1/*
2 * Copyright (c) 2004 Romain Dolbeau <romain@dolbeau.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
23#include "libavutil/attributes.h"
24#include "libavutil/cpu.h"
25#include "libavutil/intreadwrite.h"
26#include "libavutil/mem_internal.h"
27#include "libavutil/ppc/cpu.h"
28#include "libavutil/ppc/util_altivec.h"
29
30#include "libavcodec/h264qpel.h"
31
32#include "hpeldsp_altivec.h"
33
34#if HAVE_ALTIVEC
35
36#define PUT_OP_U8_ALTIVEC(d, s, dst) d = s
37#define AVG_OP_U8_ALTIVEC(d, s, dst) d = vec_avg(dst, s)
38
39#define OP_U8_ALTIVEC                          PUT_OP_U8_ALTIVEC
40#define PREFIX_h264_qpel16_h_lowpass_altivec   put_h264_qpel16_h_lowpass_altivec
41#define PREFIX_h264_qpel16_h_lowpass_num       altivec_put_h264_qpel16_h_lowpass_num
42#define PREFIX_h264_qpel16_v_lowpass_altivec   put_h264_qpel16_v_lowpass_altivec
43#define PREFIX_h264_qpel16_v_lowpass_num       altivec_put_h264_qpel16_v_lowpass_num
44#define PREFIX_h264_qpel16_hv_lowpass_altivec  put_h264_qpel16_hv_lowpass_altivec
45#define PREFIX_h264_qpel16_hv_lowpass_num      altivec_put_h264_qpel16_hv_lowpass_num
46#include "h264qpel_template.c"
47#undef OP_U8_ALTIVEC
48#undef PREFIX_h264_qpel16_h_lowpass_altivec
49#undef PREFIX_h264_qpel16_h_lowpass_num
50#undef PREFIX_h264_qpel16_v_lowpass_altivec
51#undef PREFIX_h264_qpel16_v_lowpass_num
52#undef PREFIX_h264_qpel16_hv_lowpass_altivec
53#undef PREFIX_h264_qpel16_hv_lowpass_num
54
55#define OP_U8_ALTIVEC                          AVG_OP_U8_ALTIVEC
56#define PREFIX_h264_qpel16_h_lowpass_altivec   avg_h264_qpel16_h_lowpass_altivec
57#define PREFIX_h264_qpel16_h_lowpass_num       altivec_avg_h264_qpel16_h_lowpass_num
58#define PREFIX_h264_qpel16_v_lowpass_altivec   avg_h264_qpel16_v_lowpass_altivec
59#define PREFIX_h264_qpel16_v_lowpass_num       altivec_avg_h264_qpel16_v_lowpass_num
60#define PREFIX_h264_qpel16_hv_lowpass_altivec  avg_h264_qpel16_hv_lowpass_altivec
61#define PREFIX_h264_qpel16_hv_lowpass_num      altivec_avg_h264_qpel16_hv_lowpass_num
62#include "h264qpel_template.c"
63#undef OP_U8_ALTIVEC
64#undef PREFIX_h264_qpel16_h_lowpass_altivec
65#undef PREFIX_h264_qpel16_h_lowpass_num
66#undef PREFIX_h264_qpel16_v_lowpass_altivec
67#undef PREFIX_h264_qpel16_v_lowpass_num
68#undef PREFIX_h264_qpel16_hv_lowpass_altivec
69#undef PREFIX_h264_qpel16_hv_lowpass_num
70
71#define H264_MC(OPNAME, SIZE, CODETYPE) \
72static void OPNAME ## h264_qpel ## SIZE ## _mc00_ ## CODETYPE (uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
73{\
74    ff_ ## OPNAME ## pixels ## SIZE ## _ ## CODETYPE(dst, src, stride, SIZE);\
75}\
76\
77static void OPNAME ## h264_qpel ## SIZE ## _mc10_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
78{ \
79    DECLARE_ALIGNED(16, uint8_t, half)[SIZE*SIZE];\
80    put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(half, src, SIZE, stride);\
81    OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, src, half, stride, stride, SIZE);\
82}\
83\
84static void OPNAME ## h264_qpel ## SIZE ## _mc20_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
85{\
86    OPNAME ## h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(dst, src, stride, stride);\
87}\
88\
89static void OPNAME ## h264_qpel ## SIZE ## _mc30_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
90{\
91    DECLARE_ALIGNED(16, uint8_t, half)[SIZE*SIZE];\
92    put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(half, src, SIZE, stride);\
93    OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, src+1, half, stride, stride, SIZE);\
94}\
95\
96static void OPNAME ## h264_qpel ## SIZE ## _mc01_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
97{\
98    DECLARE_ALIGNED(16, uint8_t, half)[SIZE*SIZE];\
99    put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(half, src, SIZE, stride);\
100    OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, src, half, stride, stride, SIZE);\
101}\
102\
103static void OPNAME ## h264_qpel ## SIZE ## _mc02_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
104{\
105    OPNAME ## h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(dst, src, stride, stride);\
106}\
107\
108static void OPNAME ## h264_qpel ## SIZE ## _mc03_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
109{\
110    DECLARE_ALIGNED(16, uint8_t, half)[SIZE*SIZE];\
111    put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(half, src, SIZE, stride);\
112    OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, src+stride, half, stride, stride, SIZE);\
113}\
114\
115static void OPNAME ## h264_qpel ## SIZE ## _mc11_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
116{\
117    DECLARE_ALIGNED(16, uint8_t, halfH)[SIZE*SIZE];\
118    DECLARE_ALIGNED(16, uint8_t, halfV)[SIZE*SIZE];\
119    put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src, SIZE, stride);\
120    put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src, SIZE, stride);\
121    OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfV, stride, SIZE, SIZE);\
122}\
123\
124static void OPNAME ## h264_qpel ## SIZE ## _mc31_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
125{\
126    DECLARE_ALIGNED(16, uint8_t, halfH)[SIZE*SIZE];\
127    DECLARE_ALIGNED(16, uint8_t, halfV)[SIZE*SIZE];\
128    put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src, SIZE, stride);\
129    put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src+1, SIZE, stride);\
130    OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfV, stride, SIZE, SIZE);\
131}\
132\
133static void OPNAME ## h264_qpel ## SIZE ## _mc13_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
134{\
135    DECLARE_ALIGNED(16, uint8_t, halfH)[SIZE*SIZE];\
136    DECLARE_ALIGNED(16, uint8_t, halfV)[SIZE*SIZE];\
137    put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src + stride, SIZE, stride);\
138    put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src, SIZE, stride);\
139    OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfV, stride, SIZE, SIZE);\
140}\
141\
142static void OPNAME ## h264_qpel ## SIZE ## _mc33_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
143{\
144    DECLARE_ALIGNED(16, uint8_t, halfH)[SIZE*SIZE];\
145    DECLARE_ALIGNED(16, uint8_t, halfV)[SIZE*SIZE];\
146    put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src + stride, SIZE, stride);\
147    put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src+1, SIZE, stride);\
148    OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfV, stride, SIZE, SIZE);\
149}\
150\
151static void OPNAME ## h264_qpel ## SIZE ## _mc22_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
152{\
153    DECLARE_ALIGNED(16, int16_t, tmp)[SIZE*(SIZE+8)];\
154    OPNAME ## h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(dst, tmp, src, stride, SIZE, stride);\
155}\
156\
157static void OPNAME ## h264_qpel ## SIZE ## _mc21_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
158{\
159    DECLARE_ALIGNED(16, uint8_t, halfH)[SIZE*SIZE];\
160    DECLARE_ALIGNED(16, uint8_t, halfHV)[SIZE*SIZE];\
161    DECLARE_ALIGNED(16, int16_t, tmp)[SIZE*(SIZE+8)];\
162    put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src, SIZE, stride);\
163    put_h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(halfHV, tmp, src, SIZE, SIZE, stride);\
164    OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfHV, stride, SIZE, SIZE);\
165}\
166\
167static void OPNAME ## h264_qpel ## SIZE ## _mc23_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
168{\
169    DECLARE_ALIGNED(16, uint8_t, halfH)[SIZE*SIZE];\
170    DECLARE_ALIGNED(16, uint8_t, halfHV)[SIZE*SIZE];\
171    DECLARE_ALIGNED(16, int16_t, tmp)[SIZE*(SIZE+8)];\
172    put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src + stride, SIZE, stride);\
173    put_h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(halfHV, tmp, src, SIZE, SIZE, stride);\
174    OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfHV, stride, SIZE, SIZE);\
175}\
176\
177static void OPNAME ## h264_qpel ## SIZE ## _mc12_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
178{\
179    DECLARE_ALIGNED(16, uint8_t, halfV)[SIZE*SIZE];\
180    DECLARE_ALIGNED(16, uint8_t, halfHV)[SIZE*SIZE];\
181    DECLARE_ALIGNED(16, int16_t, tmp)[SIZE*(SIZE+8)];\
182    put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src, SIZE, stride);\
183    put_h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(halfHV, tmp, src, SIZE, SIZE, stride);\
184    OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfV, halfHV, stride, SIZE, SIZE);\
185}\
186\
187static void OPNAME ## h264_qpel ## SIZE ## _mc32_ ## CODETYPE(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
188{\
189    DECLARE_ALIGNED(16, uint8_t, halfV)[SIZE*SIZE];\
190    DECLARE_ALIGNED(16, uint8_t, halfHV)[SIZE*SIZE];\
191    DECLARE_ALIGNED(16, int16_t, tmp)[SIZE*(SIZE+8)];\
192    put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src+1, SIZE, stride);\
193    put_h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(halfHV, tmp, src, SIZE, SIZE, stride);\
194    OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfV, halfHV, stride, SIZE, SIZE);\
195}\
196
197#if HAVE_BIGENDIAN
198#define put_unligned_store(s, dest) {    \
199    tmp1 = vec_ld(0, dest);              \
200    mask = vec_lvsl(0, dest);            \
201    tmp2 = vec_ld(15, dest);             \
202    edges = vec_perm(tmp2, tmp1, mask);  \
203    align = vec_lvsr(0, dest);           \
204    tmp2 = vec_perm(s, edges, align);    \
205    tmp1 = vec_perm(edges, s, align);    \
206    vec_st(tmp2, 15, dest);              \
207    vec_st(tmp1, 0 , dest);              \
208 }
209#else
210#define put_unligned_store(s, dest) vec_vsx_st(s, 0, dest);
211#endif /* HAVE_BIGENDIAN */
212
213static inline void put_pixels16_l2_altivec( uint8_t * dst, const uint8_t * src1,
214                                    const uint8_t * src2, int dst_stride,
215                                    int src_stride1, int h)
216{
217    int i;
218    vec_u8 a, b, d, mask_;
219#if HAVE_BIGENDIAN
220    vec_u8 tmp1, tmp2, mask, edges, align;
221    mask_ = vec_lvsl(0, src2);
222#endif
223
224    for (i = 0; i < h; i++) {
225        a = unaligned_load(i * src_stride1, src1);
226        b = load_with_perm_vec(i * 16, src2, mask_);
227        d = vec_avg(a, b);
228        put_unligned_store(d, dst);
229        dst += dst_stride;
230    }
231}
232
233#if HAVE_BIGENDIAN
234#define avg_unligned_store(s, dest){            \
235    tmp1 = vec_ld(0, dest);                     \
236    mask = vec_lvsl(0, dest);                   \
237    tmp2 = vec_ld(15, dest);                    \
238    a = vec_avg(vec_perm(tmp1, tmp2, mask), s); \
239    edges = vec_perm(tmp2, tmp1, mask);         \
240    align = vec_lvsr(0, dest);                  \
241    tmp2 = vec_perm(a, edges, align);           \
242    tmp1 = vec_perm(edges, a, align);           \
243    vec_st(tmp2, 15, dest);                     \
244    vec_st(tmp1, 0 , dest);                     \
245 }
246#else
247#define avg_unligned_store(s, dest){            \
248    a = vec_avg(vec_vsx_ld(0, dst), s);         \
249    vec_vsx_st(a, 0, dst);                      \
250 }
251#endif /* HAVE_BIGENDIAN */
252
253static inline void avg_pixels16_l2_altivec( uint8_t * dst, const uint8_t * src1,
254                                    const uint8_t * src2, int dst_stride,
255                                    int src_stride1, int h)
256{
257    int i;
258    vec_u8 a, b, d, mask_;
259
260#if HAVE_BIGENDIAN
261    vec_u8 tmp1, tmp2, mask, edges, align;
262    mask_ = vec_lvsl(0, src2);
263#endif
264
265    for (i = 0; i < h; i++) {
266        a = unaligned_load(i * src_stride1, src1);
267        b = load_with_perm_vec(i * 16, src2, mask_);
268        d = vec_avg(a, b);
269        avg_unligned_store(d, dst);
270        dst += dst_stride;
271    }
272}
273
274/* Implemented but could be faster
275#define put_pixels16_l2_altivec(d,s1,s2,ds,s1s,h) put_pixels16_l2(d,s1,s2,ds,s1s,16,h)
276#define avg_pixels16_l2_altivec(d,s1,s2,ds,s1s,h) avg_pixels16_l2(d,s1,s2,ds,s1s,16,h)
277 */
278
279H264_MC(put_, 16, altivec)
280H264_MC(avg_, 16, altivec)
281#endif /* HAVE_ALTIVEC */
282
283av_cold void ff_h264qpel_init_ppc(H264QpelContext *c, int bit_depth)
284{
285#if HAVE_ALTIVEC
286    const int high_bit_depth = bit_depth > 8;
287
288    if (!PPC_ALTIVEC(av_get_cpu_flags()))
289        return;
290
291    if (!high_bit_depth) {
292#define dspfunc(PFX, IDX, NUM) \
293        c->PFX ## _pixels_tab[IDX][ 0] = PFX ## NUM ## _mc00_altivec; \
294        c->PFX ## _pixels_tab[IDX][ 1] = PFX ## NUM ## _mc10_altivec; \
295        c->PFX ## _pixels_tab[IDX][ 2] = PFX ## NUM ## _mc20_altivec; \
296        c->PFX ## _pixels_tab[IDX][ 3] = PFX ## NUM ## _mc30_altivec; \
297        c->PFX ## _pixels_tab[IDX][ 4] = PFX ## NUM ## _mc01_altivec; \
298        c->PFX ## _pixels_tab[IDX][ 5] = PFX ## NUM ## _mc11_altivec; \
299        c->PFX ## _pixels_tab[IDX][ 6] = PFX ## NUM ## _mc21_altivec; \
300        c->PFX ## _pixels_tab[IDX][ 7] = PFX ## NUM ## _mc31_altivec; \
301        c->PFX ## _pixels_tab[IDX][ 8] = PFX ## NUM ## _mc02_altivec; \
302        c->PFX ## _pixels_tab[IDX][ 9] = PFX ## NUM ## _mc12_altivec; \
303        c->PFX ## _pixels_tab[IDX][10] = PFX ## NUM ## _mc22_altivec; \
304        c->PFX ## _pixels_tab[IDX][11] = PFX ## NUM ## _mc32_altivec; \
305        c->PFX ## _pixels_tab[IDX][12] = PFX ## NUM ## _mc03_altivec; \
306        c->PFX ## _pixels_tab[IDX][13] = PFX ## NUM ## _mc13_altivec; \
307        c->PFX ## _pixels_tab[IDX][14] = PFX ## NUM ## _mc23_altivec; \
308        c->PFX ## _pixels_tab[IDX][15] = PFX ## NUM ## _mc33_altivec
309
310        dspfunc(put_h264_qpel, 0, 16);
311        dspfunc(avg_h264_qpel, 0, 16);
312#undef dspfunc
313    }
314#endif /* HAVE_ALTIVEC */
315}
316