vp3dsp_altivec.c
Go to the documentation of this file.
1 /*
2  * Copyright (C) 2009 David Conrad
3  *
4  * This file is part of Libav.
5  *
6  * Libav 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  * Libav 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 Libav; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 #include "libavcodec/dsputil.h"
22 #include "util_altivec.h"
23 #include "types_altivec.h"
24 #include "dsputil_altivec.h"
25 
26 static const vec_s16 constants =
27  {0, 64277, 60547, 54491, 46341, 36410, 25080, 12785};
28 static const vec_u8 interleave_high =
29  {0, 1, 16, 17, 4, 5, 20, 21, 8, 9, 24, 25, 12, 13, 28, 29};
30 
31 #define IDCT_START \
32  vec_s16 A, B, C, D, Ad, Bd, Cd, Dd, E, F, G, H;\
33  vec_s16 Ed, Gd, Add, Bdd, Fd, Hd;\
34  vec_s16 eight = vec_splat_s16(8);\
35  vec_u16 four = vec_splat_u16(4);\
36 \
37  vec_s16 C1 = vec_splat(constants, 1);\
38  vec_s16 C2 = vec_splat(constants, 2);\
39  vec_s16 C3 = vec_splat(constants, 3);\
40  vec_s16 C4 = vec_splat(constants, 4);\
41  vec_s16 C5 = vec_splat(constants, 5);\
42  vec_s16 C6 = vec_splat(constants, 6);\
43  vec_s16 C7 = vec_splat(constants, 7);\
44 \
45  vec_s16 b0 = vec_ld(0x00, block);\
46  vec_s16 b1 = vec_ld(0x10, block);\
47  vec_s16 b2 = vec_ld(0x20, block);\
48  vec_s16 b3 = vec_ld(0x30, block);\
49  vec_s16 b4 = vec_ld(0x40, block);\
50  vec_s16 b5 = vec_ld(0x50, block);\
51  vec_s16 b6 = vec_ld(0x60, block);\
52  vec_s16 b7 = vec_ld(0x70, block);
53 
54 // these functions do (a*C)>>16
55 // things are tricky because a is signed, but C unsigned.
56 // M15 is used if C fits in 15 bit unsigned (C6,C7)
57 // M16 is used if C requires 16 bits unsigned
58 static inline vec_s16 M15(vec_s16 a, vec_s16 C)
59 {
60  return (vec_s16)vec_perm(vec_mule(a,C), vec_mulo(a,C), interleave_high);
61 }
62 static inline vec_s16 M16(vec_s16 a, vec_s16 C)
63 {
64  return vec_add(a, M15(a, C));
65 }
66 
67 #define IDCT_1D(ADD, SHIFT)\
68  A = vec_add(M16(b1, C1), M15(b7, C7));\
69  B = vec_sub(M15(b1, C7), M16(b7, C1));\
70  C = vec_add(M16(b3, C3), M16(b5, C5));\
71  D = vec_sub(M16(b5, C3), M16(b3, C5));\
72 \
73  Ad = M16(vec_sub(A, C), C4);\
74  Bd = M16(vec_sub(B, D), C4);\
75 \
76  Cd = vec_add(A, C);\
77  Dd = vec_add(B, D);\
78 \
79  E = ADD(M16(vec_add(b0, b4), C4));\
80  F = ADD(M16(vec_sub(b0, b4), C4));\
81 \
82  G = vec_add(M16(b2, C2), M15(b6, C6));\
83  H = vec_sub(M15(b2, C6), M16(b6, C2));\
84 \
85  Ed = vec_sub(E, G);\
86  Gd = vec_add(E, G);\
87 \
88  Add = vec_add(F, Ad);\
89  Bdd = vec_sub(Bd, H);\
90 \
91  Fd = vec_sub(F, Ad);\
92  Hd = vec_add(Bd, H);\
93 \
94  b0 = SHIFT(vec_add(Gd, Cd));\
95  b7 = SHIFT(vec_sub(Gd, Cd));\
96 \
97  b1 = SHIFT(vec_add(Add, Hd));\
98  b2 = SHIFT(vec_sub(Add, Hd));\
99 \
100  b3 = SHIFT(vec_add(Ed, Dd));\
101  b4 = SHIFT(vec_sub(Ed, Dd));\
102 \
103  b5 = SHIFT(vec_add(Fd, Bdd));\
104  b6 = SHIFT(vec_sub(Fd, Bdd));
105 
106 #define NOP(a) a
107 #define ADD8(a) vec_add(a, eight)
108 #define SHIFT4(a) vec_sra(a, four)
109 
111 {
112  IDCT_START
113 
114  IDCT_1D(NOP, NOP)
115  TRANSPOSE8(b0, b1, b2, b3, b4, b5, b6, b7);
117 
118  vec_st(b0, 0x00, block);
119  vec_st(b1, 0x10, block);
120  vec_st(b2, 0x20, block);
121  vec_st(b3, 0x30, block);
122  vec_st(b4, 0x40, block);
123  vec_st(b5, 0x50, block);
124  vec_st(b6, 0x60, block);
125  vec_st(b7, 0x70, block);
126 }
127 
128 void ff_vp3_idct_put_altivec(uint8_t *dst, int stride, DCTELEM block[64])
129 {
130  vec_u8 t;
131  IDCT_START
132 
133  // pixels are signed; so add 128*16 in addition to the normal 8
134  vec_s16 v2048 = vec_sl(vec_splat_s16(1), vec_splat_u16(11));
135  eight = vec_add(eight, v2048);
136 
137  IDCT_1D(NOP, NOP)
138  TRANSPOSE8(b0, b1, b2, b3, b4, b5, b6, b7);
140 
141 #define PUT(a)\
142  t = vec_packsu(a, a);\
143  vec_ste((vec_u32)t, 0, (unsigned int *)dst);\
144  vec_ste((vec_u32)t, 4, (unsigned int *)dst);
145 
146  PUT(b0) dst += stride;
147  PUT(b1) dst += stride;
148  PUT(b2) dst += stride;
149  PUT(b3) dst += stride;
150  PUT(b4) dst += stride;
151  PUT(b5) dst += stride;
152  PUT(b6) dst += stride;
153  PUT(b7)
154 }
155 
156 void ff_vp3_idct_add_altivec(uint8_t *dst, int stride, DCTELEM block[64])
157 {
158  LOAD_ZERO;
159  vec_u8 t, vdst;
160  vec_s16 vdst_16;
161  vec_u8 vdst_mask = vec_mergeh(vec_splat_u8(-1), vec_lvsl(0, dst));
162 
163  IDCT_START
164 
165  IDCT_1D(NOP, NOP)
166  TRANSPOSE8(b0, b1, b2, b3, b4, b5, b6, b7);
168 
169 #define ADD(a)\
170  vdst = vec_ld(0, dst);\
171  vdst_16 = (vec_s16)vec_perm(vdst, zero_u8v, vdst_mask);\
172  vdst_16 = vec_adds(a, vdst_16);\
173  t = vec_packsu(vdst_16, vdst_16);\
174  vec_ste((vec_u32)t, 0, (unsigned int *)dst);\
175  vec_ste((vec_u32)t, 4, (unsigned int *)dst);
176 
177  ADD(b0) dst += stride;
178  ADD(b1) dst += stride;
179  ADD(b2) dst += stride;
180  ADD(b3) dst += stride;
181  ADD(b4) dst += stride;
182  ADD(b5) dst += stride;
183  ADD(b6) dst += stride;
184  ADD(b7)
185 }