_add_add_altivec(float *dst, const float *src0, | |
72 const float *src1, const float *src2, | |
73 int src3, int len, int step) | |
74 { | |
75 int i; | |
76 vector float d, s0, s1, s2, t0, t1, edges; | |
77 vector unsigned char align = vec_lvsr(0,dst), | |
78 mask = vec_lvsl(0, dst); | |
79 | |
80 #if 0 //FIXME: there is still something wrong | |
81 if (step == 2) { | |
82 int y; | |
83 vector float d0, d1, s3, t2; | |
84 vector unsigned int sel = | |
85 vec_mergeh(vec_splat_u32(-1), vec_splat_u32(0)); | |
86 t1 = vec_ld(16, dst); | |
87 for (i=0,y=0; i<len-3; i+=4,y+=8) { | |
88 | |
89 s0 = vec_ld(0,src0+i); | |
90 s1 = vec_ld(0,src1+i); | |
91 s2 = vec_ld(0,src2+i); | |
92 | |
93 // t0 = vec_ld(0, dst+y); //[x x x|a] | |
94 // t1 = vec_ld(16, dst+y); //[b c d|e] | |
95 t2 = vec_ld(31, dst+y); //[f g h|x] | |
96 | |
97 d = vec_madd(s0,s1,s2); // [A B C D] | |
98 | |
99 // [A A B B] | |
100 | |
101 // [C C D D] | |
102 | |
103 d0 = vec_perm(t0, t1, mask); // [a b c d] | |
104 | |
105 d0 = vec_sel(vec_mergeh(d, d), d0, sel); // [A b B d] | |
106 | |
107 edges = vec_perm(t1, t0, mask); | |
108 | |
109 t0 = vec_perm(edges, d0, align); // [x x x|A] | |
110 | |
111 t1 = vec_perm(d0, edges, align); // [b B d|e] | |
112 | |
113 vec_stl(t0, 0, dst+y); | |
114 | |
115 d1 = vec_perm(t1, t2, mask); // [e f g h] | |
116 | |
117 d1 = vec_sel(vec_mergel(d, d), d1, sel); // [C f D h] | |
118 | |
119 edges = vec_perm(t2, t1, mask); | |
120 | |
121 t1 = vec_perm(edges, d1, align); // [b B d|C] | |
122 | |
123 t2 = vec_perm(d1, edges, align); // [f D h|x] | |
124 | |
125 vec_stl(t1, 16, dst+y); | |
126 | |
127 t0 = t1; | |
128 | |
129 vec_stl(t2, 31, dst+y); | |
130 | |
131 t1 = t2; | |
132 } | |
133 } else | |
134 #endif | |
135 if (step == 1 && src3 == 0) | |
136 for (i=0; i<len-3; i+=4) { | |
4387 | 137 t0 = vec_ld(0, dst+i); |
3581 | 138 t1 = vec_ld(15, dst+i); |
139 s0 = vec_ld(0, src0+i); | |
140 s1 = vec_ld(0, src1+i); | |
141 s2 = vec_ld(0, src2+i); | |
142 edges = vec_perm(t1 ,t0, mask); | |
143 d = vec_madd(s0,s1,s2); | |
3659
dd55fb216497
Proper fix for the corner case that would have been corrected before, praise&blame to me and exg in equal shares
lu_zero
parents:
3657
diff
changeset
|
144 t1 = vec_perm(d, edges, align); |
dd55fb216497
Proper fix for the corner case that would have been corrected before, praise&blame to me and exg in equal shares
lu_zero
parents:
3657
diff
changeset
|
145 t0 = vec_perm(edges, d, align); |
dd55fb216497
Proper fix for the corner case that would have been corrected before, praise&blame to me and exg in equal shares
lu_zero
parents:
3657
diff
changeset
|
146 vec_st(t1, 15, dst+i); |
3581 | 147 vec_st(t0, 0, dst+i); |
148 } | |
149 else | |
150 ff_vector_fmul_add_add_c(dst, src0, src1, src2, src3, len, step); | |
151 } | |
152 | |
8542 | 153 static void vector_fmul_window_altivec(float *dst, const float *src0, const float *src1, const float *win, float add_bias, int len) |
154 { | |
155 union { | |
156 vector float v; | |
157 float s[4]; | |
158 } vadd; | |
159 vector float vadd_bias, zero, t0, t1, s0, s1, wi, wj; | |
160 const vector unsigned char reverse = vcprm(3,2,1,0); | |
161 int i,j; | |
162 | |
163 dst += len; | |
164 win += len; | |
165 src0+= len; | |
166 | |
167 vadd.s[0] = add_bias; | |
168 vadd_bias = vec_splat(vadd.v, 0); | |
169 zero = (vector float)vec_splat_u32(0); | |
170 | |
171 for(i=-len*4, j=len*4-16; i<0; i+=16, j-=16) { | |
172 s0 = vec_ld(i, src0); | |
173 s1 = vec_ld(j, src1); | |
174 wi = vec_ld(i, win); | |
175 wj = vec_ld(j, win); | |
176 | |
177 s1 = vec_perm(s1, s1, reverse); | |
178 wj = vec_perm(wj, wj, reverse); | |
179 | |
180 t0 = vec_madd(s0, wj, vadd_bias); | |
181 t0 = vec_nmsub(s1, wi, t0); | |
182 t1 = vec_madd(s0, wi, vadd_bias); | |
183 t1 = vec_madd(s1, wj, t1); | |
184 t1 = vec_perm(t1, t1, reverse); | |
185 | |
186 vec_st(t0, i, dst); | |
187 vec_st(t1, j, dst); | |
188 } | |
189 } | |
7674 | 190 |
8365
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
191 static void int32_to_float_fmul_scalar_altivec(float *dst, const int *src, float mul, int len) |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
192 { |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
193 union { |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
194 vector float v; |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
195 float s[4]; |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
196 } mul_u; |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
197 int i; |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
198 vector float src1, src2, dst1, dst2, mul_v, zero; |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
199 |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
200 zero = (vector float)vec_splat_u32(0); |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
201 mul_u.s[0] = mul; |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
202 mul_v = vec_splat(mul_u.v, 0); |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
203 |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
204 for(i=0; i<len; i+=8) { |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
205 src1 = vec_ctf(vec_ld(0, src+i), 0); |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
206 src2 = vec_ctf(vec_ld(16, src+i), 0); |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
207 dst1 = vec_madd(src1, mul_v, zero); |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
208 dst2 = vec_madd(src2, mul_v, zero); |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
209 vec_st(dst1, 0, dst+i); |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
210 vec_st(dst2, 16, dst+i); |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
211 } |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
212 } |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
213 |
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
214 |
7674 | 215 static vector signed short |
216 float_to_int16_one_altivec(const float *src) | |
217 { | |
218 vector float s0 = vec_ld(0, src); | |
219 vector float s1 = vec_ld(16, src); | |
220 vector signed int t0 = vec_cts(s0, 0); | |
221 vector signed int t1 = vec_cts(s1, 0); | |
222 return vec_packs(t0,t1); | |
223 } | |
224 | |
8537
93a3020d9636
Fix float_to_int16_altivec prototype to match float_to_int16's in dsputil.h
gpoirier
parents:
8365
diff
changeset
|
225 static void float_to_int16_altivec(int16_t *dst, const float *src, long len) |
3581 | 226 { |
227 int i; | |
228 vector signed short d0, d1, d; | |
229 vector unsigned char align; | |
230 if(((long)dst)&15) //FIXME | |
231 for(i=0; i<len-7; i+=8) { | |
232 d0 = vec_ld(0, dst+i); | |
7674 | 233 d = float_to_int16_one_altivec(src+i); |
3581 | 234 d1 = vec_ld(15, dst+i); |
235 d1 = vec_perm(d1, d0, vec_lvsl(0,dst+i)); | |
236 align = vec_lvsr(0, dst+i); | |
3657
ff6720290478
Fix float_to_int16, unaligned case, broken by the previous commit
lu_zero
parents:
3583
diff
changeset
|
237 d0 = vec_perm(d1, d, align); |
ff6720290478
Fix float_to_int16, unaligned case, broken by the previous commit
lu_zero
parents:
3583
diff
changeset
|
238 d1 = vec_perm(d, d1, align); |
3581 | 239 vec_st(d0, 0, dst+i); |
240 vec_st(d1,15, dst+i); | |
241 } | |
242 else | |
243 for(i=0; i<len-7; i+=8) { | |
7674 | 244 d = float_to_int16_one_altivec(src+i); |
3581 | 245 vec_st(d, 0, dst+i); |
246 } | |
247 } | |
248 | |
7675
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
249 static void |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
250 float_to_int16_interleave_altivec(int16_t *dst, const float **src, |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
251 long len, int channels) |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
252 { |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
253 int i; |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
254 vector signed short d0, d1, d2, c0, c1, t0, t1; |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
255 vector unsigned char align; |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
256 if(channels == 1) |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
257 float_to_int16_altivec(dst, src[0], len); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
258 else |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
259 if (channels == 2) { |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
260 if(((long)dst)&15) |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
261 for(i=0; i<len-7; i+=8) { |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
262 d0 = vec_ld(0, dst + i); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
263 t0 = float_to_int16_one_altivec(src[0] + i); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
264 d1 = vec_ld(31, dst + i); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
265 t1 = float_to_int16_one_altivec(src[1] + i); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
266 c0 = vec_mergeh(t0, t1); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
267 c1 = vec_mergel(t0, t1); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
268 d2 = vec_perm(d1, d0, vec_lvsl(0, dst + i)); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
269 align = vec_lvsr(0, dst + i); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
270 d0 = vec_perm(d2, c0, align); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
271 d1 = vec_perm(c0, c1, align); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
272 vec_st(d0, 0, dst + i); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
273 d0 = vec_perm(c1, d2, align); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
274 vec_st(d1, 15, dst + i); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
275 vec_st(d0, 31, dst + i); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
276 dst+=8; |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
277 } |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
278 else |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
279 for(i=0; i<len-7; i+=8) { |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
280 t0 = float_to_int16_one_altivec(src[0] + i); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
281 t1 = float_to_int16_one_altivec(src[1] + i); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
282 d0 = vec_mergeh(t0, t1); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
283 d1 = vec_mergel(t0, t1); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
284 vec_st(d0, 0, dst + i); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
285 vec_st(d1, 16, dst + i); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
286 dst+=8; |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
287 } |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
288 } else { |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
289 DECLARE_ALIGNED(16, int16_t, tmp[len]); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
290 int c, j; |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
291 for (c = 0; c < channels; c++) { |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
292 float_to_int16_altivec(tmp, src[c], len); |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
293 for (i = 0, j = c; i < len; i++, j+=channels) { |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
294 dst[j] = tmp[i]; |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
295 } |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
296 } |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
297 } |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
298 } |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
299 |
3581 | 300 void float_init_altivec(DSPContext* c, AVCodecContext *avctx) |
301 { | |
302 c->vector_fmul = vector_fmul_altivec; | |
303 c->vector_fmul_reverse = vector_fmul_reverse_altivec; | |
304 c->vector_fmul_add_add = vector_fmul_add_add_altivec; | |
8365
d33b47d1f4c1
add AltiVec implementation of int32_to_float_fmul_scalar
gpoirier
parents:
7691
diff
changeset
|
305 c->int32_to_float_fmul_scalar = int32_to_float_fmul_scalar_altivec; |
7675
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
306 if(!(avctx->flags & CODEC_FLAG_BITEXACT)) { |
8542 | 307 c->vector_fmul_window = vector_fmul_window_altivec; |
3581 | 308 c->float_to_int16 = float_to_int16_altivec; |
7675
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
309 c->float_to_int16_interleave = float_to_int16_interleave_altivec; |
ad4bf45b9b63
Introduce float_to_int16_interleave_altivec, tested with vorbis
lu_zero
parents:
7674
diff
changeset
|
310 } |
3581 | 311 } |