FFmpeg coverage

Directory: ../../../ffmpeg/
File: src/libavcodec/x86/h264_qpel.c
Date: 2026-01-23 19:11:46
Exec Total Coverage
Lines: 79 79 100.0%
Functions: 146 146 100.0%
Branches: 20 20 100.0%
Line Branch Exec Source
1 /*
2 * Copyright (c) 2004-2005 Michael Niedermayer, Loren Merritt
3 * Copyright (c) 2011 Daniel Kang
4 *
5 * This file is part of FFmpeg.
6 *
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22 #include <stddef.h>
23
24 #include "libavutil/attributes.h"
25 #include "libavutil/avassert.h"
26 #include "libavutil/cpu.h"
27 #include "libavutil/mem_internal.h"
28 #include "libavutil/x86/cpu.h"
29 #include "libavcodec/h264qpel.h"
30 #include "fpel.h"
31 #include "qpel.h"
32
33 void ff_avg_pixels4_mmxext(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
34 void ff_put_pixels4x4_l2_mmxext(uint8_t *dst, const uint8_t *src1, const uint8_t *src2,
35 ptrdiff_t stride);
36 void ff_avg_pixels4x4_l2_mmxext(uint8_t *dst, const uint8_t *src1, const uint8_t *src2,
37 ptrdiff_t stride);
38 #define ff_put_pixels4x4_l2_mmxext(dst, src1, src2, dststride, src1stride) \
39 ff_put_pixels4x4_l2_mmxext((dst), (src1), (src2), (dststride))
40 #define ff_avg_pixels4x4_l2_mmxext(dst, src1, src2, dststride, src1stride) \
41 ff_avg_pixels4x4_l2_mmxext((dst), (src1), (src2), (dststride))
42 #define ff_put_pixels8x8_l2_sse2 ff_put_pixels8x8_l2_mmxext
43 #define ff_avg_pixels8x8_l2_sse2 ff_avg_pixels8x8_l2_mmxext
44
45 #define DEF_QPEL(OPNAME)\
46 void ff_ ## OPNAME ## _h264_qpel4_h_lowpass_mmxext(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride);\
47 void ff_ ## OPNAME ## _h264_qpel8_h_lowpass_ssse3(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride);\
48 void ff_ ## OPNAME ## _h264_qpel4_h_lowpass_l2_mmxext(uint8_t *dst, const uint8_t *src, const uint8_t *src2, ptrdiff_t dstStride, ptrdiff_t src2Stride);\
49 void ff_ ## OPNAME ## _h264_qpel8_h_lowpass_l2_sse2(uint8_t *dst, const uint8_t *src, const uint8_t *src2, ptrdiff_t dstStride, ptrdiff_t src2Stride);\
50 void ff_ ## OPNAME ## _h264_qpel16_h_lowpass_l2_sse2(uint8_t *dst, const uint8_t *src, const uint8_t *src2, ptrdiff_t dstStride, ptrdiff_t src2Stride);\
51 void ff_ ## OPNAME ## _h264_qpel8_h_lowpass_l2_ssse3(uint8_t *dst, const uint8_t *src, const uint8_t *src2, ptrdiff_t dstStride, ptrdiff_t src2Stride);\
52 void ff_ ## OPNAME ## _h264_qpel4_v_lowpass_mmxext(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride);\
53 void ff_ ## OPNAME ## _h264_qpel8or16_v_lowpass_sse2(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride, int h);\
54 void ff_ ## OPNAME ## _h264_qpel4_hv_lowpass_h_mmxext(int16_t *tmp, uint8_t *dst, ptrdiff_t dstStride);\
55 void ff_ ## OPNAME ## _h264_qpel8or16_hv1_lowpass_op_sse2(const uint8_t *src, int16_t *tmp, ptrdiff_t srcStride, int size);\
56 void ff_ ## OPNAME ## _h264_qpel8_hv2_lowpass_sse2(uint8_t *dst, int16_t *tmp, ptrdiff_t dstStride);\
57 void ff_ ## OPNAME ## _h264_qpel16_hv2_lowpass_sse2(uint8_t *dst, int16_t *tmp, ptrdiff_t dstStride);\
58 void ff_ ## OPNAME ## _h264_qpel8_hv2_lowpass_ssse3(uint8_t *dst, int16_t *tmp, ptrdiff_t dstStride);\
59 void ff_ ## OPNAME ## _h264_qpel16_hv2_lowpass_ssse3(uint8_t *dst, int16_t *tmp, ptrdiff_t dstStride);\
60 void ff_ ## OPNAME ## _pixels4_l2_shift5_mmxext(uint8_t *dst, const int16_t *src16, const uint8_t *src8, ptrdiff_t dstStride);\
61 void ff_ ## OPNAME ## _pixels8_l2_shift5_sse2(uint8_t *dst, const int16_t *src16, const uint8_t *src8, ptrdiff_t dstStride);\
62 void ff_ ## OPNAME ## _pixels16_l2_shift5_sse2(uint8_t *dst, const int16_t *src16, const uint8_t *src8, ptrdiff_t dstStride);\
63
64 void ff_put_h264_qpel4_hv_lowpass_v_mmxext(const uint8_t *src, int16_t *tmp, ptrdiff_t srcStride);
65
66 DEF_QPEL(avg)
67 DEF_QPEL(put)
68
69 #define QPEL_H264(OPNAME, MMX)\
70 static av_always_inline void OPNAME ## h264_qpel4_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
71 {\
72 src -= 2*srcStride+2;\
73 ff_put_h264_qpel4_hv_lowpass_v_mmxext(src, tmp, srcStride);\
74 ff_ ## OPNAME ## h264_qpel4_hv_lowpass_h_mmxext(tmp, dst, dstStride);\
75 }\
76
77 #define QPEL_H264_H16(OPNAME, EXT) \
78 static av_always_inline void ff_ ## OPNAME ## h264_qpel16_h_lowpass_l2_ ## EXT(uint8_t *dst, const uint8_t *src, const uint8_t *src2, ptrdiff_t dstStride, ptrdiff_t src2Stride)\
79 {\
80 ff_ ## OPNAME ## h264_qpel8_h_lowpass_l2_ ## EXT(dst , src , src2 , dstStride, src2Stride);\
81 ff_ ## OPNAME ## h264_qpel8_h_lowpass_l2_ ## EXT(dst+8, src+8, src2+8, dstStride, src2Stride);\
82 src += 8*dstStride;\
83 dst += 8*dstStride;\
84 src2 += 8*src2Stride;\
85 ff_ ## OPNAME ## h264_qpel8_h_lowpass_l2_ ## EXT(dst , src , src2 , dstStride, src2Stride);\
86 ff_ ## OPNAME ## h264_qpel8_h_lowpass_l2_ ## EXT(dst+8, src+8, src2+8, dstStride, src2Stride);\
87 }\
88
89
90 #if ARCH_X86_64
91 #define QPEL_H264_H16_XMM(OPNAME, MMX)\
92
93 void ff_avg_h264_qpel16_h_lowpass_l2_ssse3(uint8_t *dst, const uint8_t *src, const uint8_t *src2, ptrdiff_t dstStride, ptrdiff_t src2Stride);
94 void ff_put_h264_qpel16_h_lowpass_l2_ssse3(uint8_t *dst, const uint8_t *src, const uint8_t *src2, ptrdiff_t dstStride, ptrdiff_t src2Stride);
95
96 #else // ARCH_X86_64
97 #define QPEL_H264_H16_XMM(OPNAME, EXT) QPEL_H264_H16(OPNAME, EXT)
98 #endif // ARCH_X86_64
99
100 #define QPEL_H264_H_XMM(OPNAME, MMX)\
101 QPEL_H264_H16_XMM(OPNAME, MMX)\
102 static av_always_inline void ff_ ## OPNAME ## h264_qpel16_h_lowpass_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
103 {\
104 ff_ ## OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst , src , dstStride, srcStride);\
105 ff_ ## OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst+8, src+8, dstStride, srcStride);\
106 src += 8*srcStride;\
107 dst += 8*dstStride;\
108 ff_ ## OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst , src , dstStride, srcStride);\
109 ff_ ## OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst+8, src+8, dstStride, srcStride);\
110 }\
111
112 #define QPEL_H264_V_XMM(OPNAME, XMM, XMM2)\
113 static av_always_inline void ff_ ## OPNAME ## h264_qpel8_v_lowpass_ ## XMM(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
114 {\
115 ff_ ## OPNAME ## h264_qpel8or16_v_lowpass_ ## XMM2(dst , src , dstStride, srcStride, 8);\
116 }\
117 static av_always_inline void ff_ ## OPNAME ## h264_qpel16_v_lowpass_ ## XMM(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
118 {\
119 ff_ ## OPNAME ## h264_qpel8or16_v_lowpass_ ## XMM2(dst , src , dstStride, srcStride, 16);\
120 ff_ ## OPNAME ## h264_qpel8or16_v_lowpass_ ## XMM2(dst+8, src+8, dstStride, srcStride, 16);\
121 }
122
123 161167 static av_always_inline void put_h264_qpel8or16_hv1_lowpass_sse2(int16_t *tmp,
124 const uint8_t *src,
125 ptrdiff_t srcStride,
126 int size)
127 {
128 161167 int w = (size+8)>>3;
129 161167 src -= 2*srcStride+2;
130
2/2
✓ Branch 0 taken 388862 times.
✓ Branch 1 taken 161167 times.
 550029 while(w--){
131 388862 ff_put_h264_qpel8or16_hv1_lowpass_op_sse2(src, tmp, srcStride, size);
132 388862 tmp += 8;
133 388862 src += 8;
134 }
135 161167 }
136
137 #define QPEL_H264_HV_XMM(OPNAME, MMX)\
138 static av_always_inline void OPNAME ## h264_qpel8_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
139 {\
140 put_h264_qpel8or16_hv1_lowpass_sse2(tmp, src, srcStride, 8);\
141 ff_ ## OPNAME ## h264_qpel8_hv2_lowpass_ ## MMX(dst, tmp, dstStride);\
142 }\
143 static av_always_inline void OPNAME ## h264_qpel16_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
144 {\
145 put_h264_qpel8or16_hv1_lowpass_sse2(tmp, src, srcStride, 16);\
146 ff_ ## OPNAME ## h264_qpel16_hv2_lowpass_ ## MMX(dst, tmp, dstStride);\
147 }\
148
149 #define H264_MC_V_H_HV(OPNAME, SIZE, MMX, ALIGN, SHIFT5_EXT) \
150 H264_MC_V(OPNAME, SIZE, MMX, ALIGN, SHIFT5_EXT)\
151 H264_MC_H(OPNAME, SIZE, MMX, ALIGN, SHIFT5_EXT)\
152 H264_MC_HV(OPNAME, SIZE, MMX, ALIGN, SHIFT5_EXT)\
153
154 #define H264_MC_H(OPNAME, SIZE, MMX, ALIGN, UNUSED) \
155 static void OPNAME ## h264_qpel ## SIZE ## _mc10_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
156 {\
157 ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src, src, stride, stride);\
158 }\
159 \
160 static void OPNAME ## h264_qpel ## SIZE ## _mc20_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
161 {\
162 ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_ ## MMX(dst, src, stride, stride);\
163 }\
164 \
165 static void OPNAME ## h264_qpel ## SIZE ## _mc30_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
166 {\
167 ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src, src+1, stride, stride);\
168 }\
169
170 #define H264_MC_V(OPNAME, SIZE, MMX, ALIGN, UNUSED) \
171 static void OPNAME ## h264_qpel ## SIZE ## _mc01_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
172 {\
173 LOCAL_ALIGNED(ALIGN, uint8_t, temp, [SIZE*SIZE]);\
174 ff_put_h264_qpel ## SIZE ## _v_lowpass_ ## MMX(temp, src, SIZE, stride);\
175 ff_ ## OPNAME ## pixels ## SIZE ## x ## SIZE ## _l2_ ## MMX(dst, src, temp, stride, stride);\
176 }\
177 \
178 static void OPNAME ## h264_qpel ## SIZE ## _mc02_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
179 {\
180 ff_ ## OPNAME ## h264_qpel ## SIZE ## _v_lowpass_ ## MMX(dst, src, stride, stride);\
181 }\
182 \
183 static void OPNAME ## h264_qpel ## SIZE ## _mc03_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
184 {\
185 LOCAL_ALIGNED(ALIGN, uint8_t, temp, [SIZE*SIZE]);\
186 ff_put_h264_qpel ## SIZE ## _v_lowpass_ ## MMX(temp, src, SIZE, stride);\
187 ff_ ## OPNAME ## pixels ## SIZE ## x ## SIZE ## _l2_ ## MMX(dst, src+stride, temp, stride, stride);\
188 }\
189
190 #define H264_MC_HV(OPNAME, SIZE, MMX, ALIGN, SHIFT5_EXT) \
191 static void OPNAME ## h264_qpel ## SIZE ## _mc11_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
192 {\
193 LOCAL_ALIGNED(ALIGN, uint8_t, temp, [SIZE*SIZE]);\
194 ff_put_h264_qpel ## SIZE ## _v_lowpass_ ## MMX(temp, src, SIZE, stride);\
195 ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src, temp, stride, SIZE);\
196 }\
197 \
198 static void OPNAME ## h264_qpel ## SIZE ## _mc31_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
199 {\
200 LOCAL_ALIGNED(ALIGN, uint8_t, temp, [SIZE*SIZE]);\
201 ff_put_h264_qpel ## SIZE ## _v_lowpass_ ## MMX(temp, src+1, SIZE, stride);\
202 ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src, temp, stride, SIZE);\
203 }\
204 \
205 static void OPNAME ## h264_qpel ## SIZE ## _mc13_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
206 {\
207 LOCAL_ALIGNED(ALIGN, uint8_t, temp, [SIZE*SIZE]);\
208 ff_put_h264_qpel ## SIZE ## _v_lowpass_ ## MMX(temp, src, SIZE, stride);\
209 ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src+stride, temp, stride, SIZE);\
210 }\
211 \
212 static void OPNAME ## h264_qpel ## SIZE ## _mc33_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
213 {\
214 LOCAL_ALIGNED(ALIGN, uint8_t, temp, [SIZE*SIZE]);\
215 ff_put_h264_qpel ## SIZE ## _v_lowpass_ ## MMX(temp, src+1, SIZE, stride);\
216 ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src+stride, temp, stride, SIZE);\
217 }\
218 \
219 static void OPNAME ## h264_qpel ## SIZE ## _mc22_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
220 {\
221 LOCAL_ALIGNED(ALIGN, uint16_t, temp, [SIZE*(SIZE<8?12:24)]);\
222 OPNAME ## h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(dst, temp, src, stride, stride);\
223 }\
224 \
225 static void OPNAME ## h264_qpel ## SIZE ## _mc21_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
226 {\
227 LOCAL_ALIGNED(ALIGN, uint8_t, temp, [SIZE*(SIZE<8?12:24)*2 + SIZE*SIZE]);\
228 uint8_t * const halfHV= temp;\
229 int16_t * const halfV= (int16_t*)(temp + SIZE*SIZE);\
230 av_assert2(((uintptr_t)temp & 7) == 0);\
231 put_h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(halfHV, halfV, src, SIZE, stride);\
232 ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src, halfHV, stride, SIZE);\
233 }\
234 \
235 static void OPNAME ## h264_qpel ## SIZE ## _mc23_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
236 {\
237 LOCAL_ALIGNED(ALIGN, uint8_t, temp, [SIZE*(SIZE<8?12:24)*2 + SIZE*SIZE]);\
238 uint8_t * const halfHV= temp;\
239 int16_t * const halfV= (int16_t*)(temp + SIZE*SIZE);\
240 av_assert2(((uintptr_t)temp & 7) == 0);\
241 put_h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(halfHV, halfV, src, SIZE, stride);\
242 ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src+stride, halfHV, stride, SIZE);\
243 }\
244 \
245 static void OPNAME ## h264_qpel ## SIZE ## _mc12_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
246 {\
247 LOCAL_ALIGNED(ALIGN, uint8_t, temp, [SIZE*(SIZE<8?12:24)*2 + SIZE*SIZE]);\
248 uint8_t * const halfHV= temp;\
249 int16_t * const halfV= (int16_t*)(temp + SIZE*SIZE);\
250 av_assert2(((uintptr_t)temp & 7) == 0);\
251 put_h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(halfHV, halfV, src, SIZE, stride);\
252 ff_ ## OPNAME ## pixels ## SIZE ## _l2_shift5_ ## SHIFT5_EXT(dst, halfV+2, halfHV, stride);\
253 }\
254 \
255 static void OPNAME ## h264_qpel ## SIZE ## _mc32_ ## MMX(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
256 {\
257 LOCAL_ALIGNED(ALIGN, uint8_t, temp, [SIZE*(SIZE<8?12:24)*2 + SIZE*SIZE]);\
258 uint8_t * const halfHV= temp;\
259 int16_t * const halfV= (int16_t*)(temp + SIZE*SIZE);\
260 av_assert2(((uintptr_t)temp & 7) == 0);\
261 put_h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(halfHV, halfV, src, SIZE, stride);\
262 ff_ ## OPNAME ## pixels ## SIZE ## _l2_shift5_ ## SHIFT5_EXT(dst, halfV+3, halfHV, stride);\
263 }\
264
265 #define H264_MC(QPEL, SIZE, MMX, ALIGN, SHIFT5_EXT)\
266 QPEL(put_, SIZE, MMX, ALIGN, SHIFT5_EXT) \
267 QPEL(avg_, SIZE, MMX, ALIGN, SHIFT5_EXT) \
268
269 #define H264_MC_816(QPEL, XMM, SHIFT5_EXT)\
270 QPEL(put_, 8, XMM, 16, SHIFT5_EXT)\
271 QPEL(put_, 16,XMM, 16, SHIFT5_EXT)\
272 QPEL(avg_, 8, XMM, 16, SHIFT5_EXT)\
273 QPEL(avg_, 16,XMM, 16, SHIFT5_EXT)\
274
275 337 QPEL_H264(put_, mmxext)
276 1 QPEL_H264(avg_, mmxext)
277 177686 QPEL_H264_V_XMM(put_, sse2, sse2)
278 3756 QPEL_H264_V_XMM(avg_, sse2, sse2)
279 72 QPEL_H264_HV_XMM(put_, sse2)
280 8 QPEL_H264_HV_XMM(avg_, sse2)
281 13934 QPEL_H264_H_XMM(put_, ssse3)
282 3344 QPEL_H264_H_XMM(avg_, ssse3)
283 320780 QPEL_H264_V_XMM(put_, ssse3, sse2)
284 316678 QPEL_H264_HV_XMM(put_, ssse3)
285 5576 QPEL_H264_HV_XMM(avg_, ssse3)
286
287 2284 H264_MC(H264_MC_V_H_HV, 4, mmxext, 8, mmxext)
288 181378 H264_MC_816(H264_MC_V, sse2, sse2)
289 144 H264_MC_816(H264_MC_HV, sse2, sse2)
290 233994 H264_MC_816(H264_MC_H, ssse3, sse2)
291 643034 H264_MC_816(H264_MC_HV, ssse3, sse2)
292
293
294 //10bit
295 #define LUMA_MC_OP(OP, NUM, DEPTH, TYPE, OPT) \
296 void ff_ ## OP ## _h264_qpel ## NUM ## _ ## TYPE ## _ ## DEPTH ## _ ## OPT \
297 (uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
298
299 #define LUMA_MC_4(DEPTH, TYPE, OPT) \
300 LUMA_MC_OP(put, 4, DEPTH, TYPE, OPT) \
301 LUMA_MC_OP(avg, 4, DEPTH, TYPE, OPT)
302
303 #define LUMA_MC_816(DEPTH, TYPE, OPT) \
304 LUMA_MC_OP(put, 8, DEPTH, TYPE, OPT) \
305 LUMA_MC_OP(avg, 8, DEPTH, TYPE, OPT) \
306 LUMA_MC_OP(put, 16, DEPTH, TYPE, OPT) \
307 LUMA_MC_OP(avg, 16, DEPTH, TYPE, OPT)
308
309 LUMA_MC_4(10, mc00, mmxext)
310 LUMA_MC_4(10, mc10, mmxext)
311 LUMA_MC_4(10, mc20, mmxext)
312 LUMA_MC_4(10, mc30, mmxext)
313 LUMA_MC_4(10, mc01, mmxext)
314 LUMA_MC_4(10, mc11, mmxext)
315 LUMA_MC_4(10, mc21, mmxext)
316 LUMA_MC_4(10, mc31, mmxext)
317 LUMA_MC_4(10, mc02, mmxext)
318 LUMA_MC_4(10, mc12, mmxext)
319 LUMA_MC_4(10, mc22, mmxext)
320 LUMA_MC_4(10, mc32, mmxext)
321 LUMA_MC_4(10, mc03, mmxext)
322 LUMA_MC_4(10, mc13, mmxext)
323 LUMA_MC_4(10, mc23, mmxext)
324 LUMA_MC_4(10, mc33, mmxext)
325
326 LUMA_MC_816(10, mc00, sse2)
327 LUMA_MC_816(10, mc10, sse2)
328 LUMA_MC_816(10, mc10, ssse3_cache64)
329 LUMA_MC_816(10, mc20, sse2)
330 LUMA_MC_816(10, mc20, ssse3_cache64)
331 LUMA_MC_816(10, mc30, sse2)
332 LUMA_MC_816(10, mc30, ssse3_cache64)
333 LUMA_MC_816(10, mc01, sse2)
334 LUMA_MC_816(10, mc11, sse2)
335 LUMA_MC_816(10, mc21, sse2)
336 LUMA_MC_816(10, mc31, sse2)
337 LUMA_MC_816(10, mc02, sse2)
338 LUMA_MC_816(10, mc12, sse2)
339 LUMA_MC_816(10, mc22, sse2)
340 LUMA_MC_816(10, mc32, sse2)
341 LUMA_MC_816(10, mc03, sse2)
342 LUMA_MC_816(10, mc13, sse2)
343 LUMA_MC_816(10, mc23, sse2)
344 LUMA_MC_816(10, mc33, sse2)
345
346 #define SET_QPEL_FUNCS_1PP(PFX, IDX, SIZE, CPU, PREFIX) \
347 do { \
348 c->PFX ## _pixels_tab[IDX][ 1] = PREFIX ## PFX ## SIZE ## _mc10_ ## CPU; \
349 c->PFX ## _pixels_tab[IDX][ 2] = PREFIX ## PFX ## SIZE ## _mc20_ ## CPU; \
350 c->PFX ## _pixels_tab[IDX][ 3] = PREFIX ## PFX ## SIZE ## _mc30_ ## CPU; \
351 c->PFX ## _pixels_tab[IDX][ 4] = PREFIX ## PFX ## SIZE ## _mc01_ ## CPU; \
352 c->PFX ## _pixels_tab[IDX][ 5] = PREFIX ## PFX ## SIZE ## _mc11_ ## CPU; \
353 c->PFX ## _pixels_tab[IDX][ 6] = PREFIX ## PFX ## SIZE ## _mc21_ ## CPU; \
354 c->PFX ## _pixels_tab[IDX][ 7] = PREFIX ## PFX ## SIZE ## _mc31_ ## CPU; \
355 c->PFX ## _pixels_tab[IDX][ 8] = PREFIX ## PFX ## SIZE ## _mc02_ ## CPU; \
356 c->PFX ## _pixels_tab[IDX][ 9] = PREFIX ## PFX ## SIZE ## _mc12_ ## CPU; \
357 c->PFX ## _pixels_tab[IDX][10] = PREFIX ## PFX ## SIZE ## _mc22_ ## CPU; \
358 c->PFX ## _pixels_tab[IDX][11] = PREFIX ## PFX ## SIZE ## _mc32_ ## CPU; \
359 c->PFX ## _pixels_tab[IDX][12] = PREFIX ## PFX ## SIZE ## _mc03_ ## CPU; \
360 c->PFX ## _pixels_tab[IDX][13] = PREFIX ## PFX ## SIZE ## _mc13_ ## CPU; \
361 c->PFX ## _pixels_tab[IDX][14] = PREFIX ## PFX ## SIZE ## _mc23_ ## CPU; \
362 c->PFX ## _pixels_tab[IDX][15] = PREFIX ## PFX ## SIZE ## _mc33_ ## CPU; \
363 } while (0)
364 #define SET_QPEL_FUNCS(PFX, IDX, SIZE, CPU, PREFIX) \
365 do { \
366 c->PFX ## _pixels_tab[IDX][ 0] = PREFIX ## PFX ## SIZE ## _mc00_ ## CPU; \
367 SET_QPEL_FUNCS_1PP(PFX, IDX, SIZE, CPU, PREFIX); \
368 } while (0)
369
370 #define H264_QPEL_FUNCS(x, y, CPU) \
371 do { \
372 c->put_h264_qpel_pixels_tab[0][x + y * 4] = put_h264_qpel16_mc ## x ## y ## _ ## CPU; \
373 c->put_h264_qpel_pixels_tab[1][x + y * 4] = put_h264_qpel8_mc ## x ## y ## _ ## CPU; \
374 c->avg_h264_qpel_pixels_tab[0][x + y * 4] = avg_h264_qpel16_mc ## x ## y ## _ ## CPU; \
375 c->avg_h264_qpel_pixels_tab[1][x + y * 4] = avg_h264_qpel8_mc ## x ## y ## _ ## CPU; \
376 } while (0)
377
378 #define H264_QPEL_FUNCS_10(x, y, CPU) \
379 do { \
380 c->put_h264_qpel_pixels_tab[0][x + y * 4] = ff_put_h264_qpel16_mc ## x ## y ## _10_ ## CPU; \
381 c->put_h264_qpel_pixels_tab[1][x + y * 4] = ff_put_h264_qpel8_mc ## x ## y ## _10_ ## CPU; \
382 c->avg_h264_qpel_pixels_tab[0][x + y * 4] = ff_avg_h264_qpel16_mc ## x ## y ## _10_ ## CPU; \
383 c->avg_h264_qpel_pixels_tab[1][x + y * 4] = ff_avg_h264_qpel8_mc ## x ## y ## _10_ ## CPU; \
384 } while (0)
385
386 755 av_cold void ff_h264qpel_init_x86(H264QpelContext *c, int bit_depth)
387 {
388 755 int high_bit_depth = bit_depth > 8;
389 755 int cpu_flags = av_get_cpu_flags();
390
391
2/2
✓ Branch 0 taken 141 times.
✓ Branch 1 taken 614 times.
 755 if (EXTERNAL_MMXEXT(cpu_flags)) {
392
2/2
✓ Branch 0 taken 81 times.
✓ Branch 1 taken 60 times.
 141 if (!high_bit_depth) {
393 81 SET_QPEL_FUNCS_1PP(put_h264_qpel, 2, 4, mmxext, );
394 81 c->avg_h264_qpel_pixels_tab[1][0] = ff_avg_pixels8x8_mmxext;
395 81 SET_QPEL_FUNCS_1PP(avg_h264_qpel, 2, 4, mmxext, );
396 81 c->avg_h264_qpel_pixels_tab[2][0] = ff_avg_pixels4_mmxext;
397
2/2
✓ Branch 0 taken 36 times.
✓ Branch 1 taken 24 times.
 60 } else if (bit_depth == 10) {
398 36 SET_QPEL_FUNCS(put_h264_qpel, 2, 4, 10_mmxext, ff_);
399 36 SET_QPEL_FUNCS(avg_h264_qpel, 2, 4, 10_mmxext, ff_);
400 }
401 }
402
403
2/2
✓ Branch 0 taken 127 times.
✓ Branch 1 taken 628 times.
 755 if (EXTERNAL_SSE2(cpu_flags)) {
404
2/2
✓ Branch 0 taken 75 times.
✓ Branch 1 taken 52 times.
 127 if (!high_bit_depth) {
405 75 H264_QPEL_FUNCS(0, 1, sse2);
406 75 H264_QPEL_FUNCS(0, 2, sse2);
407 75 H264_QPEL_FUNCS(0, 3, sse2);
408 75 H264_QPEL_FUNCS(1, 1, sse2);
409 75 H264_QPEL_FUNCS(1, 2, sse2);
410 75 H264_QPEL_FUNCS(1, 3, sse2);
411 75 H264_QPEL_FUNCS(2, 1, sse2);
412 75 H264_QPEL_FUNCS(2, 2, sse2);
413 75 H264_QPEL_FUNCS(2, 3, sse2);
414 75 H264_QPEL_FUNCS(3, 1, sse2);
415 75 H264_QPEL_FUNCS(3, 2, sse2);
416 75 H264_QPEL_FUNCS(3, 3, sse2);
417 75 c->put_h264_qpel_pixels_tab[0][0] = ff_put_pixels16x16_sse2;
418 75 c->avg_h264_qpel_pixels_tab[0][0] = ff_avg_pixels16x16_sse2;
419 }
420
421
2/2
✓ Branch 0 taken 32 times.
✓ Branch 1 taken 95 times.
 127 if (bit_depth == 10) {
422 32 SET_QPEL_FUNCS(put_h264_qpel, 0, 16, 10_sse2, ff_);
423 32 SET_QPEL_FUNCS(put_h264_qpel, 1, 8, 10_sse2, ff_);
424 32 SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, 10_sse2, ff_);
425 32 SET_QPEL_FUNCS(avg_h264_qpel, 1, 8, 10_sse2, ff_);
426 32 H264_QPEL_FUNCS_10(1, 0, sse2);
427 32 H264_QPEL_FUNCS_10(2, 0, sse2);
428 32 H264_QPEL_FUNCS_10(3, 0, sse2);
429 }
430 }
431
432
2/2
✓ Branch 0 taken 113 times.
✓ Branch 1 taken 642 times.
 755 if (EXTERNAL_SSSE3(cpu_flags)) {
433
2/2
✓ Branch 0 taken 69 times.
✓ Branch 1 taken 44 times.
 113 if (!high_bit_depth) {
434 69 H264_QPEL_FUNCS(1, 0, ssse3);
435 69 H264_QPEL_FUNCS(1, 1, ssse3);
436 69 H264_QPEL_FUNCS(1, 2, ssse3);
437 69 H264_QPEL_FUNCS(1, 3, ssse3);
438 69 H264_QPEL_FUNCS(2, 0, ssse3);
439 69 H264_QPEL_FUNCS(2, 1, ssse3);
440 69 H264_QPEL_FUNCS(2, 2, ssse3);
441 69 H264_QPEL_FUNCS(2, 3, ssse3);
442 69 H264_QPEL_FUNCS(3, 0, ssse3);
443 69 H264_QPEL_FUNCS(3, 1, ssse3);
444 69 H264_QPEL_FUNCS(3, 2, ssse3);
445 69 H264_QPEL_FUNCS(3, 3, ssse3);
446 }
447
448
2/2
✓ Branch 0 taken 28 times.
✓ Branch 1 taken 85 times.
 113 if (bit_depth == 10) {
449 28 H264_QPEL_FUNCS_10(1, 0, ssse3_cache64);
450 28 H264_QPEL_FUNCS_10(2, 0, ssse3_cache64);
451 28 H264_QPEL_FUNCS_10(3, 0, ssse3_cache64);
452 }
453 }
454 755 }
455