23 #include "libavutil/attributes.h"
24 #include "libavutil/cpu.h"
33 int dstStride,
int src1Stride,
int h);
36 int src1Stride,
int h);
38 int dstStride,
int src1Stride,
int h);
40 int dstStride,
int src1Stride,
int h);
42 int dstStride,
int src1Stride,
int h);
44 int dstStride,
int src1Stride,
int h);
46 int dstStride,
int srcStride,
int h);
48 int dstStride,
int srcStride,
int h);
50 int dstStride,
int srcStride,
53 int dstStride,
int srcStride,
int h);
55 int dstStride,
int srcStride,
int h);
57 int dstStride,
int srcStride,
60 int dstStride,
int srcStride);
62 int dstStride,
int srcStride);
64 int dstStride,
int srcStride);
66 int dstStride,
int srcStride);
68 int dstStride,
int srcStride);
70 int dstStride,
int srcStride);
71 #define ff_put_no_rnd_pixels16_mmxext ff_put_pixels16_mmxext
72 #define ff_put_no_rnd_pixels8_mmxext ff_put_pixels8_mmxext
92 const int16_t *window,
unsigned int len);
94 const int16_t *window,
unsigned int len);
96 const int16_t *window,
unsigned int len);
98 const int16_t *window,
unsigned int len);
100 const int16_t *window,
unsigned int len);
102 const int16_t *window,
unsigned int len);
109 int *left,
int *left_top);
126 PIXELS16(
static, ff_avg, , , _mmxext)
127 PIXELS16(static, ff_put, , , _mmxext)
129 #define QPEL_OP(OPNAME, RND, MMX) \
130 static void OPNAME ## qpel8_mc00_ ## MMX (uint8_t *dst, uint8_t *src, \
133 ff_ ## OPNAME ## pixels8_ ## MMX(dst, src, stride, 8); \
136 static void OPNAME ## qpel8_mc10_ ## MMX(uint8_t *dst, uint8_t *src, \
140 uint8_t * const half = (uint8_t*)temp; \
141 ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(half, src, 8, \
143 ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src, half, \
144 stride, stride, 8); \
147 static void OPNAME ## qpel8_mc20_ ## MMX(uint8_t *dst, uint8_t *src, \
150 ff_ ## OPNAME ## mpeg4_qpel8_h_lowpass_ ## MMX(dst, src, stride, \
154 static void OPNAME ## qpel8_mc30_ ## MMX(uint8_t *dst, uint8_t *src, \
158 uint8_t * const half = (uint8_t*)temp; \
159 ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(half, src, 8, \
161 ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src + 1, half, stride, \
165 static void OPNAME ## qpel8_mc01_ ## MMX(uint8_t *dst, uint8_t *src, \
169 uint8_t * const half = (uint8_t*)temp; \
170 ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(half, src, \
172 ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src, half, \
173 stride, stride, 8); \
176 static void OPNAME ## qpel8_mc02_ ## MMX(uint8_t *dst, uint8_t *src, \
179 ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, src, \
183 static void OPNAME ## qpel8_mc03_ ## MMX(uint8_t *dst, uint8_t *src, \
187 uint8_t * const half = (uint8_t*)temp; \
188 ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(half, src, \
190 ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src + stride, half, stride,\
194 static void OPNAME ## qpel8_mc11_ ## MMX(uint8_t *dst, uint8_t *src, \
197 uint64_t half[8 + 9]; \
198 uint8_t * const halfH = ((uint8_t*)half) + 64; \
199 uint8_t * const halfHV = ((uint8_t*)half); \
200 ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \
202 ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, 8, \
204 ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
205 ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV, \
209 static void OPNAME ## qpel8_mc31_ ## MMX(uint8_t *dst, uint8_t *src, \
212 uint64_t half[8 + 9]; \
213 uint8_t * const halfH = ((uint8_t*)half) + 64; \
214 uint8_t * const halfHV = ((uint8_t*)half); \
215 ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \
217 ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8, \
219 ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
220 ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV, \
224 static void OPNAME ## qpel8_mc13_ ## MMX(uint8_t *dst, uint8_t *src, \
227 uint64_t half[8 + 9]; \
228 uint8_t * const halfH = ((uint8_t*)half) + 64; \
229 uint8_t * const halfHV = ((uint8_t*)half); \
230 ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \
232 ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, 8, \
234 ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
235 ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV, \
239 static void OPNAME ## qpel8_mc33_ ## MMX(uint8_t *dst, uint8_t *src, \
242 uint64_t half[8 + 9]; \
243 uint8_t * const halfH = ((uint8_t*)half) + 64; \
244 uint8_t * const halfHV = ((uint8_t*)half); \
245 ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \
247 ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8, \
249 ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
250 ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV, \
254 static void OPNAME ## qpel8_mc21_ ## MMX(uint8_t *dst, uint8_t *src, \
257 uint64_t half[8 + 9]; \
258 uint8_t * const halfH = ((uint8_t*)half) + 64; \
259 uint8_t * const halfHV = ((uint8_t*)half); \
260 ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \
262 ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
263 ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV, \
267 static void OPNAME ## qpel8_mc23_ ## MMX(uint8_t *dst, uint8_t *src, \
270 uint64_t half[8 + 9]; \
271 uint8_t * const halfH = ((uint8_t*)half) + 64; \
272 uint8_t * const halfHV = ((uint8_t*)half); \
273 ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \
275 ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
276 ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV, \
280 static void OPNAME ## qpel8_mc12_ ## MMX(uint8_t *dst, uint8_t *src, \
283 uint64_t half[8 + 9]; \
284 uint8_t * const halfH = ((uint8_t*)half); \
285 ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \
287 ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, \
289 ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH, \
293 static void OPNAME ## qpel8_mc32_ ## MMX(uint8_t *dst, uint8_t *src, \
296 uint64_t half[8 + 9]; \
297 uint8_t * const halfH = ((uint8_t*)half); \
298 ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \
300 ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8, \
302 ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH, \
306 static void OPNAME ## qpel8_mc22_ ## MMX(uint8_t *dst, uint8_t *src, \
310 uint8_t * const halfH = ((uint8_t*)half); \
311 ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \
313 ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH, \
317 static void OPNAME ## qpel16_mc00_ ## MMX (uint8_t *dst, uint8_t *src, \
320 ff_ ## OPNAME ## pixels16_ ## MMX(dst, src, stride, 16); \
323 static void OPNAME ## qpel16_mc10_ ## MMX(uint8_t *dst, uint8_t *src, \
327 uint8_t * const half = (uint8_t*)temp; \
328 ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(half, src, 16, \
330 ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src, half, stride, \
334 static void OPNAME ## qpel16_mc20_ ## MMX(uint8_t *dst, uint8_t *src, \
337 ff_ ## OPNAME ## mpeg4_qpel16_h_lowpass_ ## MMX(dst, src, \
338 stride, stride, 16);\
341 static void OPNAME ## qpel16_mc30_ ## MMX(uint8_t *dst, uint8_t *src, \
345 uint8_t * const half = (uint8_t*)temp; \
346 ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(half, src, 16, \
348 ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src + 1, half, \
349 stride, stride, 16); \
352 static void OPNAME ## qpel16_mc01_ ## MMX(uint8_t *dst, uint8_t *src, \
356 uint8_t * const half = (uint8_t*)temp; \
357 ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(half, src, 16, \
359 ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src, half, stride, \
363 static void OPNAME ## qpel16_mc02_ ## MMX(uint8_t *dst, uint8_t *src, \
366 ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, src, \
370 static void OPNAME ## qpel16_mc03_ ## MMX(uint8_t *dst, uint8_t *src, \
374 uint8_t * const half = (uint8_t*)temp; \
375 ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(half, src, 16, \
377 ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src+stride, half, \
378 stride, stride, 16); \
381 static void OPNAME ## qpel16_mc11_ ## MMX(uint8_t *dst, uint8_t *src, \
384 uint64_t half[16 * 2 + 17 * 2]; \
385 uint8_t * const halfH = ((uint8_t*)half) + 256; \
386 uint8_t * const halfHV = ((uint8_t*)half); \
387 ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \
389 ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16, \
391 ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \
393 ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV, \
397 static void OPNAME ## qpel16_mc31_ ## MMX(uint8_t *dst, uint8_t *src, \
400 uint64_t half[16 * 2 + 17 * 2]; \
401 uint8_t * const halfH = ((uint8_t*)half) + 256; \
402 uint8_t * const halfHV = ((uint8_t*)half); \
403 ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \
405 ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16, \
407 ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \
409 ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV, \
413 static void OPNAME ## qpel16_mc13_ ## MMX(uint8_t *dst, uint8_t *src, \
416 uint64_t half[16 * 2 + 17 * 2]; \
417 uint8_t * const halfH = ((uint8_t*)half) + 256; \
418 uint8_t * const halfHV = ((uint8_t*)half); \
419 ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \
421 ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16, \
423 ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \
425 ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV, \
429 static void OPNAME ## qpel16_mc33_ ## MMX(uint8_t *dst, uint8_t *src, \
432 uint64_t half[16 * 2 + 17 * 2]; \
433 uint8_t * const halfH = ((uint8_t*)half) + 256; \
434 uint8_t * const halfHV = ((uint8_t*)half); \
435 ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \
437 ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16, \
439 ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \
441 ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV, \
445 static void OPNAME ## qpel16_mc21_ ## MMX(uint8_t *dst, uint8_t *src, \
448 uint64_t half[16 * 2 + 17 * 2]; \
449 uint8_t * const halfH = ((uint8_t*)half) + 256; \
450 uint8_t * const halfHV = ((uint8_t*)half); \
451 ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \
453 ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \
455 ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV, \
459 static void OPNAME ## qpel16_mc23_ ## MMX(uint8_t *dst, uint8_t *src, \
462 uint64_t half[16 * 2 + 17 * 2]; \
463 uint8_t * const halfH = ((uint8_t*)half) + 256; \
464 uint8_t * const halfHV = ((uint8_t*)half); \
465 ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \
467 ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \
469 ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV, \
473 static void OPNAME ## qpel16_mc12_ ## MMX(uint8_t *dst, uint8_t *src, \
476 uint64_t half[17 * 2]; \
477 uint8_t * const halfH = ((uint8_t*)half); \
478 ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \
480 ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16, \
482 ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH, \
486 static void OPNAME ## qpel16_mc32_ ## MMX(uint8_t *dst, uint8_t *src, \
489 uint64_t half[17 * 2]; \
490 uint8_t * const halfH = ((uint8_t*)half); \
491 ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \
493 ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16, \
495 ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH, \
499 static void OPNAME ## qpel16_mc22_ ## MMX(uint8_t *dst, uint8_t *src, \
502 uint64_t half[17 * 2]; \
503 uint8_t * const halfH = ((uint8_t*)half); \
504 ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \
506 ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH, \
510 QPEL_OP(put_,
_, mmxext)
511 QPEL_OP(avg_,
_, mmxext)
512 QPEL_OP(put_no_rnd_, _no_rnd_, mmxext)
515 #define SET_QPEL_FUNCS(PFX, IDX, SIZE, CPU, PREFIX) \
517 c->PFX ## _pixels_tab[IDX][ 0] = PREFIX ## PFX ## SIZE ## _mc00_ ## CPU; \
518 c->PFX ## _pixels_tab[IDX][ 1] = PREFIX ## PFX ## SIZE ## _mc10_ ## CPU; \
519 c->PFX ## _pixels_tab[IDX][ 2] = PREFIX ## PFX ## SIZE ## _mc20_ ## CPU; \
520 c->PFX ## _pixels_tab[IDX][ 3] = PREFIX ## PFX ## SIZE ## _mc30_ ## CPU; \
521 c->PFX ## _pixels_tab[IDX][ 4] = PREFIX ## PFX ## SIZE ## _mc01_ ## CPU; \
522 c->PFX ## _pixels_tab[IDX][ 5] = PREFIX ## PFX ## SIZE ## _mc11_ ## CPU; \
523 c->PFX ## _pixels_tab[IDX][ 6] = PREFIX ## PFX ## SIZE ## _mc21_ ## CPU; \
524 c->PFX ## _pixels_tab[IDX][ 7] = PREFIX ## PFX ## SIZE ## _mc31_ ## CPU; \
525 c->PFX ## _pixels_tab[IDX][ 8] = PREFIX ## PFX ## SIZE ## _mc02_ ## CPU; \
526 c->PFX ## _pixels_tab[IDX][ 9] = PREFIX ## PFX ## SIZE ## _mc12_ ## CPU; \
527 c->PFX ## _pixels_tab[IDX][10] = PREFIX ## PFX ## SIZE ## _mc22_ ## CPU; \
528 c->PFX ## _pixels_tab[IDX][11] = PREFIX ## PFX ## SIZE ## _mc32_ ## CPU; \
529 c->PFX ## _pixels_tab[IDX][12] = PREFIX ## PFX ## SIZE ## _mc03_ ## CPU; \
530 c->PFX ## _pixels_tab[IDX][13] = PREFIX ## PFX ## SIZE ## _mc13_ ## CPU; \
531 c->PFX ## _pixels_tab[IDX][14] = PREFIX ## PFX ## SIZE ## _mc23_ ## CPU; \
532 c->PFX ## _pixels_tab[IDX][15] = PREFIX ## PFX ## SIZE ## _mc33_ ## CPU; \
545 if (!high_bit_depth) {
551 #if CONFIG_VIDEODSP && (ARCH_X86_32 || !HAVE_YASM)
558 #if HAVE_MMX_EXTERNAL
571 #if HAVE_MMXEXT_INLINE
581 #if HAVE_MMXEXT_EXTERNAL
611 if (!high_bit_depth) {
623 #if HAVE_INLINE_ASM && CONFIG_VIDEODSP
643 #if HAVE_SSE2_EXTERNAL
663 #if HAVE_SSSE3_EXTERNAL
681 #if HAVE_SSE4_EXTERNAL
690 #if HAVE_7REGS && HAVE_INLINE_ASM
void ff_put_no_rnd_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
void ff_apply_window_int16_sse2(int16_t *output, const int16_t *input, const int16_t *window, unsigned int len)
void ff_apply_window_int16_round_sse2(int16_t *output, const int16_t *input, const int16_t *window, unsigned int len)
void ff_bswap32_buf_ssse3(uint32_t *dst, const uint32_t *src, int w)
#define SET_QPEL_FUNCS(PFX, IDX, SIZE, CPU, PREFIX)
void(* put_signed_pixels_clamped)(const int16_t *block, uint8_t *pixels, int line_size)
void ff_apply_window_int16_mmxext(int16_t *output, const int16_t *input, const int16_t *window, unsigned int len)
#define CONFIG_MPEG_XVMC_DECODER
void ff_put_no_rnd_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride)
void ff_vector_clip_int32_sse4(int32_t *dst, const int32_t *src, int32_t min, int32_t max, unsigned int len)
static av_cold void dsputil_init_mmx(DSPContext *c, AVCodecContext *avctx, int cpu_flags)
#define FF_IDCT_SIMPLEMMX
int32_t ff_scalarproduct_int16_sse2(const int16_t *v1, const int16_t *v2, int order)
#define AV_CPU_FLAG_CMOV
supports cmov instruction
void ff_idct_xvid_sse2(short *block)
void(* draw_edges)(uint8_t *buf, int wrap, int width, int height, int w, int h, int sides)
void(* apply_window_int16)(int16_t *output, const int16_t *input, const int16_t *window, unsigned int len)
Apply symmetric window in 16-bit fixed-point.
int bits_per_raw_sample
Bits per sample/pixel of internal libavcodec pixel/sample format.
void(* idct_add)(uint8_t *dest, int line_size, int16_t *block)
block -> idct -> add dest -> clip to unsigned 8 bit -> dest.
void ff_gmc_sse(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy, int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height)
#define AV_CPU_FLAG_3DNOW
AMD 3DNOW.
void ff_clear_blocks_mmx(int16_t *blocks)
void ff_clear_blocks_sse(int16_t *blocks)
void(* idct_put)(uint8_t *dest, int line_size, int16_t *block)
block -> idct -> clip to unsigned 8 bit -> dest.
void ff_idct_xvid_sse2_put(uint8_t *dest, int line_size, short *block)
void ff_add_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels, int line_size)
void(* add_bytes)(uint8_t *dst, uint8_t *src, int w)
void(* add_pixels_clamped)(const int16_t *block, uint8_t *pixels, int line_size)
void ff_put_signed_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels, int line_size)
#define EXTERNAL_SSSE3(flags)
#define CODEC_FLAG_BITEXACT
Use only bitexact stuff (except (I)DCT).
int lowres
low resolution decoding, 1-> 1/2 size, 2->1/4 size
void ff_vector_clip_int32_mmx(int32_t *dst, const int32_t *src, int32_t min, int32_t max, unsigned int len)
void ff_dsputil_init_x86(DSPContext *c, AVCodecContext *avctx)
void ff_idct_xvid_mmxext_put(uint8_t *dest, int line_size, int16_t *block)
int av_get_cpu_flags(void)
Return the flags which specify extensions supported by the CPU.
void(* h263_h_loop_filter)(uint8_t *src, int stride, int qscale)
void ff_dsputilenc_init_mmx(DSPContext *c, AVCodecContext *avctx)
void ff_clear_block_sse(int16_t *block)
int32_t(* scalarproduct_int16)(const int16_t *v1, const int16_t *v2, int len)
Calculate scalar product of two vectors.
void ff_gmc_mmx(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy, int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height)
#define FF_SSE2_IDCT_PERM
#define AV_CPU_FLAG_SSE4
Penryn SSE4.1 functions.
static av_cold void dsputil_init_sse2(DSPContext *c, AVCodecContext *avctx, int cpu_flags)
void ff_vector_clip_int32_int_sse2(int32_t *dst, const int32_t *src, int32_t min, int32_t max, unsigned int len)
void ff_idct_xvid_sse2_add(uint8_t *dest, int line_size, short *block)
void(* bswap_buf)(uint32_t *dst, const uint32_t *src, int w)
static av_cold void dsputil_init_ssse3(DSPContext *c, AVCodecContext *avctx, int cpu_flags)
int(* add_hfyu_left_prediction)(uint8_t *dst, const uint8_t *src, int w, int left)
void ff_add_hfyu_median_prediction_mmxext(uint8_t *dst, const uint8_t *top, const uint8_t *diff, int w, int *left, int *left_top)
void ff_simple_idct_mmx(int16_t *block)
void(* idct)(int16_t *block)
void ff_simple_idct_put_mmx(uint8_t *dest, int line_size, int16_t *block)
void ff_apply_window_int16_round_mmxext(int16_t *output, const int16_t *input, const int16_t *window, unsigned int len)
void(* add_hfyu_median_prediction)(uint8_t *dst, const uint8_t *top, const uint8_t *diff, int w, int *left, int *left_top)
void ff_apply_window_int16_ssse3(int16_t *output, const int16_t *input, const int16_t *window, unsigned int len)
void(* clear_blocks)(int16_t *blocks)
int idct_algo
IDCT algorithm, see FF_IDCT_* below.
void ff_avg_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride)
int ff_add_hfyu_left_prediction_ssse3(uint8_t *dst, const uint8_t *src, int w, int left)
void ff_simple_idct_add_mmx(uint8_t *dest, int line_size, int16_t *block)
void ff_avg_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
int32_t ff_scalarproduct_int16_mmxext(const int16_t *v1, const int16_t *v2, int order)
#define EXTERNAL_SSE4(flags)
#define AV_CPU_FLAG_ATOM
Atom processor, some SSSE3 instructions are slower.
#define AV_CPU_FLAG_SSE42
Nehalem SSE4.2 functions.
int xvmc_acceleration
XVideo Motion Acceleration.
void ff_h263_v_loop_filter_mmx(uint8_t *src, int stride, int qscale)
void ff_put_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h)
void(* put_pixels_clamped)(const int16_t *block, uint8_t *pixels, int line_size)
int idct_permutation_type
void(* vector_clipf)(float *dst, const float *src, float min, float max, int len)
main external API structure.
#define FF_SIMPLE_IDCT_PERM
int ff_add_hfyu_left_prediction_sse4(uint8_t *dst, const uint8_t *src, int w, int left)
int32_t ff_scalarproduct_and_madd_int16_mmxext(int16_t *v1, const int16_t *v2, const int16_t *v3, int order, int mul)
#define CONFIG_H263_DECODER
void ff_avg_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h)
void ff_avg_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride)
void ff_h263_h_loop_filter_mmx(uint8_t *src, int stride, int qscale)
int32_t ff_scalarproduct_and_madd_int16_ssse3(int16_t *v1, const int16_t *v2, const int16_t *v3, int order, int mul)
void ff_idct_xvid_mmx_add(uint8_t *dest, int line_size, int16_t *block)
static av_cold void dsputil_init_sse(DSPContext *c, AVCodecContext *avctx, int cpu_flags)
void ff_idct_xvid_mmx_put(uint8_t *dest, int line_size, int16_t *block)
void ff_idct_xvid_mmxext_add(uint8_t *dest, int line_size, int16_t *block)
void ff_avg_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
void ff_put_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h)
void(* h263_v_loop_filter)(uint8_t *src, int stride, int qscale)
void ff_clear_block_mmx(int16_t *block)
void ff_apply_window_int16_ssse3_atom(int16_t *output, const int16_t *input, const int16_t *window, unsigned int len)
void ff_avg_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h)
void ff_vector_clip_int32_sse2(int32_t *dst, const int32_t *src, int32_t min, int32_t max, unsigned int len)
#define X86_MMXEXT(flags)
header for Xvid IDCT functions
void(* vector_clip_int32)(int32_t *dst, const int32_t *src, int32_t min, int32_t max, unsigned int len)
Clip each element in an array of int32_t to a given minimum and maximum value.
void ff_put_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride)
void(* gmc)(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy, int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height)
global motion compensation.
void ff_add_hfyu_median_prediction_cmov(uint8_t *dst, const uint8_t *top, const uint8_t *diff, int w, int *left, int *left_top)
void(* clear_block)(int16_t *block)
int32_t(* scalarproduct_and_madd_int16)(int16_t *v1, const int16_t *v2, const int16_t *v3, int len, int mul)
Calculate scalar product of v1 and v2, and v1[i] += v3[i] * mul.
#define CONFIG_H263_ENCODER
void ff_put_no_rnd_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h)
static av_cold void dsputil_init_mmxext(DSPContext *c, AVCodecContext *avctx, int cpu_flags)
void ff_put_no_rnd_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride)
void ff_put_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
void ff_draw_edges_mmx(uint8_t *buf, int wrap, int width, int height, int w, int h, int sides)
void ff_put_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels, int line_size)
void ff_bswap32_buf_sse2(uint32_t *dst, const uint32_t *src, int w)
void ff_put_no_rnd_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
#define AV_CPU_FLAG_SSE2SLOW
SSE2 supported, but usually not faster.
static av_cold void dsputil_init_sse4(DSPContext *c, AVCodecContext *avctx, int cpu_flags)
void ff_idct_xvid_mmx(short *block)
void ff_vector_clipf_sse(float *dst, const float *src, float min, float max, int len)
void ff_idct_xvid_mmxext(short *block)
void ff_put_no_rnd_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h)
void ff_add_bytes_mmx(uint8_t *dst, uint8_t *src, int w)
void ff_put_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride)
int32_t ff_scalarproduct_and_madd_int16_sse2(int16_t *v1, const int16_t *v2, const int16_t *v3, int order, int mul)
void ff_put_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
#define PIXELS16(STATIC, PFX1, PFX2, TYPE, CPUEXT)