FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
fmtconvert_altivec.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2006 Luca Barbato <lu_zero@gentoo.org>
3  *
4  * This file is part of FFmpeg.
5  *
6  * FFmpeg is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * FFmpeg is distributed in the hope that it will be useful,
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
17  * License along with FFmpeg; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 #include "config.h"
22 #include "libavutil/attributes.h"
23 #include "libavutil/mem.h"
25 #include "libavcodec/fmtconvert.h"
26 #include "dsputil_altivec.h"
27 
28 #if HAVE_ALTIVEC
29 
30 static void int32_to_float_fmul_scalar_altivec(float *dst, const int32_t *src,
31  float mul, int len)
32 {
33  union {
34  vector float v;
35  float s[4];
36  } mul_u;
37  int i;
38  vector float src1, src2, dst1, dst2, mul_v, zero;
39 
40  zero = (vector float)vec_splat_u32(0);
41  mul_u.s[0] = mul;
42  mul_v = vec_splat(mul_u.v, 0);
43 
44  for (i = 0; i < len; i += 8) {
45  src1 = vec_ctf(vec_ld(0, src+i), 0);
46  src2 = vec_ctf(vec_ld(16, src+i), 0);
47  dst1 = vec_madd(src1, mul_v, zero);
48  dst2 = vec_madd(src2, mul_v, zero);
49  vec_st(dst1, 0, dst+i);
50  vec_st(dst2, 16, dst+i);
51  }
52 }
53 
54 
55 static vector signed short float_to_int16_one_altivec(const float *src)
56 {
57  vector float s0 = vec_ld(0, src);
58  vector float s1 = vec_ld(16, src);
59  vector signed int t0 = vec_cts(s0, 0);
60  vector signed int t1 = vec_cts(s1, 0);
61  return vec_packs(t0,t1);
62 }
63 
64 static void float_to_int16_altivec(int16_t *dst, const float *src, long len)
65 {
66  int i;
67  vector signed short d0, d1, d;
68  vector unsigned char align;
69  if (((long)dst) & 15) { //FIXME
70  for (i = 0; i < len - 7; i += 8) {
71  d0 = vec_ld(0, dst+i);
72  d = float_to_int16_one_altivec(src + i);
73  d1 = vec_ld(15, dst+i);
74  d1 = vec_perm(d1, d0, vec_lvsl(0, dst + i));
75  align = vec_lvsr(0, dst + i);
76  d0 = vec_perm(d1, d, align);
77  d1 = vec_perm(d, d1, align);
78  vec_st(d0, 0, dst + i);
79  vec_st(d1, 15, dst + i);
80  }
81  } else {
82  for (i = 0; i < len - 7; i += 8) {
83  d = float_to_int16_one_altivec(src + i);
84  vec_st(d, 0, dst + i);
85  }
86  }
87 }
88 
89 #define VSTE_INC(dst, v, elem, inc) do { \
90  vector signed short s = vec_splat(v, elem); \
91  vec_ste(s, 0, dst); \
92  dst += inc; \
93  } while (0)
94 
95 static void float_to_int16_stride_altivec(int16_t *dst, const float *src,
96  long len, int stride)
97 {
98  int i;
99  vector signed short d;
100 
101  for (i = 0; i < len - 7; i += 8) {
102  d = float_to_int16_one_altivec(src + i);
103  VSTE_INC(dst, d, 0, stride);
104  VSTE_INC(dst, d, 1, stride);
105  VSTE_INC(dst, d, 2, stride);
106  VSTE_INC(dst, d, 3, stride);
107  VSTE_INC(dst, d, 4, stride);
108  VSTE_INC(dst, d, 5, stride);
109  VSTE_INC(dst, d, 6, stride);
110  VSTE_INC(dst, d, 7, stride);
111  }
112 }
113 
114 static void float_to_int16_interleave_altivec(int16_t *dst, const float **src,
115  long len, int channels)
116 {
117  int i;
118  vector signed short d0, d1, d2, c0, c1, t0, t1;
119  vector unsigned char align;
120 
121  if (channels == 1)
122  float_to_int16_altivec(dst, src[0], len);
123  else {
124  if (channels == 2) {
125  if (((long)dst) & 15) {
126  for (i = 0; i < len - 7; i += 8) {
127  d0 = vec_ld(0, dst + i);
128  t0 = float_to_int16_one_altivec(src[0] + i);
129  d1 = vec_ld(31, dst + i);
130  t1 = float_to_int16_one_altivec(src[1] + i);
131  c0 = vec_mergeh(t0, t1);
132  c1 = vec_mergel(t0, t1);
133  d2 = vec_perm(d1, d0, vec_lvsl(0, dst + i));
134  align = vec_lvsr(0, dst + i);
135  d0 = vec_perm(d2, c0, align);
136  d1 = vec_perm(c0, c1, align);
137  vec_st(d0, 0, dst + i);
138  d0 = vec_perm(c1, d2, align);
139  vec_st(d1, 15, dst + i);
140  vec_st(d0, 31, dst + i);
141  dst += 8;
142  }
143  } else {
144  for (i = 0; i < len - 7; i += 8) {
145  t0 = float_to_int16_one_altivec(src[0] + i);
146  t1 = float_to_int16_one_altivec(src[1] + i);
147  d0 = vec_mergeh(t0, t1);
148  d1 = vec_mergel(t0, t1);
149  vec_st(d0, 0, dst + i);
150  vec_st(d1, 16, dst + i);
151  dst += 8;
152  }
153  }
154  } else {
155  for (i = 0; i < channels; i++)
156  float_to_int16_stride_altivec(dst + i, src[i], len, channels);
157  }
158  }
159 }
160 
161 #endif /* HAVE_ALTIVEC */
162 
164  AVCodecContext *avctx)
165 {
166 #if HAVE_ALTIVEC
167  c->int32_to_float_fmul_scalar = int32_to_float_fmul_scalar_altivec;
168  if (!(avctx->flags & CODEC_FLAG_BITEXACT)) {
169  c->float_to_int16 = float_to_int16_altivec;
170  c->float_to_int16_interleave = float_to_int16_interleave_altivec;
171  }
172 #endif /* HAVE_ALTIVEC */
173 }