Browse code
dsputil: Split off quarterpel bits into their own context
Diego Biurrun authored on 2014/01/08 22:00:10Showing 30 changed files
- configure
- libavcodec/Makefile
- libavcodec/cavs.c
- libavcodec/cavsdsp.h
- libavcodec/dsputil.c
- libavcodec/dsputil.h
- libavcodec/dsputil_template.c
- libavcodec/h263dec.c
- libavcodec/h264.h
- libavcodec/h264_mb.c
- libavcodec/h264qpel.h
- libavcodec/motion_est.c
- libavcodec/mpegvideo.c
- libavcodec/mpegvideo.h
- libavcodec/mpegvideo_enc.c
- libavcodec/mpegvideo_motion.c
- libavcodec/mss2.c
- libavcodec/qpel_template.c
- libavcodec/qpeldsp.c
- libavcodec/qpeldsp.h
- libavcodec/rv34.c
- libavcodec/rv34dsp.h
- libavcodec/vc1dec.c
- libavcodec/vc1dsp.c
- libavcodec/wmv2dsp.h
- libavcodec/x86/Makefile
- libavcodec/x86/dsputil_init.c
- libavcodec/x86/mpeg4qpel.asm
- libavcodec/x86/qpeldsp.asm
- libavcodec/x86/qpeldsp_init.c
| ... | ... |
@@ -1551,6 +1551,7 @@ CONFIG_EXTRA=" |
| 1551 | 1551 |
mpegvideo |
| 1552 | 1552 |
mpegvideoenc |
| 1553 | 1553 |
nettle |
| 1554 |
+ qpeldsp |
|
| 1554 | 1555 |
rangecoder |
| 1555 | 1556 |
riffdec |
| 1556 | 1557 |
riffenc |
| ... | ... |
@@ -1704,7 +1705,7 @@ rdft_select="fft" |
| 1704 | 1704 |
mpegaudio_select="mpegaudiodsp" |
| 1705 | 1705 |
mpegaudiodsp_select="dct" |
| 1706 | 1706 |
mpegvideo_select="dsputil hpeldsp videodsp" |
| 1707 |
-mpegvideoenc_select="dsputil mpegvideo" |
|
| 1707 |
+mpegvideoenc_select="dsputil mpegvideo qpeldsp" |
|
| 1708 | 1708 |
|
| 1709 | 1709 |
# decoders / encoders |
| 1710 | 1710 |
aac_decoder_select="mdct sinewin" |
| ... | ... |
@@ -1730,7 +1731,7 @@ atrac3p_decoder_select="mdct sinewin" |
| 1730 | 1730 |
bink_decoder_select="dsputil hpeldsp" |
| 1731 | 1731 |
binkaudio_dct_decoder_select="mdct rdft dct sinewin" |
| 1732 | 1732 |
binkaudio_rdft_decoder_select="mdct rdft sinewin" |
| 1733 |
-cavs_decoder_select="dsputil golomb h264chroma videodsp" |
|
| 1733 |
+cavs_decoder_select="dsputil golomb h264chroma qpeldsp videodsp" |
|
| 1734 | 1734 |
cllc_decoder_select="dsputil" |
| 1735 | 1735 |
comfortnoise_encoder_select="lpc" |
| 1736 | 1736 |
cook_decoder_select="dsputil mdct sinewin" |
| ... | ... |
@@ -1766,7 +1767,7 @@ g2m_decoder_deps="zlib" |
| 1766 | 1766 |
g2m_decoder_select="dsputil" |
| 1767 | 1767 |
h261_decoder_select="error_resilience mpegvideo" |
| 1768 | 1768 |
h261_encoder_select="aandcttables mpegvideoenc" |
| 1769 |
-h263_decoder_select="error_resilience h263_parser h263dsp mpegvideo" |
|
| 1769 |
+h263_decoder_select="error_resilience h263_parser h263dsp mpegvideo qpeldsp" |
|
| 1770 | 1770 |
h263_encoder_select="aandcttables h263dsp mpegvideoenc" |
| 1771 | 1771 |
h263i_decoder_select="h263_decoder" |
| 1772 | 1772 |
h263p_encoder_select="h263_encoder" |
| ... | ... |
@@ -1818,7 +1819,7 @@ msmpeg4v2_decoder_select="h263_decoder" |
| 1818 | 1818 |
msmpeg4v2_encoder_select="h263_encoder" |
| 1819 | 1819 |
msmpeg4v3_decoder_select="h263_decoder" |
| 1820 | 1820 |
msmpeg4v3_encoder_select="h263_encoder" |
| 1821 |
-mss2_decoder_select="error_resilience vc1_decoder" |
|
| 1821 |
+mss2_decoder_select="error_resilience qpeldsp vc1_decoder" |
|
| 1822 | 1822 |
mxpeg_decoder_select="mjpeg_decoder" |
| 1823 | 1823 |
nellymoser_decoder_select="mdct sinewin" |
| 1824 | 1824 |
nellymoser_encoder_select="audio_frame_queue mdct sinewin" |
| ... | ... |
@@ -1860,7 +1861,7 @@ twinvq_decoder_select="mdct lsp sinewin" |
| 1860 | 1860 |
utvideo_decoder_select="dsputil" |
| 1861 | 1861 |
utvideo_encoder_select="dsputil huffman huffyuvencdsp" |
| 1862 | 1862 |
vble_decoder_select="huffyuvdsp" |
| 1863 |
-vc1_decoder_select="error_resilience h263_decoder h264chroma h264qpel intrax8" |
|
| 1863 |
+vc1_decoder_select="error_resilience h263_decoder h264chroma h264qpel intrax8 qpeldsp" |
|
| 1864 | 1864 |
vc1image_decoder_select="vc1_decoder" |
| 1865 | 1865 |
vorbis_decoder_select="mdct" |
| 1866 | 1866 |
vorbis_encoder_select="mdct" |
| ... | ... |
@@ -1936,8 +1937,8 @@ wmv3_vdpau_hwaccel_select="vc1_vdpau_hwaccel" |
| 1936 | 1936 |
|
| 1937 | 1937 |
# parsers |
| 1938 | 1938 |
h264_parser_select="h264_decoder" |
| 1939 |
-mpeg4video_parser_select="error_resilience h263dsp mpegvideo" |
|
| 1940 | 1939 |
mpegvideo_parser_select="error_resilience mpegvideo" |
| 1940 |
+mpeg4video_parser_select="error_resilience h263dsp mpegvideo qpeldsp" |
|
| 1941 | 1941 |
vc1_parser_select="mpegvideo" |
| 1942 | 1942 |
|
| 1943 | 1943 |
# external libraries |
| ... | ... |
@@ -63,6 +63,7 @@ OBJS-$(CONFIG_MPEGVIDEO) += mpegvideo.o mpegvideo_motion.o \ |
| 63 | 63 |
mpegutils.o |
| 64 | 64 |
OBJS-$(CONFIG_MPEGVIDEOENC) += mpegvideo_enc.o mpeg12data.o \ |
| 65 | 65 |
motion_est.o ratecontrol.o |
| 66 |
+OBJS-$(CONFIG_QPELDSP) += qpeldsp.o |
|
| 66 | 67 |
OBJS-$(CONFIG_RANGECODER) += rangecoder.o |
| 67 | 68 |
RDFT-OBJS-$(CONFIG_HARDCODED_TABLES) += sin_tables.o |
| 68 | 69 |
OBJS-$(CONFIG_RDFT) += rdft.o $(RDFT-OBJS-yes) |
| ... | ... |
@@ -48,9 +48,6 @@ uint32_t ff_square_tab[512] = { 0, };
|
| 48 | 48 |
#undef BIT_DEPTH |
| 49 | 49 |
|
| 50 | 50 |
#define BIT_DEPTH 8 |
| 51 |
-#include "hpel_template.c" |
|
| 52 |
-#include "tpel_template.c" |
|
| 53 |
-#include "dsputil_template.c" |
|
| 54 | 51 |
#include "dsputilenc_template.c" |
| 55 | 52 |
|
| 56 | 53 |
/* Input permutation for the simple_idct_mmx */ |
| ... | ... |
@@ -485,701 +482,6 @@ void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy, |
| 485 | 485 |
} |
| 486 | 486 |
} |
| 487 | 487 |
|
| 488 |
-#define QPEL_MC(r, OPNAME, RND, OP) \ |
|
| 489 |
-static void OPNAME ## mpeg4_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, \ |
|
| 490 |
- int dstStride, int srcStride, \ |
|
| 491 |
- int h) \ |
|
| 492 |
-{ \
|
|
| 493 |
- const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ |
|
| 494 |
- int i; \ |
|
| 495 |
- \ |
|
| 496 |
- for (i = 0; i < h; i++) { \
|
|
| 497 |
- OP(dst[0], (src[0] + src[1]) * 20 - (src[0] + src[2]) * 6 + (src[1] + src[3]) * 3 - (src[2] + src[4])); \ |
|
| 498 |
- OP(dst[1], (src[1] + src[2]) * 20 - (src[0] + src[3]) * 6 + (src[0] + src[4]) * 3 - (src[1] + src[5])); \ |
|
| 499 |
- OP(dst[2], (src[2] + src[3]) * 20 - (src[1] + src[4]) * 6 + (src[0] + src[5]) * 3 - (src[0] + src[6])); \ |
|
| 500 |
- OP(dst[3], (src[3] + src[4]) * 20 - (src[2] + src[5]) * 6 + (src[1] + src[6]) * 3 - (src[0] + src[7])); \ |
|
| 501 |
- OP(dst[4], (src[4] + src[5]) * 20 - (src[3] + src[6]) * 6 + (src[2] + src[7]) * 3 - (src[1] + src[8])); \ |
|
| 502 |
- OP(dst[5], (src[5] + src[6]) * 20 - (src[4] + src[7]) * 6 + (src[3] + src[8]) * 3 - (src[2] + src[8])); \ |
|
| 503 |
- OP(dst[6], (src[6] + src[7]) * 20 - (src[5] + src[8]) * 6 + (src[4] + src[8]) * 3 - (src[3] + src[7])); \ |
|
| 504 |
- OP(dst[7], (src[7] + src[8]) * 20 - (src[6] + src[8]) * 6 + (src[5] + src[7]) * 3 - (src[4] + src[6])); \ |
|
| 505 |
- dst += dstStride; \ |
|
| 506 |
- src += srcStride; \ |
|
| 507 |
- } \ |
|
| 508 |
-} \ |
|
| 509 |
- \ |
|
| 510 |
-static void OPNAME ## mpeg4_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, \ |
|
| 511 |
- int dstStride, int srcStride) \ |
|
| 512 |
-{ \
|
|
| 513 |
- const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ |
|
| 514 |
- const int w = 8; \ |
|
| 515 |
- int i; \ |
|
| 516 |
- \ |
|
| 517 |
- for (i = 0; i < w; i++) { \
|
|
| 518 |
- const int src0 = src[0 * srcStride]; \ |
|
| 519 |
- const int src1 = src[1 * srcStride]; \ |
|
| 520 |
- const int src2 = src[2 * srcStride]; \ |
|
| 521 |
- const int src3 = src[3 * srcStride]; \ |
|
| 522 |
- const int src4 = src[4 * srcStride]; \ |
|
| 523 |
- const int src5 = src[5 * srcStride]; \ |
|
| 524 |
- const int src6 = src[6 * srcStride]; \ |
|
| 525 |
- const int src7 = src[7 * srcStride]; \ |
|
| 526 |
- const int src8 = src[8 * srcStride]; \ |
|
| 527 |
- OP(dst[0 * dstStride], (src0 + src1) * 20 - (src0 + src2) * 6 + (src1 + src3) * 3 - (src2 + src4)); \ |
|
| 528 |
- OP(dst[1 * dstStride], (src1 + src2) * 20 - (src0 + src3) * 6 + (src0 + src4) * 3 - (src1 + src5)); \ |
|
| 529 |
- OP(dst[2 * dstStride], (src2 + src3) * 20 - (src1 + src4) * 6 + (src0 + src5) * 3 - (src0 + src6)); \ |
|
| 530 |
- OP(dst[3 * dstStride], (src3 + src4) * 20 - (src2 + src5) * 6 + (src1 + src6) * 3 - (src0 + src7)); \ |
|
| 531 |
- OP(dst[4 * dstStride], (src4 + src5) * 20 - (src3 + src6) * 6 + (src2 + src7) * 3 - (src1 + src8)); \ |
|
| 532 |
- OP(dst[5 * dstStride], (src5 + src6) * 20 - (src4 + src7) * 6 + (src3 + src8) * 3 - (src2 + src8)); \ |
|
| 533 |
- OP(dst[6 * dstStride], (src6 + src7) * 20 - (src5 + src8) * 6 + (src4 + src8) * 3 - (src3 + src7)); \ |
|
| 534 |
- OP(dst[7 * dstStride], (src7 + src8) * 20 - (src6 + src8) * 6 + (src5 + src7) * 3 - (src4 + src6)); \ |
|
| 535 |
- dst++; \ |
|
| 536 |
- src++; \ |
|
| 537 |
- } \ |
|
| 538 |
-} \ |
|
| 539 |
- \ |
|
| 540 |
-static void OPNAME ## mpeg4_qpel16_h_lowpass(uint8_t *dst, uint8_t *src, \ |
|
| 541 |
- int dstStride, int srcStride, \ |
|
| 542 |
- int h) \ |
|
| 543 |
-{ \
|
|
| 544 |
- const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ |
|
| 545 |
- int i; \ |
|
| 546 |
- \ |
|
| 547 |
- for (i = 0; i < h; i++) { \
|
|
| 548 |
- OP(dst[0], (src[0] + src[1]) * 20 - (src[0] + src[2]) * 6 + (src[1] + src[3]) * 3 - (src[2] + src[4])); \ |
|
| 549 |
- OP(dst[1], (src[1] + src[2]) * 20 - (src[0] + src[3]) * 6 + (src[0] + src[4]) * 3 - (src[1] + src[5])); \ |
|
| 550 |
- OP(dst[2], (src[2] + src[3]) * 20 - (src[1] + src[4]) * 6 + (src[0] + src[5]) * 3 - (src[0] + src[6])); \ |
|
| 551 |
- OP(dst[3], (src[3] + src[4]) * 20 - (src[2] + src[5]) * 6 + (src[1] + src[6]) * 3 - (src[0] + src[7])); \ |
|
| 552 |
- OP(dst[4], (src[4] + src[5]) * 20 - (src[3] + src[6]) * 6 + (src[2] + src[7]) * 3 - (src[1] + src[8])); \ |
|
| 553 |
- OP(dst[5], (src[5] + src[6]) * 20 - (src[4] + src[7]) * 6 + (src[3] + src[8]) * 3 - (src[2] + src[9])); \ |
|
| 554 |
- OP(dst[6], (src[6] + src[7]) * 20 - (src[5] + src[8]) * 6 + (src[4] + src[9]) * 3 - (src[3] + src[10])); \ |
|
| 555 |
- OP(dst[7], (src[7] + src[8]) * 20 - (src[6] + src[9]) * 6 + (src[5] + src[10]) * 3 - (src[4] + src[11])); \ |
|
| 556 |
- OP(dst[8], (src[8] + src[9]) * 20 - (src[7] + src[10]) * 6 + (src[6] + src[11]) * 3 - (src[5] + src[12])); \ |
|
| 557 |
- OP(dst[9], (src[9] + src[10]) * 20 - (src[8] + src[11]) * 6 + (src[7] + src[12]) * 3 - (src[6] + src[13])); \ |
|
| 558 |
- OP(dst[10], (src[10] + src[11]) * 20 - (src[9] + src[12]) * 6 + (src[8] + src[13]) * 3 - (src[7] + src[14])); \ |
|
| 559 |
- OP(dst[11], (src[11] + src[12]) * 20 - (src[10] + src[13]) * 6 + (src[9] + src[14]) * 3 - (src[8] + src[15])); \ |
|
| 560 |
- OP(dst[12], (src[12] + src[13]) * 20 - (src[11] + src[14]) * 6 + (src[10] + src[15]) * 3 - (src[9] + src[16])); \ |
|
| 561 |
- OP(dst[13], (src[13] + src[14]) * 20 - (src[12] + src[15]) * 6 + (src[11] + src[16]) * 3 - (src[10] + src[16])); \ |
|
| 562 |
- OP(dst[14], (src[14] + src[15]) * 20 - (src[13] + src[16]) * 6 + (src[12] + src[16]) * 3 - (src[11] + src[15])); \ |
|
| 563 |
- OP(dst[15], (src[15] + src[16]) * 20 - (src[14] + src[16]) * 6 + (src[13] + src[15]) * 3 - (src[12] + src[14])); \ |
|
| 564 |
- dst += dstStride; \ |
|
| 565 |
- src += srcStride; \ |
|
| 566 |
- } \ |
|
| 567 |
-} \ |
|
| 568 |
- \ |
|
| 569 |
-static void OPNAME ## mpeg4_qpel16_v_lowpass(uint8_t *dst, uint8_t *src, \ |
|
| 570 |
- int dstStride, int srcStride) \ |
|
| 571 |
-{ \
|
|
| 572 |
- const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ |
|
| 573 |
- const int w = 16; \ |
|
| 574 |
- int i; \ |
|
| 575 |
- \ |
|
| 576 |
- for (i = 0; i < w; i++) { \
|
|
| 577 |
- const int src0 = src[0 * srcStride]; \ |
|
| 578 |
- const int src1 = src[1 * srcStride]; \ |
|
| 579 |
- const int src2 = src[2 * srcStride]; \ |
|
| 580 |
- const int src3 = src[3 * srcStride]; \ |
|
| 581 |
- const int src4 = src[4 * srcStride]; \ |
|
| 582 |
- const int src5 = src[5 * srcStride]; \ |
|
| 583 |
- const int src6 = src[6 * srcStride]; \ |
|
| 584 |
- const int src7 = src[7 * srcStride]; \ |
|
| 585 |
- const int src8 = src[8 * srcStride]; \ |
|
| 586 |
- const int src9 = src[9 * srcStride]; \ |
|
| 587 |
- const int src10 = src[10 * srcStride]; \ |
|
| 588 |
- const int src11 = src[11 * srcStride]; \ |
|
| 589 |
- const int src12 = src[12 * srcStride]; \ |
|
| 590 |
- const int src13 = src[13 * srcStride]; \ |
|
| 591 |
- const int src14 = src[14 * srcStride]; \ |
|
| 592 |
- const int src15 = src[15 * srcStride]; \ |
|
| 593 |
- const int src16 = src[16 * srcStride]; \ |
|
| 594 |
- OP(dst[0 * dstStride], (src0 + src1) * 20 - (src0 + src2) * 6 + (src1 + src3) * 3 - (src2 + src4)); \ |
|
| 595 |
- OP(dst[1 * dstStride], (src1 + src2) * 20 - (src0 + src3) * 6 + (src0 + src4) * 3 - (src1 + src5)); \ |
|
| 596 |
- OP(dst[2 * dstStride], (src2 + src3) * 20 - (src1 + src4) * 6 + (src0 + src5) * 3 - (src0 + src6)); \ |
|
| 597 |
- OP(dst[3 * dstStride], (src3 + src4) * 20 - (src2 + src5) * 6 + (src1 + src6) * 3 - (src0 + src7)); \ |
|
| 598 |
- OP(dst[4 * dstStride], (src4 + src5) * 20 - (src3 + src6) * 6 + (src2 + src7) * 3 - (src1 + src8)); \ |
|
| 599 |
- OP(dst[5 * dstStride], (src5 + src6) * 20 - (src4 + src7) * 6 + (src3 + src8) * 3 - (src2 + src9)); \ |
|
| 600 |
- OP(dst[6 * dstStride], (src6 + src7) * 20 - (src5 + src8) * 6 + (src4 + src9) * 3 - (src3 + src10)); \ |
|
| 601 |
- OP(dst[7 * dstStride], (src7 + src8) * 20 - (src6 + src9) * 6 + (src5 + src10) * 3 - (src4 + src11)); \ |
|
| 602 |
- OP(dst[8 * dstStride], (src8 + src9) * 20 - (src7 + src10) * 6 + (src6 + src11) * 3 - (src5 + src12)); \ |
|
| 603 |
- OP(dst[9 * dstStride], (src9 + src10) * 20 - (src8 + src11) * 6 + (src7 + src12) * 3 - (src6 + src13)); \ |
|
| 604 |
- OP(dst[10 * dstStride], (src10 + src11) * 20 - (src9 + src12) * 6 + (src8 + src13) * 3 - (src7 + src14)); \ |
|
| 605 |
- OP(dst[11 * dstStride], (src11 + src12) * 20 - (src10 + src13) * 6 + (src9 + src14) * 3 - (src8 + src15)); \ |
|
| 606 |
- OP(dst[12 * dstStride], (src12 + src13) * 20 - (src11 + src14) * 6 + (src10 + src15) * 3 - (src9 + src16)); \ |
|
| 607 |
- OP(dst[13 * dstStride], (src13 + src14) * 20 - (src12 + src15) * 6 + (src11 + src16) * 3 - (src10 + src16)); \ |
|
| 608 |
- OP(dst[14 * dstStride], (src14 + src15) * 20 - (src13 + src16) * 6 + (src12 + src16) * 3 - (src11 + src15)); \ |
|
| 609 |
- OP(dst[15 * dstStride], (src15 + src16) * 20 - (src14 + src16) * 6 + (src13 + src15) * 3 - (src12 + src14)); \ |
|
| 610 |
- dst++; \ |
|
| 611 |
- src++; \ |
|
| 612 |
- } \ |
|
| 613 |
-} \ |
|
| 614 |
- \ |
|
| 615 |
-static void OPNAME ## qpel8_mc10_c(uint8_t *dst, uint8_t *src, \ |
|
| 616 |
- ptrdiff_t stride) \ |
|
| 617 |
-{ \
|
|
| 618 |
- uint8_t half[64]; \ |
|
| 619 |
- \ |
|
| 620 |
- put ## RND ## mpeg4_qpel8_h_lowpass(half, src, 8, stride, 8); \ |
|
| 621 |
- OPNAME ## pixels8_l2_8(dst, src, half, stride, stride, 8, 8); \ |
|
| 622 |
-} \ |
|
| 623 |
- \ |
|
| 624 |
-static void OPNAME ## qpel8_mc20_c(uint8_t *dst, uint8_t *src, \ |
|
| 625 |
- ptrdiff_t stride) \ |
|
| 626 |
-{ \
|
|
| 627 |
- OPNAME ## mpeg4_qpel8_h_lowpass(dst, src, stride, stride, 8); \ |
|
| 628 |
-} \ |
|
| 629 |
- \ |
|
| 630 |
-static void OPNAME ## qpel8_mc30_c(uint8_t *dst, uint8_t *src, \ |
|
| 631 |
- ptrdiff_t stride) \ |
|
| 632 |
-{ \
|
|
| 633 |
- uint8_t half[64]; \ |
|
| 634 |
- \ |
|
| 635 |
- put ## RND ## mpeg4_qpel8_h_lowpass(half, src, 8, stride, 8); \ |
|
| 636 |
- OPNAME ## pixels8_l2_8(dst, src + 1, half, stride, stride, 8, 8); \ |
|
| 637 |
-} \ |
|
| 638 |
- \ |
|
| 639 |
-static void OPNAME ## qpel8_mc01_c(uint8_t *dst, uint8_t *src, \ |
|
| 640 |
- ptrdiff_t stride) \ |
|
| 641 |
-{ \
|
|
| 642 |
- uint8_t full[16 * 9]; \ |
|
| 643 |
- uint8_t half[64]; \ |
|
| 644 |
- \ |
|
| 645 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 646 |
- put ## RND ## mpeg4_qpel8_v_lowpass(half, full, 8, 16); \ |
|
| 647 |
- OPNAME ## pixels8_l2_8(dst, full, half, stride, 16, 8, 8); \ |
|
| 648 |
-} \ |
|
| 649 |
- \ |
|
| 650 |
-static void OPNAME ## qpel8_mc02_c(uint8_t *dst, uint8_t *src, \ |
|
| 651 |
- ptrdiff_t stride) \ |
|
| 652 |
-{ \
|
|
| 653 |
- uint8_t full[16 * 9]; \ |
|
| 654 |
- \ |
|
| 655 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 656 |
- OPNAME ## mpeg4_qpel8_v_lowpass(dst, full, stride, 16); \ |
|
| 657 |
-} \ |
|
| 658 |
- \ |
|
| 659 |
-static void OPNAME ## qpel8_mc03_c(uint8_t *dst, uint8_t *src, \ |
|
| 660 |
- ptrdiff_t stride) \ |
|
| 661 |
-{ \
|
|
| 662 |
- uint8_t full[16 * 9]; \ |
|
| 663 |
- uint8_t half[64]; \ |
|
| 664 |
- \ |
|
| 665 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 666 |
- put ## RND ## mpeg4_qpel8_v_lowpass(half, full, 8, 16); \ |
|
| 667 |
- OPNAME ## pixels8_l2_8(dst, full + 16, half, stride, 16, 8, 8); \ |
|
| 668 |
-} \ |
|
| 669 |
- \ |
|
| 670 |
-void ff_ ## OPNAME ## qpel8_mc11_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 671 |
- ptrdiff_t stride) \ |
|
| 672 |
-{ \
|
|
| 673 |
- uint8_t full[16 * 9]; \ |
|
| 674 |
- uint8_t halfH[72]; \ |
|
| 675 |
- uint8_t halfV[64]; \ |
|
| 676 |
- uint8_t halfHV[64]; \ |
|
| 677 |
- \ |
|
| 678 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 679 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 680 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full, 8, 16); \ |
|
| 681 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 682 |
- OPNAME ## pixels8_l4_8(dst, full, halfH, halfV, halfHV, \ |
|
| 683 |
- stride, 16, 8, 8, 8, 8); \ |
|
| 684 |
-} \ |
|
| 685 |
- \ |
|
| 686 |
-static void OPNAME ## qpel8_mc11_c(uint8_t *dst, uint8_t *src, \ |
|
| 687 |
- ptrdiff_t stride) \ |
|
| 688 |
-{ \
|
|
| 689 |
- uint8_t full[16 * 9]; \ |
|
| 690 |
- uint8_t halfH[72]; \ |
|
| 691 |
- uint8_t halfHV[64]; \ |
|
| 692 |
- \ |
|
| 693 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 694 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 695 |
- put ## RND ## pixels8_l2_8(halfH, halfH, full, 8, 8, 16, 9); \ |
|
| 696 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 697 |
- OPNAME ## pixels8_l2_8(dst, halfH, halfHV, stride, 8, 8, 8); \ |
|
| 698 |
-} \ |
|
| 699 |
- \ |
|
| 700 |
-void ff_ ## OPNAME ## qpel8_mc31_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 701 |
- ptrdiff_t stride) \ |
|
| 702 |
-{ \
|
|
| 703 |
- uint8_t full[16 * 9]; \ |
|
| 704 |
- uint8_t halfH[72]; \ |
|
| 705 |
- uint8_t halfV[64]; \ |
|
| 706 |
- uint8_t halfHV[64]; \ |
|
| 707 |
- \ |
|
| 708 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 709 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 710 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full + 1, 8, 16); \ |
|
| 711 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 712 |
- OPNAME ## pixels8_l4_8(dst, full + 1, halfH, halfV, halfHV, \ |
|
| 713 |
- stride, 16, 8, 8, 8, 8); \ |
|
| 714 |
-} \ |
|
| 715 |
- \ |
|
| 716 |
-static void OPNAME ## qpel8_mc31_c(uint8_t *dst, uint8_t *src, \ |
|
| 717 |
- ptrdiff_t stride) \ |
|
| 718 |
-{ \
|
|
| 719 |
- uint8_t full[16 * 9]; \ |
|
| 720 |
- uint8_t halfH[72]; \ |
|
| 721 |
- uint8_t halfHV[64]; \ |
|
| 722 |
- \ |
|
| 723 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 724 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 725 |
- put ## RND ## pixels8_l2_8(halfH, halfH, full + 1, 8, 8, 16, 9); \ |
|
| 726 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 727 |
- OPNAME ## pixels8_l2_8(dst, halfH, halfHV, stride, 8, 8, 8); \ |
|
| 728 |
-} \ |
|
| 729 |
- \ |
|
| 730 |
-void ff_ ## OPNAME ## qpel8_mc13_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 731 |
- ptrdiff_t stride) \ |
|
| 732 |
-{ \
|
|
| 733 |
- uint8_t full[16 * 9]; \ |
|
| 734 |
- uint8_t halfH[72]; \ |
|
| 735 |
- uint8_t halfV[64]; \ |
|
| 736 |
- uint8_t halfHV[64]; \ |
|
| 737 |
- \ |
|
| 738 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 739 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 740 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full, 8, 16); \ |
|
| 741 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 742 |
- OPNAME ## pixels8_l4_8(dst, full + 16, halfH + 8, halfV, halfHV, \ |
|
| 743 |
- stride, 16, 8, 8, 8, 8); \ |
|
| 744 |
-} \ |
|
| 745 |
- \ |
|
| 746 |
-static void OPNAME ## qpel8_mc13_c(uint8_t *dst, uint8_t *src, \ |
|
| 747 |
- ptrdiff_t stride) \ |
|
| 748 |
-{ \
|
|
| 749 |
- uint8_t full[16 * 9]; \ |
|
| 750 |
- uint8_t halfH[72]; \ |
|
| 751 |
- uint8_t halfHV[64]; \ |
|
| 752 |
- \ |
|
| 753 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 754 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 755 |
- put ## RND ## pixels8_l2_8(halfH, halfH, full, 8, 8, 16, 9); \ |
|
| 756 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 757 |
- OPNAME ## pixels8_l2_8(dst, halfH + 8, halfHV, stride, 8, 8, 8); \ |
|
| 758 |
-} \ |
|
| 759 |
- \ |
|
| 760 |
-void ff_ ## OPNAME ## qpel8_mc33_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 761 |
- ptrdiff_t stride) \ |
|
| 762 |
-{ \
|
|
| 763 |
- uint8_t full[16 * 9]; \ |
|
| 764 |
- uint8_t halfH[72]; \ |
|
| 765 |
- uint8_t halfV[64]; \ |
|
| 766 |
- uint8_t halfHV[64]; \ |
|
| 767 |
- \ |
|
| 768 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 769 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 770 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full + 1, 8, 16); \ |
|
| 771 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 772 |
- OPNAME ## pixels8_l4_8(dst, full + 17, halfH + 8, halfV, halfHV, \ |
|
| 773 |
- stride, 16, 8, 8, 8, 8); \ |
|
| 774 |
-} \ |
|
| 775 |
- \ |
|
| 776 |
-static void OPNAME ## qpel8_mc33_c(uint8_t *dst, uint8_t *src, \ |
|
| 777 |
- ptrdiff_t stride) \ |
|
| 778 |
-{ \
|
|
| 779 |
- uint8_t full[16 * 9]; \ |
|
| 780 |
- uint8_t halfH[72]; \ |
|
| 781 |
- uint8_t halfHV[64]; \ |
|
| 782 |
- \ |
|
| 783 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 784 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 785 |
- put ## RND ## pixels8_l2_8(halfH, halfH, full + 1, 8, 8, 16, 9); \ |
|
| 786 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 787 |
- OPNAME ## pixels8_l2_8(dst, halfH + 8, halfHV, stride, 8, 8, 8); \ |
|
| 788 |
-} \ |
|
| 789 |
- \ |
|
| 790 |
-static void OPNAME ## qpel8_mc21_c(uint8_t *dst, uint8_t *src, \ |
|
| 791 |
- ptrdiff_t stride) \ |
|
| 792 |
-{ \
|
|
| 793 |
- uint8_t halfH[72]; \ |
|
| 794 |
- uint8_t halfHV[64]; \ |
|
| 795 |
- \ |
|
| 796 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9); \ |
|
| 797 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 798 |
- OPNAME ## pixels8_l2_8(dst, halfH, halfHV, stride, 8, 8, 8); \ |
|
| 799 |
-} \ |
|
| 800 |
- \ |
|
| 801 |
-static void OPNAME ## qpel8_mc23_c(uint8_t *dst, uint8_t *src, \ |
|
| 802 |
- ptrdiff_t stride) \ |
|
| 803 |
-{ \
|
|
| 804 |
- uint8_t halfH[72]; \ |
|
| 805 |
- uint8_t halfHV[64]; \ |
|
| 806 |
- \ |
|
| 807 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9); \ |
|
| 808 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 809 |
- OPNAME ## pixels8_l2_8(dst, halfH + 8, halfHV, stride, 8, 8, 8); \ |
|
| 810 |
-} \ |
|
| 811 |
- \ |
|
| 812 |
-void ff_ ## OPNAME ## qpel8_mc12_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 813 |
- ptrdiff_t stride) \ |
|
| 814 |
-{ \
|
|
| 815 |
- uint8_t full[16 * 9]; \ |
|
| 816 |
- uint8_t halfH[72]; \ |
|
| 817 |
- uint8_t halfV[64]; \ |
|
| 818 |
- uint8_t halfHV[64]; \ |
|
| 819 |
- \ |
|
| 820 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 821 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 822 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full, 8, 16); \ |
|
| 823 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 824 |
- OPNAME ## pixels8_l2_8(dst, halfV, halfHV, stride, 8, 8, 8); \ |
|
| 825 |
-} \ |
|
| 826 |
- \ |
|
| 827 |
-static void OPNAME ## qpel8_mc12_c(uint8_t *dst, uint8_t *src, \ |
|
| 828 |
- ptrdiff_t stride) \ |
|
| 829 |
-{ \
|
|
| 830 |
- uint8_t full[16 * 9]; \ |
|
| 831 |
- uint8_t halfH[72]; \ |
|
| 832 |
- \ |
|
| 833 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 834 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 835 |
- put ## RND ## pixels8_l2_8(halfH, halfH, full, 8, 8, 16, 9); \ |
|
| 836 |
- OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8); \ |
|
| 837 |
-} \ |
|
| 838 |
- \ |
|
| 839 |
-void ff_ ## OPNAME ## qpel8_mc32_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 840 |
- ptrdiff_t stride) \ |
|
| 841 |
-{ \
|
|
| 842 |
- uint8_t full[16 * 9]; \ |
|
| 843 |
- uint8_t halfH[72]; \ |
|
| 844 |
- uint8_t halfV[64]; \ |
|
| 845 |
- uint8_t halfHV[64]; \ |
|
| 846 |
- \ |
|
| 847 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 848 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 849 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full + 1, 8, 16); \ |
|
| 850 |
- put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 851 |
- OPNAME ## pixels8_l2_8(dst, halfV, halfHV, stride, 8, 8, 8); \ |
|
| 852 |
-} \ |
|
| 853 |
- \ |
|
| 854 |
-static void OPNAME ## qpel8_mc32_c(uint8_t *dst, uint8_t *src, \ |
|
| 855 |
- ptrdiff_t stride) \ |
|
| 856 |
-{ \
|
|
| 857 |
- uint8_t full[16 * 9]; \ |
|
| 858 |
- uint8_t halfH[72]; \ |
|
| 859 |
- \ |
|
| 860 |
- copy_block9(full, src, 16, stride, 9); \ |
|
| 861 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 862 |
- put ## RND ## pixels8_l2_8(halfH, halfH, full + 1, 8, 8, 16, 9); \ |
|
| 863 |
- OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8); \ |
|
| 864 |
-} \ |
|
| 865 |
- \ |
|
| 866 |
-static void OPNAME ## qpel8_mc22_c(uint8_t *dst, uint8_t *src, \ |
|
| 867 |
- ptrdiff_t stride) \ |
|
| 868 |
-{ \
|
|
| 869 |
- uint8_t halfH[72]; \ |
|
| 870 |
- \ |
|
| 871 |
- put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9); \ |
|
| 872 |
- OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8); \ |
|
| 873 |
-} \ |
|
| 874 |
- \ |
|
| 875 |
-static void OPNAME ## qpel16_mc10_c(uint8_t *dst, uint8_t *src, \ |
|
| 876 |
- ptrdiff_t stride) \ |
|
| 877 |
-{ \
|
|
| 878 |
- uint8_t half[256]; \ |
|
| 879 |
- \ |
|
| 880 |
- put ## RND ## mpeg4_qpel16_h_lowpass(half, src, 16, stride, 16); \ |
|
| 881 |
- OPNAME ## pixels16_l2_8(dst, src, half, stride, stride, 16, 16); \ |
|
| 882 |
-} \ |
|
| 883 |
- \ |
|
| 884 |
-static void OPNAME ## qpel16_mc20_c(uint8_t *dst, uint8_t *src, \ |
|
| 885 |
- ptrdiff_t stride) \ |
|
| 886 |
-{ \
|
|
| 887 |
- OPNAME ## mpeg4_qpel16_h_lowpass(dst, src, stride, stride, 16); \ |
|
| 888 |
-} \ |
|
| 889 |
- \ |
|
| 890 |
-static void OPNAME ## qpel16_mc30_c(uint8_t *dst, uint8_t *src, \ |
|
| 891 |
- ptrdiff_t stride) \ |
|
| 892 |
-{ \
|
|
| 893 |
- uint8_t half[256]; \ |
|
| 894 |
- \ |
|
| 895 |
- put ## RND ## mpeg4_qpel16_h_lowpass(half, src, 16, stride, 16); \ |
|
| 896 |
- OPNAME ## pixels16_l2_8(dst, src + 1, half, stride, stride, 16, 16); \ |
|
| 897 |
-} \ |
|
| 898 |
- \ |
|
| 899 |
-static void OPNAME ## qpel16_mc01_c(uint8_t *dst, uint8_t *src, \ |
|
| 900 |
- ptrdiff_t stride) \ |
|
| 901 |
-{ \
|
|
| 902 |
- uint8_t full[24 * 17]; \ |
|
| 903 |
- uint8_t half[256]; \ |
|
| 904 |
- \ |
|
| 905 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 906 |
- put ## RND ## mpeg4_qpel16_v_lowpass(half, full, 16, 24); \ |
|
| 907 |
- OPNAME ## pixels16_l2_8(dst, full, half, stride, 24, 16, 16); \ |
|
| 908 |
-} \ |
|
| 909 |
- \ |
|
| 910 |
-static void OPNAME ## qpel16_mc02_c(uint8_t *dst, uint8_t *src, \ |
|
| 911 |
- ptrdiff_t stride) \ |
|
| 912 |
-{ \
|
|
| 913 |
- uint8_t full[24 * 17]; \ |
|
| 914 |
- \ |
|
| 915 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 916 |
- OPNAME ## mpeg4_qpel16_v_lowpass(dst, full, stride, 24); \ |
|
| 917 |
-} \ |
|
| 918 |
- \ |
|
| 919 |
-static void OPNAME ## qpel16_mc03_c(uint8_t *dst, uint8_t *src, \ |
|
| 920 |
- ptrdiff_t stride) \ |
|
| 921 |
-{ \
|
|
| 922 |
- uint8_t full[24 * 17]; \ |
|
| 923 |
- uint8_t half[256]; \ |
|
| 924 |
- \ |
|
| 925 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 926 |
- put ## RND ## mpeg4_qpel16_v_lowpass(half, full, 16, 24); \ |
|
| 927 |
- OPNAME ## pixels16_l2_8(dst, full + 24, half, stride, 24, 16, 16); \ |
|
| 928 |
-} \ |
|
| 929 |
- \ |
|
| 930 |
-void ff_ ## OPNAME ## qpel16_mc11_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 931 |
- ptrdiff_t stride) \ |
|
| 932 |
-{ \
|
|
| 933 |
- uint8_t full[24 * 17]; \ |
|
| 934 |
- uint8_t halfH[272]; \ |
|
| 935 |
- uint8_t halfV[256]; \ |
|
| 936 |
- uint8_t halfHV[256]; \ |
|
| 937 |
- \ |
|
| 938 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 939 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 940 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full, 16, 24); \ |
|
| 941 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 942 |
- OPNAME ## pixels16_l4_8(dst, full, halfH, halfV, halfHV, \ |
|
| 943 |
- stride, 24, 16, 16, 16, 16); \ |
|
| 944 |
-} \ |
|
| 945 |
- \ |
|
| 946 |
-static void OPNAME ## qpel16_mc11_c(uint8_t *dst, uint8_t *src, \ |
|
| 947 |
- ptrdiff_t stride) \ |
|
| 948 |
-{ \
|
|
| 949 |
- uint8_t full[24 * 17]; \ |
|
| 950 |
- uint8_t halfH[272]; \ |
|
| 951 |
- uint8_t halfHV[256]; \ |
|
| 952 |
- \ |
|
| 953 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 954 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 955 |
- put ## RND ## pixels16_l2_8(halfH, halfH, full, 16, 16, 24, 17); \ |
|
| 956 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 957 |
- OPNAME ## pixels16_l2_8(dst, halfH, halfHV, stride, 16, 16, 16); \ |
|
| 958 |
-} \ |
|
| 959 |
- \ |
|
| 960 |
-void ff_ ## OPNAME ## qpel16_mc31_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 961 |
- ptrdiff_t stride) \ |
|
| 962 |
-{ \
|
|
| 963 |
- uint8_t full[24 * 17]; \ |
|
| 964 |
- uint8_t halfH[272]; \ |
|
| 965 |
- uint8_t halfV[256]; \ |
|
| 966 |
- uint8_t halfHV[256]; \ |
|
| 967 |
- \ |
|
| 968 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 969 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 970 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full + 1, 16, 24); \ |
|
| 971 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 972 |
- OPNAME ## pixels16_l4_8(dst, full + 1, halfH, halfV, halfHV, \ |
|
| 973 |
- stride, 24, 16, 16, 16, 16); \ |
|
| 974 |
-} \ |
|
| 975 |
- \ |
|
| 976 |
-static void OPNAME ## qpel16_mc31_c(uint8_t *dst, uint8_t *src, \ |
|
| 977 |
- ptrdiff_t stride) \ |
|
| 978 |
-{ \
|
|
| 979 |
- uint8_t full[24 * 17]; \ |
|
| 980 |
- uint8_t halfH[272]; \ |
|
| 981 |
- uint8_t halfHV[256]; \ |
|
| 982 |
- \ |
|
| 983 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 984 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 985 |
- put ## RND ## pixels16_l2_8(halfH, halfH, full + 1, 16, 16, 24, 17); \ |
|
| 986 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 987 |
- OPNAME ## pixels16_l2_8(dst, halfH, halfHV, stride, 16, 16, 16); \ |
|
| 988 |
-} \ |
|
| 989 |
- \ |
|
| 990 |
-void ff_ ## OPNAME ## qpel16_mc13_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 991 |
- ptrdiff_t stride) \ |
|
| 992 |
-{ \
|
|
| 993 |
- uint8_t full[24 * 17]; \ |
|
| 994 |
- uint8_t halfH[272]; \ |
|
| 995 |
- uint8_t halfV[256]; \ |
|
| 996 |
- uint8_t halfHV[256]; \ |
|
| 997 |
- \ |
|
| 998 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 999 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 1000 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full, 16, 24); \ |
|
| 1001 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 1002 |
- OPNAME ## pixels16_l4_8(dst, full + 24, halfH + 16, halfV, halfHV, \ |
|
| 1003 |
- stride, 24, 16, 16, 16, 16); \ |
|
| 1004 |
-} \ |
|
| 1005 |
- \ |
|
| 1006 |
-static void OPNAME ## qpel16_mc13_c(uint8_t *dst, uint8_t *src, \ |
|
| 1007 |
- ptrdiff_t stride) \ |
|
| 1008 |
-{ \
|
|
| 1009 |
- uint8_t full[24 * 17]; \ |
|
| 1010 |
- uint8_t halfH[272]; \ |
|
| 1011 |
- uint8_t halfHV[256]; \ |
|
| 1012 |
- \ |
|
| 1013 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 1014 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 1015 |
- put ## RND ## pixels16_l2_8(halfH, halfH, full, 16, 16, 24, 17); \ |
|
| 1016 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 1017 |
- OPNAME ## pixels16_l2_8(dst, halfH + 16, halfHV, stride, 16, 16, 16); \ |
|
| 1018 |
-} \ |
|
| 1019 |
- \ |
|
| 1020 |
-void ff_ ## OPNAME ## qpel16_mc33_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 1021 |
- ptrdiff_t stride) \ |
|
| 1022 |
-{ \
|
|
| 1023 |
- uint8_t full[24 * 17]; \ |
|
| 1024 |
- uint8_t halfH[272]; \ |
|
| 1025 |
- uint8_t halfV[256]; \ |
|
| 1026 |
- uint8_t halfHV[256]; \ |
|
| 1027 |
- \ |
|
| 1028 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 1029 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 1030 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full + 1, 16, 24); \ |
|
| 1031 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 1032 |
- OPNAME ## pixels16_l4_8(dst, full + 25, halfH + 16, halfV, halfHV, \ |
|
| 1033 |
- stride, 24, 16, 16, 16, 16); \ |
|
| 1034 |
-} \ |
|
| 1035 |
- \ |
|
| 1036 |
-static void OPNAME ## qpel16_mc33_c(uint8_t *dst, uint8_t *src, \ |
|
| 1037 |
- ptrdiff_t stride) \ |
|
| 1038 |
-{ \
|
|
| 1039 |
- uint8_t full[24 * 17]; \ |
|
| 1040 |
- uint8_t halfH[272]; \ |
|
| 1041 |
- uint8_t halfHV[256]; \ |
|
| 1042 |
- \ |
|
| 1043 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 1044 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 1045 |
- put ## RND ## pixels16_l2_8(halfH, halfH, full + 1, 16, 16, 24, 17); \ |
|
| 1046 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 1047 |
- OPNAME ## pixels16_l2_8(dst, halfH + 16, halfHV, stride, 16, 16, 16); \ |
|
| 1048 |
-} \ |
|
| 1049 |
- \ |
|
| 1050 |
-static void OPNAME ## qpel16_mc21_c(uint8_t *dst, uint8_t *src, \ |
|
| 1051 |
- ptrdiff_t stride) \ |
|
| 1052 |
-{ \
|
|
| 1053 |
- uint8_t halfH[272]; \ |
|
| 1054 |
- uint8_t halfHV[256]; \ |
|
| 1055 |
- \ |
|
| 1056 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17); \ |
|
| 1057 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 1058 |
- OPNAME ## pixels16_l2_8(dst, halfH, halfHV, stride, 16, 16, 16); \ |
|
| 1059 |
-} \ |
|
| 1060 |
- \ |
|
| 1061 |
-static void OPNAME ## qpel16_mc23_c(uint8_t *dst, uint8_t *src, \ |
|
| 1062 |
- ptrdiff_t stride) \ |
|
| 1063 |
-{ \
|
|
| 1064 |
- uint8_t halfH[272]; \ |
|
| 1065 |
- uint8_t halfHV[256]; \ |
|
| 1066 |
- \ |
|
| 1067 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17); \ |
|
| 1068 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 1069 |
- OPNAME ## pixels16_l2_8(dst, halfH + 16, halfHV, stride, 16, 16, 16); \ |
|
| 1070 |
-} \ |
|
| 1071 |
- \ |
|
| 1072 |
-void ff_ ## OPNAME ## qpel16_mc12_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 1073 |
- ptrdiff_t stride) \ |
|
| 1074 |
-{ \
|
|
| 1075 |
- uint8_t full[24 * 17]; \ |
|
| 1076 |
- uint8_t halfH[272]; \ |
|
| 1077 |
- uint8_t halfV[256]; \ |
|
| 1078 |
- uint8_t halfHV[256]; \ |
|
| 1079 |
- \ |
|
| 1080 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 1081 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 1082 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full, 16, 24); \ |
|
| 1083 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 1084 |
- OPNAME ## pixels16_l2_8(dst, halfV, halfHV, stride, 16, 16, 16); \ |
|
| 1085 |
-} \ |
|
| 1086 |
- \ |
|
| 1087 |
-static void OPNAME ## qpel16_mc12_c(uint8_t *dst, uint8_t *src, \ |
|
| 1088 |
- ptrdiff_t stride) \ |
|
| 1089 |
-{ \
|
|
| 1090 |
- uint8_t full[24 * 17]; \ |
|
| 1091 |
- uint8_t halfH[272]; \ |
|
| 1092 |
- \ |
|
| 1093 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 1094 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 1095 |
- put ## RND ## pixels16_l2_8(halfH, halfH, full, 16, 16, 24, 17); \ |
|
| 1096 |
- OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16); \ |
|
| 1097 |
-} \ |
|
| 1098 |
- \ |
|
| 1099 |
-void ff_ ## OPNAME ## qpel16_mc32_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 1100 |
- ptrdiff_t stride) \ |
|
| 1101 |
-{ \
|
|
| 1102 |
- uint8_t full[24 * 17]; \ |
|
| 1103 |
- uint8_t halfH[272]; \ |
|
| 1104 |
- uint8_t halfV[256]; \ |
|
| 1105 |
- uint8_t halfHV[256]; \ |
|
| 1106 |
- \ |
|
| 1107 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 1108 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 1109 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full + 1, 16, 24); \ |
|
| 1110 |
- put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 1111 |
- OPNAME ## pixels16_l2_8(dst, halfV, halfHV, stride, 16, 16, 16); \ |
|
| 1112 |
-} \ |
|
| 1113 |
- \ |
|
| 1114 |
-static void OPNAME ## qpel16_mc32_c(uint8_t *dst, uint8_t *src, \ |
|
| 1115 |
- ptrdiff_t stride) \ |
|
| 1116 |
-{ \
|
|
| 1117 |
- uint8_t full[24 * 17]; \ |
|
| 1118 |
- uint8_t halfH[272]; \ |
|
| 1119 |
- \ |
|
| 1120 |
- copy_block17(full, src, 24, stride, 17); \ |
|
| 1121 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 1122 |
- put ## RND ## pixels16_l2_8(halfH, halfH, full + 1, 16, 16, 24, 17); \ |
|
| 1123 |
- OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16); \ |
|
| 1124 |
-} \ |
|
| 1125 |
- \ |
|
| 1126 |
-static void OPNAME ## qpel16_mc22_c(uint8_t *dst, uint8_t *src, \ |
|
| 1127 |
- ptrdiff_t stride) \ |
|
| 1128 |
-{ \
|
|
| 1129 |
- uint8_t halfH[272]; \ |
|
| 1130 |
- \ |
|
| 1131 |
- put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17); \ |
|
| 1132 |
- OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16); \ |
|
| 1133 |
-} |
|
| 1134 |
- |
|
| 1135 |
-#define op_avg(a, b) a = (((a) + cm[((b) + 16) >> 5] + 1) >> 1) |
|
| 1136 |
-#define op_avg_no_rnd(a, b) a = (((a) + cm[((b) + 15) >> 5]) >> 1) |
|
| 1137 |
-#define op_put(a, b) a = cm[((b) + 16) >> 5] |
|
| 1138 |
-#define op_put_no_rnd(a, b) a = cm[((b) + 15) >> 5] |
|
| 1139 |
- |
|
| 1140 |
-QPEL_MC(0, put_, _, op_put) |
|
| 1141 |
-QPEL_MC(1, put_no_rnd_, _no_rnd_, op_put_no_rnd) |
|
| 1142 |
-QPEL_MC(0, avg_, _, op_avg) |
|
| 1143 |
- |
|
| 1144 |
-#undef op_avg |
|
| 1145 |
-#undef op_put |
|
| 1146 |
-#undef op_put_no_rnd |
|
| 1147 |
- |
|
| 1148 |
-void ff_put_pixels8x8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) |
|
| 1149 |
-{
|
|
| 1150 |
- put_pixels8_8_c(dst, src, stride, 8); |
|
| 1151 |
-} |
|
| 1152 |
- |
|
| 1153 |
-void ff_avg_pixels8x8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) |
|
| 1154 |
-{
|
|
| 1155 |
- avg_pixels8_8_c(dst, src, stride, 8); |
|
| 1156 |
-} |
|
| 1157 |
- |
|
| 1158 |
-void ff_put_pixels16x16_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) |
|
| 1159 |
-{
|
|
| 1160 |
- put_pixels16_8_c(dst, src, stride, 16); |
|
| 1161 |
-} |
|
| 1162 |
- |
|
| 1163 |
-void ff_avg_pixels16x16_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) |
|
| 1164 |
-{
|
|
| 1165 |
- avg_pixels16_8_c(dst, src, stride, 16); |
|
| 1166 |
-} |
|
| 1167 |
- |
|
| 1168 |
-#define put_qpel8_mc00_c ff_put_pixels8x8_c |
|
| 1169 |
-#define avg_qpel8_mc00_c ff_avg_pixels8x8_c |
|
| 1170 |
-#define put_qpel16_mc00_c ff_put_pixels16x16_c |
|
| 1171 |
-#define avg_qpel16_mc00_c ff_avg_pixels16x16_c |
|
| 1172 |
-#define put_no_rnd_qpel8_mc00_c ff_put_pixels8x8_c |
|
| 1173 |
-#define put_no_rnd_qpel16_mc00_c ff_put_pixels16x16_c |
|
| 1174 |
- |
|
| 1175 |
-void ff_put_pixels8_l2_8(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, |
|
| 1176 |
- int dst_stride, int src_stride1, int src_stride2, |
|
| 1177 |
- int h) |
|
| 1178 |
-{
|
|
| 1179 |
- put_pixels8_l2_8(dst, src1, src2, dst_stride, src_stride1, src_stride2, h); |
|
| 1180 |
- |
|
| 1181 |
-} |
|
| 1182 |
- |
|
| 1183 | 488 |
static inline int pix_abs16_c(MpegEncContext *v, uint8_t *pix1, uint8_t *pix2, |
| 1184 | 489 |
int line_size, int h) |
| 1185 | 490 |
{
|
| ... | ... |
@@ -2198,35 +1500,6 @@ av_cold void ff_dsputil_init(DSPContext *c, AVCodecContext *avctx) |
| 2198 | 2198 |
c->pix_abs[1][2] = pix_abs8_y2_c; |
| 2199 | 2199 |
c->pix_abs[1][3] = pix_abs8_xy2_c; |
| 2200 | 2200 |
|
| 2201 |
-#define dspfunc(PFX, IDX, NUM) \ |
|
| 2202 |
- c->PFX ## _pixels_tab[IDX][0] = PFX ## NUM ## _mc00_c; \ |
|
| 2203 |
- c->PFX ## _pixels_tab[IDX][1] = PFX ## NUM ## _mc10_c; \ |
|
| 2204 |
- c->PFX ## _pixels_tab[IDX][2] = PFX ## NUM ## _mc20_c; \ |
|
| 2205 |
- c->PFX ## _pixels_tab[IDX][3] = PFX ## NUM ## _mc30_c; \ |
|
| 2206 |
- c->PFX ## _pixels_tab[IDX][4] = PFX ## NUM ## _mc01_c; \ |
|
| 2207 |
- c->PFX ## _pixels_tab[IDX][5] = PFX ## NUM ## _mc11_c; \ |
|
| 2208 |
- c->PFX ## _pixels_tab[IDX][6] = PFX ## NUM ## _mc21_c; \ |
|
| 2209 |
- c->PFX ## _pixels_tab[IDX][7] = PFX ## NUM ## _mc31_c; \ |
|
| 2210 |
- c->PFX ## _pixels_tab[IDX][8] = PFX ## NUM ## _mc02_c; \ |
|
| 2211 |
- c->PFX ## _pixels_tab[IDX][9] = PFX ## NUM ## _mc12_c; \ |
|
| 2212 |
- c->PFX ## _pixels_tab[IDX][10] = PFX ## NUM ## _mc22_c; \ |
|
| 2213 |
- c->PFX ## _pixels_tab[IDX][11] = PFX ## NUM ## _mc32_c; \ |
|
| 2214 |
- c->PFX ## _pixels_tab[IDX][12] = PFX ## NUM ## _mc03_c; \ |
|
| 2215 |
- c->PFX ## _pixels_tab[IDX][13] = PFX ## NUM ## _mc13_c; \ |
|
| 2216 |
- c->PFX ## _pixels_tab[IDX][14] = PFX ## NUM ## _mc23_c; \ |
|
| 2217 |
- c->PFX ## _pixels_tab[IDX][15] = PFX ## NUM ## _mc33_c |
|
| 2218 |
- |
|
| 2219 |
- dspfunc(put_qpel, 0, 16); |
|
| 2220 |
- dspfunc(put_qpel, 1, 8); |
|
| 2221 |
- |
|
| 2222 |
- dspfunc(put_no_rnd_qpel, 0, 16); |
|
| 2223 |
- dspfunc(put_no_rnd_qpel, 1, 8); |
|
| 2224 |
- |
|
| 2225 |
- dspfunc(avg_qpel, 0, 16); |
|
| 2226 |
- dspfunc(avg_qpel, 1, 8); |
|
| 2227 |
- |
|
| 2228 |
-#undef dspfunc |
|
| 2229 |
- |
|
| 2230 | 2201 |
#define SET_CMP_FUNC(name) \ |
| 2231 | 2202 |
c->name[0] = name ## 16_c; \ |
| 2232 | 2203 |
c->name[1] = name ## 8x8_c; |
| ... | ... |
@@ -34,15 +34,6 @@ |
| 34 | 34 |
|
| 35 | 35 |
extern uint32_t ff_square_tab[512]; |
| 36 | 36 |
|
| 37 |
-void ff_put_pixels8x8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); |
|
| 38 |
-void ff_avg_pixels8x8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); |
|
| 39 |
-void ff_put_pixels16x16_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); |
|
| 40 |
-void ff_avg_pixels16x16_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); |
|
| 41 |
- |
|
| 42 |
-void ff_put_pixels8_l2_8(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, |
|
| 43 |
- int dst_stride, int src_stride1, int src_stride2, |
|
| 44 |
- int h); |
|
| 45 |
- |
|
| 46 | 37 |
void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy, |
| 47 | 38 |
int dxx, int dxy, int dyx, int dyy, int shift, int r, |
| 48 | 39 |
int width, int height); |
| ... | ... |
@@ -64,33 +55,9 @@ void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy, |
| 64 | 64 |
* Block sizes for op_pixels_func are 8x4,8x8 16x8 16x16. |
| 65 | 65 |
* h for op_pixels_func is limited to { width / 2, width },
|
| 66 | 66 |
* but never larger than 16 and never smaller than 4. */ |
| 67 |
-typedef void (*qpel_mc_func)(uint8_t *dst /* align width (8 or 16) */, |
|
| 68 |
- uint8_t *src /* align 1 */, ptrdiff_t stride); |
|
| 69 |
- |
|
| 70 | 67 |
typedef void (*op_fill_func)(uint8_t *block /* align width (8 or 16) */, |
| 71 | 68 |
uint8_t value, int line_size, int h); |
| 72 | 69 |
|
| 73 |
-#define DEF_OLD_QPEL(name) \ |
|
| 74 |
- void ff_put_ ## name(uint8_t *dst /* align width (8 or 16) */, \ |
|
| 75 |
- uint8_t *src /* align 1 */, ptrdiff_t stride); \ |
|
| 76 |
- void ff_put_no_rnd_ ## name(uint8_t *dst /* align width (8 or 16) */, \ |
|
| 77 |
- uint8_t *src /* align 1 */, ptrdiff_t stride); \ |
|
| 78 |
- void ff_avg_ ## name(uint8_t *dst /* align width (8 or 16) */, \ |
|
| 79 |
- uint8_t *src /* align 1 */, ptrdiff_t stride); |
|
| 80 |
- |
|
| 81 |
-DEF_OLD_QPEL(qpel16_mc11_old_c) |
|
| 82 |
-DEF_OLD_QPEL(qpel16_mc31_old_c) |
|
| 83 |
-DEF_OLD_QPEL(qpel16_mc12_old_c) |
|
| 84 |
-DEF_OLD_QPEL(qpel16_mc32_old_c) |
|
| 85 |
-DEF_OLD_QPEL(qpel16_mc13_old_c) |
|
| 86 |
-DEF_OLD_QPEL(qpel16_mc33_old_c) |
|
| 87 |
-DEF_OLD_QPEL(qpel8_mc11_old_c) |
|
| 88 |
-DEF_OLD_QPEL(qpel8_mc31_old_c) |
|
| 89 |
-DEF_OLD_QPEL(qpel8_mc12_old_c) |
|
| 90 |
-DEF_OLD_QPEL(qpel8_mc32_old_c) |
|
| 91 |
-DEF_OLD_QPEL(qpel8_mc13_old_c) |
|
| 92 |
-DEF_OLD_QPEL(qpel8_mc33_old_c) |
|
| 93 |
- |
|
| 94 | 70 |
struct MpegEncContext; |
| 95 | 71 |
/* Motion estimation: |
| 96 | 72 |
* h is limited to { width / 2, width, 2 * width },
|
| ... | ... |
@@ -174,10 +141,6 @@ typedef struct DSPContext {
|
| 174 | 174 |
me_cmp_func ildct_cmp[6]; // only width 16 used |
| 175 | 175 |
me_cmp_func frame_skip_cmp[6]; // only width 8 used |
| 176 | 176 |
|
| 177 |
- qpel_mc_func put_qpel_pixels_tab[2][16]; |
|
| 178 |
- qpel_mc_func avg_qpel_pixels_tab[2][16]; |
|
| 179 |
- qpel_mc_func put_no_rnd_qpel_pixels_tab[2][16]; |
|
| 180 |
- |
|
| 181 | 177 |
me_cmp_func pix_abs[2][4]; |
| 182 | 178 |
|
| 183 | 179 |
void (*bswap_buf)(uint32_t *dst, const uint32_t *src, int w); |
| 184 | 180 |
deleted file mode 100644 |
| ... | ... |
@@ -1,223 +0,0 @@ |
| 1 |
-/* |
|
| 2 |
- * DSP utils |
|
| 3 |
- * Copyright (c) 2000, 2001 Fabrice Bellard |
|
| 4 |
- * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at> |
|
| 5 |
- * |
|
| 6 |
- * gmc & q-pel & 32/64 bit based MC by Michael Niedermayer <michaelni@gmx.at> |
|
| 7 |
- * |
|
| 8 |
- * This file is part of Libav. |
|
| 9 |
- * |
|
| 10 |
- * Libav is free software; you can redistribute it and/or |
|
| 11 |
- * modify it under the terms of the GNU Lesser General Public |
|
| 12 |
- * License as published by the Free Software Foundation; either |
|
| 13 |
- * version 2.1 of the License, or (at your option) any later version. |
|
| 14 |
- * |
|
| 15 |
- * Libav is distributed in the hope that it will be useful, |
|
| 16 |
- * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
| 17 |
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
| 18 |
- * Lesser General Public License for more details. |
|
| 19 |
- * |
|
| 20 |
- * You should have received a copy of the GNU Lesser General Public |
|
| 21 |
- * License along with Libav; if not, write to the Free Software |
|
| 22 |
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
| 23 |
- */ |
|
| 24 |
- |
|
| 25 |
-/** |
|
| 26 |
- * @file |
|
| 27 |
- * DSP utils |
|
| 28 |
- */ |
|
| 29 |
- |
|
| 30 |
-#define PIXOP2(OPNAME, OP) \ |
|
| 31 |
-static inline void OPNAME ## _no_rnd_pixels8_l2_8(uint8_t *dst, \ |
|
| 32 |
- const uint8_t *src1, \ |
|
| 33 |
- const uint8_t *src2, \ |
|
| 34 |
- int dst_stride, \ |
|
| 35 |
- int src_stride1, \ |
|
| 36 |
- int src_stride2, \ |
|
| 37 |
- int h) \ |
|
| 38 |
-{ \
|
|
| 39 |
- int i; \ |
|
| 40 |
- \ |
|
| 41 |
- for (i = 0; i < h; i++) { \
|
|
| 42 |
- uint32_t a, b; \ |
|
| 43 |
- a = AV_RN32(&src1[i * src_stride1]); \ |
|
| 44 |
- b = AV_RN32(&src2[i * src_stride2]); \ |
|
| 45 |
- OP(*((uint32_t *) &dst[i * dst_stride]), \ |
|
| 46 |
- no_rnd_avg32(a, b)); \ |
|
| 47 |
- a = AV_RN32(&src1[i * src_stride1 + 4]); \ |
|
| 48 |
- b = AV_RN32(&src2[i * src_stride2 + 4]); \ |
|
| 49 |
- OP(*((uint32_t *) &dst[i * dst_stride + 4]), \ |
|
| 50 |
- no_rnd_avg32(a, b)); \ |
|
| 51 |
- } \ |
|
| 52 |
-} \ |
|
| 53 |
- \ |
|
| 54 |
-static inline void OPNAME ## _no_rnd_pixels16_l2_8(uint8_t *dst, \ |
|
| 55 |
- const uint8_t *src1, \ |
|
| 56 |
- const uint8_t *src2, \ |
|
| 57 |
- int dst_stride, \ |
|
| 58 |
- int src_stride1, \ |
|
| 59 |
- int src_stride2, \ |
|
| 60 |
- int h) \ |
|
| 61 |
-{ \
|
|
| 62 |
- OPNAME ## _no_rnd_pixels8_l2_8(dst, src1, src2, dst_stride, \ |
|
| 63 |
- src_stride1, src_stride2, h); \ |
|
| 64 |
- OPNAME ## _no_rnd_pixels8_l2_8(dst + 8, \ |
|
| 65 |
- src1 + 8, \ |
|
| 66 |
- src2 + 8, \ |
|
| 67 |
- dst_stride, src_stride1, \ |
|
| 68 |
- src_stride2, h); \ |
|
| 69 |
-} \ |
|
| 70 |
- \ |
|
| 71 |
-static inline void OPNAME ## _pixels8_l4_8(uint8_t *dst, \ |
|
| 72 |
- const uint8_t *src1, \ |
|
| 73 |
- const uint8_t *src2, \ |
|
| 74 |
- const uint8_t *src3, \ |
|
| 75 |
- const uint8_t *src4, \ |
|
| 76 |
- int dst_stride, \ |
|
| 77 |
- int src_stride1, \ |
|
| 78 |
- int src_stride2, \ |
|
| 79 |
- int src_stride3, \ |
|
| 80 |
- int src_stride4, \ |
|
| 81 |
- int h) \ |
|
| 82 |
-{ \
|
|
| 83 |
- /* FIXME HIGH BIT DEPTH */ \ |
|
| 84 |
- int i; \ |
|
| 85 |
- \ |
|
| 86 |
- for (i = 0; i < h; i++) { \
|
|
| 87 |
- uint32_t a, b, c, d, l0, l1, h0, h1; \ |
|
| 88 |
- a = AV_RN32(&src1[i * src_stride1]); \ |
|
| 89 |
- b = AV_RN32(&src2[i * src_stride2]); \ |
|
| 90 |
- c = AV_RN32(&src3[i * src_stride3]); \ |
|
| 91 |
- d = AV_RN32(&src4[i * src_stride4]); \ |
|
| 92 |
- l0 = (a & 0x03030303UL) + \ |
|
| 93 |
- (b & 0x03030303UL) + \ |
|
| 94 |
- 0x02020202UL; \ |
|
| 95 |
- h0 = ((a & 0xFCFCFCFCUL) >> 2) + \ |
|
| 96 |
- ((b & 0xFCFCFCFCUL) >> 2); \ |
|
| 97 |
- l1 = (c & 0x03030303UL) + \ |
|
| 98 |
- (d & 0x03030303UL); \ |
|
| 99 |
- h1 = ((c & 0xFCFCFCFCUL) >> 2) + \ |
|
| 100 |
- ((d & 0xFCFCFCFCUL) >> 2); \ |
|
| 101 |
- OP(*((uint32_t *) &dst[i * dst_stride]), \ |
|
| 102 |
- h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL)); \ |
|
| 103 |
- a = AV_RN32(&src1[i * src_stride1 + 4]); \ |
|
| 104 |
- b = AV_RN32(&src2[i * src_stride2 + 4]); \ |
|
| 105 |
- c = AV_RN32(&src3[i * src_stride3 + 4]); \ |
|
| 106 |
- d = AV_RN32(&src4[i * src_stride4 + 4]); \ |
|
| 107 |
- l0 = (a & 0x03030303UL) + \ |
|
| 108 |
- (b & 0x03030303UL) + \ |
|
| 109 |
- 0x02020202UL; \ |
|
| 110 |
- h0 = ((a & 0xFCFCFCFCUL) >> 2) + \ |
|
| 111 |
- ((b & 0xFCFCFCFCUL) >> 2); \ |
|
| 112 |
- l1 = (c & 0x03030303UL) + \ |
|
| 113 |
- (d & 0x03030303UL); \ |
|
| 114 |
- h1 = ((c & 0xFCFCFCFCUL) >> 2) + \ |
|
| 115 |
- ((d & 0xFCFCFCFCUL) >> 2); \ |
|
| 116 |
- OP(*((uint32_t *) &dst[i * dst_stride + 4]), \ |
|
| 117 |
- h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL)); \ |
|
| 118 |
- } \ |
|
| 119 |
-} \ |
|
| 120 |
- \ |
|
| 121 |
-static inline void OPNAME ## _no_rnd_pixels8_l4_8(uint8_t *dst, \ |
|
| 122 |
- const uint8_t *src1, \ |
|
| 123 |
- const uint8_t *src2, \ |
|
| 124 |
- const uint8_t *src3, \ |
|
| 125 |
- const uint8_t *src4, \ |
|
| 126 |
- int dst_stride, \ |
|
| 127 |
- int src_stride1, \ |
|
| 128 |
- int src_stride2, \ |
|
| 129 |
- int src_stride3, \ |
|
| 130 |
- int src_stride4, \ |
|
| 131 |
- int h) \ |
|
| 132 |
-{ \
|
|
| 133 |
- /* FIXME HIGH BIT DEPTH */ \ |
|
| 134 |
- int i; \ |
|
| 135 |
- \ |
|
| 136 |
- for (i = 0; i < h; i++) { \
|
|
| 137 |
- uint32_t a, b, c, d, l0, l1, h0, h1; \ |
|
| 138 |
- a = AV_RN32(&src1[i * src_stride1]); \ |
|
| 139 |
- b = AV_RN32(&src2[i * src_stride2]); \ |
|
| 140 |
- c = AV_RN32(&src3[i * src_stride3]); \ |
|
| 141 |
- d = AV_RN32(&src4[i * src_stride4]); \ |
|
| 142 |
- l0 = (a & 0x03030303UL) + \ |
|
| 143 |
- (b & 0x03030303UL) + \ |
|
| 144 |
- 0x01010101UL; \ |
|
| 145 |
- h0 = ((a & 0xFCFCFCFCUL) >> 2) + \ |
|
| 146 |
- ((b & 0xFCFCFCFCUL) >> 2); \ |
|
| 147 |
- l1 = (c & 0x03030303UL) + \ |
|
| 148 |
- (d & 0x03030303UL); \ |
|
| 149 |
- h1 = ((c & 0xFCFCFCFCUL) >> 2) + \ |
|
| 150 |
- ((d & 0xFCFCFCFCUL) >> 2); \ |
|
| 151 |
- OP(*((uint32_t *) &dst[i * dst_stride]), \ |
|
| 152 |
- h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL)); \ |
|
| 153 |
- a = AV_RN32(&src1[i * src_stride1 + 4]); \ |
|
| 154 |
- b = AV_RN32(&src2[i * src_stride2 + 4]); \ |
|
| 155 |
- c = AV_RN32(&src3[i * src_stride3 + 4]); \ |
|
| 156 |
- d = AV_RN32(&src4[i * src_stride4 + 4]); \ |
|
| 157 |
- l0 = (a & 0x03030303UL) + \ |
|
| 158 |
- (b & 0x03030303UL) + \ |
|
| 159 |
- 0x01010101UL; \ |
|
| 160 |
- h0 = ((a & 0xFCFCFCFCUL) >> 2) + \ |
|
| 161 |
- ((b & 0xFCFCFCFCUL) >> 2); \ |
|
| 162 |
- l1 = (c & 0x03030303UL) + \ |
|
| 163 |
- (d & 0x03030303UL); \ |
|
| 164 |
- h1 = ((c & 0xFCFCFCFCUL) >> 2) + \ |
|
| 165 |
- ((d & 0xFCFCFCFCUL) >> 2); \ |
|
| 166 |
- OP(*((uint32_t *) &dst[i * dst_stride + 4]), \ |
|
| 167 |
- h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL)); \ |
|
| 168 |
- } \ |
|
| 169 |
-} \ |
|
| 170 |
- \ |
|
| 171 |
-static inline void OPNAME ## _pixels16_l4_8(uint8_t *dst, \ |
|
| 172 |
- const uint8_t *src1, \ |
|
| 173 |
- const uint8_t *src2, \ |
|
| 174 |
- const uint8_t *src3, \ |
|
| 175 |
- const uint8_t *src4, \ |
|
| 176 |
- int dst_stride, \ |
|
| 177 |
- int src_stride1, \ |
|
| 178 |
- int src_stride2, \ |
|
| 179 |
- int src_stride3, \ |
|
| 180 |
- int src_stride4, \ |
|
| 181 |
- int h) \ |
|
| 182 |
-{ \
|
|
| 183 |
- OPNAME ## _pixels8_l4_8(dst, src1, src2, src3, src4, dst_stride, \ |
|
| 184 |
- src_stride1, src_stride2, src_stride3, \ |
|
| 185 |
- src_stride4, h); \ |
|
| 186 |
- OPNAME ## _pixels8_l4_8(dst + 8, \ |
|
| 187 |
- src1 + 8, src2 + 8, \ |
|
| 188 |
- src3 + 8, src4 + 8, \ |
|
| 189 |
- dst_stride, src_stride1, src_stride2, \ |
|
| 190 |
- src_stride3, src_stride4, h); \ |
|
| 191 |
-} \ |
|
| 192 |
- \ |
|
| 193 |
-static inline void OPNAME ## _no_rnd_pixels16_l4_8(uint8_t *dst, \ |
|
| 194 |
- const uint8_t *src1, \ |
|
| 195 |
- const uint8_t *src2, \ |
|
| 196 |
- const uint8_t *src3, \ |
|
| 197 |
- const uint8_t *src4, \ |
|
| 198 |
- int dst_stride, \ |
|
| 199 |
- int src_stride1, \ |
|
| 200 |
- int src_stride2, \ |
|
| 201 |
- int src_stride3, \ |
|
| 202 |
- int src_stride4, \ |
|
| 203 |
- int h) \ |
|
| 204 |
-{ \
|
|
| 205 |
- OPNAME ## _no_rnd_pixels8_l4_8(dst, src1, src2, src3, src4, \ |
|
| 206 |
- dst_stride, src_stride1, \ |
|
| 207 |
- src_stride2, src_stride3, \ |
|
| 208 |
- src_stride4, h); \ |
|
| 209 |
- OPNAME ## _no_rnd_pixels8_l4_8(dst + 8, \ |
|
| 210 |
- src1 + 8, src2 + 8, \ |
|
| 211 |
- src3 + 8, src4 + 8, \ |
|
| 212 |
- dst_stride, src_stride1, \ |
|
| 213 |
- src_stride2, src_stride3, \ |
|
| 214 |
- src_stride4, h); \ |
|
| 215 |
-} \ |
|
| 216 |
- |
|
| 217 |
-#define op_avg(a, b) a = rnd_avg32(a, b) |
|
| 218 |
-#define op_put(a, b) a = b |
|
| 219 |
-#define put_no_rnd_pixels8_8_c put_pixels8_8_c |
|
| 220 |
-PIXOP2(avg, op_avg) |
|
| 221 |
-PIXOP2(put, op_put) |
|
| 222 |
-#undef op_avg |
|
| 223 |
-#undef op_put |
| ... | ... |
@@ -36,6 +36,7 @@ |
| 36 | 36 |
#include "mpeg4video_parser.h" |
| 37 | 37 |
#include "mpegvideo.h" |
| 38 | 38 |
#include "msmpeg4.h" |
| 39 |
+#include "qpeldsp.h" |
|
| 39 | 40 |
#include "thread.h" |
| 40 | 41 |
|
| 41 | 42 |
av_cold int ff_h263_decode_init(AVCodecContext *avctx) |
| ... | ... |
@@ -116,6 +117,7 @@ av_cold int ff_h263_decode_init(AVCodecContext *avctx) |
| 116 | 116 |
return ret; |
| 117 | 117 |
|
| 118 | 118 |
ff_h263dsp_init(&s->h263dsp); |
| 119 |
+ ff_qpeldsp_init(&s->qdsp); |
|
| 119 | 120 |
ff_h263_decode_init_vlc(); |
| 120 | 121 |
|
| 121 | 122 |
return 0; |
| ... | ... |
@@ -461,9 +463,9 @@ int ff_h263_decode_frame(AVCodecContext *avctx, void *data, int *got_frame, |
| 461 | 461 |
avctx->has_b_frames = !s->low_delay; |
| 462 | 462 |
|
| 463 | 463 |
#define SET_QPEL_FUNC(postfix1, postfix2) \ |
| 464 |
- s->dsp.put_ ## postfix1 = ff_put_ ## postfix2; \ |
|
| 465 |
- s->dsp.put_no_rnd_ ## postfix1 = ff_put_no_rnd_ ## postfix2; \ |
|
| 466 |
- s->dsp.avg_ ## postfix1 = ff_avg_ ## postfix2; |
|
| 464 |
+ s->qdsp.put_ ## postfix1 = ff_put_ ## postfix2; \ |
|
| 465 |
+ s->qdsp.put_no_rnd_ ## postfix1 = ff_put_no_rnd_ ## postfix2; \ |
|
| 466 |
+ s->qdsp.avg_ ## postfix1 = ff_avg_ ## postfix2; |
|
| 467 | 467 |
|
| 468 | 468 |
if (s->workaround_bugs & FF_BUG_STD_QPEL) {
|
| 469 | 469 |
SET_QPEL_FUNC(qpel_pixels_tab[0][5], qpel16_mc11_old_c) |
| ... | ... |
@@ -527,11 +529,11 @@ int ff_h263_decode_frame(AVCodecContext *avctx, void *data, int *got_frame, |
| 527 | 527 |
} |
| 528 | 528 |
|
| 529 | 529 |
if ((!s->no_rounding) || s->pict_type == AV_PICTURE_TYPE_B) {
|
| 530 |
- s->me.qpel_put = s->dsp.put_qpel_pixels_tab; |
|
| 531 |
- s->me.qpel_avg = s->dsp.avg_qpel_pixels_tab; |
|
| 530 |
+ s->me.qpel_put = s->qdsp.put_qpel_pixels_tab; |
|
| 531 |
+ s->me.qpel_avg = s->qdsp.avg_qpel_pixels_tab; |
|
| 532 | 532 |
} else {
|
| 533 |
- s->me.qpel_put = s->dsp.put_no_rnd_qpel_pixels_tab; |
|
| 534 |
- s->me.qpel_avg = s->dsp.avg_qpel_pixels_tab; |
|
| 533 |
+ s->me.qpel_put = s->qdsp.put_no_rnd_qpel_pixels_tab; |
|
| 534 |
+ s->me.qpel_avg = s->qdsp.avg_qpel_pixels_tab; |
|
| 535 | 535 |
} |
| 536 | 536 |
|
| 537 | 537 |
if ((ret = ff_MPV_frame_start(s, avctx)) < 0) |
| ... | ... |
@@ -329,9 +329,11 @@ int ff_init_me(MpegEncContext *s){
|
| 329 | 329 |
/*FIXME s->no_rounding b_type*/ |
| 330 | 330 |
if(s->flags&CODEC_FLAG_QPEL){
|
| 331 | 331 |
c->sub_motion_search= qpel_motion_search; |
| 332 |
- c->qpel_avg= s->dsp.avg_qpel_pixels_tab; |
|
| 333 |
- if(s->no_rounding) c->qpel_put= s->dsp.put_no_rnd_qpel_pixels_tab; |
|
| 334 |
- else c->qpel_put= s->dsp.put_qpel_pixels_tab; |
|
| 332 |
+ c->qpel_avg = s->qdsp.avg_qpel_pixels_tab; |
|
| 333 |
+ if (s->no_rounding) |
|
| 334 |
+ c->qpel_put = s->qdsp.put_no_rnd_qpel_pixels_tab; |
|
| 335 |
+ else |
|
| 336 |
+ c->qpel_put = s->qdsp.put_qpel_pixels_tab; |
|
| 335 | 337 |
}else{
|
| 336 | 338 |
if(c->avctx->me_sub_cmp&FF_CMP_CHROMA) |
| 337 | 339 |
c->sub_motion_search= hpel_motion_search; |
| ... | ... |
@@ -622,9 +624,9 @@ static inline int h263_mv4_search(MpegEncContext *s, int mx, int my, int shift) |
| 622 | 622 |
dxy = ((my4 & 3) << 2) | (mx4 & 3); |
| 623 | 623 |
|
| 624 | 624 |
if(s->no_rounding) |
| 625 |
- s->dsp.put_no_rnd_qpel_pixels_tab[1][dxy](dest_y , ref , stride); |
|
| 625 |
+ s->qdsp.put_no_rnd_qpel_pixels_tab[1][dxy](dest_y, ref, stride); |
|
| 626 | 626 |
else |
| 627 |
- s->dsp.put_qpel_pixels_tab [1][dxy](dest_y , ref , stride); |
|
| 627 |
+ s->qdsp.put_qpel_pixels_tab[1][dxy](dest_y, ref, stride); |
|
| 628 | 628 |
}else{
|
| 629 | 629 |
uint8_t *ref= c->ref[block][0] + (mx4>>1) + (my4>>1)*stride; |
| 630 | 630 |
dxy = ((my4 & 1) << 1) | (mx4 & 1); |
| ... | ... |
@@ -1208,14 +1210,14 @@ static inline int check_bidir_mv(MpegEncContext * s, |
| 1208 | 1208 |
src_y = motion_fy >> 2; |
| 1209 | 1209 |
|
| 1210 | 1210 |
ptr = ref_data[0] + (src_y * stride) + src_x; |
| 1211 |
- s->dsp.put_qpel_pixels_tab[0][dxy](dest_y , ptr , stride); |
|
| 1211 |
+ s->qdsp.put_qpel_pixels_tab[0][dxy](dest_y, ptr, stride); |
|
| 1212 | 1212 |
|
| 1213 | 1213 |
dxy = ((motion_by & 3) << 2) | (motion_bx & 3); |
| 1214 | 1214 |
src_x = motion_bx >> 2; |
| 1215 | 1215 |
src_y = motion_by >> 2; |
| 1216 | 1216 |
|
| 1217 | 1217 |
ptr = ref2_data[0] + (src_y * stride) + src_x; |
| 1218 |
- s->dsp.avg_qpel_pixels_tab[size][dxy](dest_y , ptr , stride); |
|
| 1218 |
+ s->qdsp.avg_qpel_pixels_tab[size][dxy](dest_y, ptr, stride); |
|
| 1219 | 1219 |
}else{
|
| 1220 | 1220 |
dxy = ((motion_fy & 1) << 1) | (motion_fx & 1); |
| 1221 | 1221 |
src_x = motion_fx >> 1; |
| ... | ... |
@@ -38,6 +38,7 @@ |
| 38 | 38 |
#include "ratecontrol.h" |
| 39 | 39 |
#include "parser.h" |
| 40 | 40 |
#include "mpeg12data.h" |
| 41 |
+#include "qpeldsp.h" |
|
| 41 | 42 |
#include "rl.h" |
| 42 | 43 |
#include "thread.h" |
| 43 | 44 |
#include "videodsp.h" |
| ... | ... |
@@ -348,6 +349,7 @@ typedef struct MpegEncContext {
|
| 348 | 348 |
|
| 349 | 349 |
DSPContext dsp; ///< pointers for accelerated dsp functions |
| 350 | 350 |
HpelDSPContext hdsp; |
| 351 |
+ QpelDSPContext qdsp; |
|
| 351 | 352 |
VideoDSPContext vdsp; |
| 352 | 353 |
H263DSPContext h263dsp; |
| 353 | 354 |
int f_code; ///< forward MV resolution |
| ... | ... |
@@ -46,6 +46,7 @@ |
| 46 | 46 |
#include "mpegutils.h" |
| 47 | 47 |
#include "mjpegenc.h" |
| 48 | 48 |
#include "msmpeg4.h" |
| 49 |
+#include "qpeldsp.h" |
|
| 49 | 50 |
#include "faandct.h" |
| 50 | 51 |
#include "thread.h" |
| 51 | 52 |
#include "aandcttab.h" |
| ... | ... |
@@ -687,6 +688,8 @@ av_cold int ff_MPV_encode_init(AVCodecContext *avctx) |
| 687 | 687 |
if (ARCH_X86) |
| 688 | 688 |
ff_MPV_encode_init_x86(s); |
| 689 | 689 |
|
| 690 |
+ ff_qpeldsp_init(&s->qdsp); |
|
| 691 |
+ |
|
| 690 | 692 |
s->avctx->coded_frame = s->current_picture.f; |
| 691 | 693 |
|
| 692 | 694 |
if (s->msmpeg4_version) {
|
| ... | ... |
@@ -1944,10 +1947,10 @@ static av_always_inline void encode_mb_internal(MpegEncContext *s, |
| 1944 | 1944 |
|
| 1945 | 1945 |
if ((!s->no_rounding) || s->pict_type == AV_PICTURE_TYPE_B) {
|
| 1946 | 1946 |
op_pix = s->hdsp.put_pixels_tab; |
| 1947 |
- op_qpix = s->dsp.put_qpel_pixels_tab; |
|
| 1947 |
+ op_qpix = s->qdsp.put_qpel_pixels_tab; |
|
| 1948 | 1948 |
} else {
|
| 1949 | 1949 |
op_pix = s->hdsp.put_no_rnd_pixels_tab; |
| 1950 |
- op_qpix = s->dsp.put_no_rnd_qpel_pixels_tab; |
|
| 1950 |
+ op_qpix = s->qdsp.put_no_rnd_qpel_pixels_tab; |
|
| 1951 | 1951 |
} |
| 1952 | 1952 |
|
| 1953 | 1953 |
if (s->mv_dir & MV_DIR_FORWARD) {
|
| ... | ... |
@@ -1955,7 +1958,7 @@ static av_always_inline void encode_mb_internal(MpegEncContext *s, |
| 1955 | 1955 |
s->last_picture.f->data, |
| 1956 | 1956 |
op_pix, op_qpix); |
| 1957 | 1957 |
op_pix = s->hdsp.avg_pixels_tab; |
| 1958 |
- op_qpix = s->dsp.avg_qpel_pixels_tab; |
|
| 1958 |
+ op_qpix = s->qdsp.avg_qpel_pixels_tab; |
|
| 1959 | 1959 |
} |
| 1960 | 1960 |
if (s->mv_dir & MV_DIR_BACKWARD) {
|
| 1961 | 1961 |
ff_MPV_motion(s, dest_y, dest_cb, dest_cr, 1, |
| ... | ... |
@@ -27,6 +27,7 @@ |
| 27 | 27 |
#include "error_resilience.h" |
| 28 | 28 |
#include "internal.h" |
| 29 | 29 |
#include "msmpeg4data.h" |
| 30 |
+#include "qpeldsp.h" |
|
| 30 | 31 |
#include "vc1.h" |
| 31 | 32 |
#include "mss12.h" |
| 32 | 33 |
#include "mss2dsp.h" |
| ... | ... |
@@ -37,6 +38,7 @@ typedef struct MSS2Context {
|
| 37 | 37 |
AVFrame *last_pic; |
| 38 | 38 |
MSS12Context c; |
| 39 | 39 |
MSS2DSPContext dsp; |
| 40 |
+ QpelDSPContext qdsp; |
|
| 40 | 41 |
SliceContext sc[2]; |
| 41 | 42 |
} MSS2Context; |
| 42 | 43 |
|
| ... | ... |
@@ -787,8 +789,8 @@ static av_cold int wmv9_init(AVCodecContext *avctx) |
| 787 | 787 |
return ret; |
| 788 | 788 |
|
| 789 | 789 |
/* error concealment */ |
| 790 |
- v->s.me.qpel_put = v->s.dsp.put_qpel_pixels_tab; |
|
| 791 |
- v->s.me.qpel_avg = v->s.dsp.avg_qpel_pixels_tab; |
|
| 790 |
+ v->s.me.qpel_put = v->s.qdsp.put_qpel_pixels_tab; |
|
| 791 |
+ v->s.me.qpel_avg = v->s.qdsp.avg_qpel_pixels_tab; |
|
| 792 | 792 |
|
| 793 | 793 |
return 0; |
| 794 | 794 |
} |
| ... | ... |
@@ -827,6 +829,7 @@ static av_cold int mss2_decode_init(AVCodecContext *avctx) |
| 827 | 827 |
return ret; |
| 828 | 828 |
} |
| 829 | 829 |
ff_mss2dsp_init(&ctx->dsp); |
| 830 |
+ ff_qpeldsp_init(&ctx->qdsp); |
|
| 830 | 831 |
|
| 831 | 832 |
avctx->pix_fmt = c->free_colours == 127 ? AV_PIX_FMT_RGB555 |
| 832 | 833 |
: AV_PIX_FMT_RGB24; |
| 833 | 834 |
new file mode 100644 |
| ... | ... |
@@ -0,0 +1,219 @@ |
| 0 |
+/* |
|
| 1 |
+ * quarterpel DSP function templates |
|
| 2 |
+ * |
|
| 3 |
+ * This file is part of Libav. |
|
| 4 |
+ * |
|
| 5 |
+ * Libav is free software; you can redistribute it and/or |
|
| 6 |
+ * modify it under the terms of the GNU Lesser General Public |
|
| 7 |
+ * License as published by the Free Software Foundation; either |
|
| 8 |
+ * version 2.1 of the License, or (at your option) any later version. |
|
| 9 |
+ * |
|
| 10 |
+ * Libav is distributed in the hope that it will be useful, |
|
| 11 |
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
| 12 |
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
| 13 |
+ * Lesser General Public License for more details. |
|
| 14 |
+ * |
|
| 15 |
+ * You should have received a copy of the GNU Lesser General Public |
|
| 16 |
+ * License along with Libav; if not, write to the Free Software |
|
| 17 |
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
| 18 |
+ */ |
|
| 19 |
+ |
|
| 20 |
+/** |
|
| 21 |
+ * @file |
|
| 22 |
+ * quarterpel DSP function templates |
|
| 23 |
+ */ |
|
| 24 |
+ |
|
| 25 |
+#define PIXOP2(OPNAME, OP) \ |
|
| 26 |
+static inline void OPNAME ## _no_rnd_pixels8_l2_8(uint8_t *dst, \ |
|
| 27 |
+ const uint8_t *src1, \ |
|
| 28 |
+ const uint8_t *src2, \ |
|
| 29 |
+ int dst_stride, \ |
|
| 30 |
+ int src_stride1, \ |
|
| 31 |
+ int src_stride2, \ |
|
| 32 |
+ int h) \ |
|
| 33 |
+{ \
|
|
| 34 |
+ int i; \ |
|
| 35 |
+ \ |
|
| 36 |
+ for (i = 0; i < h; i++) { \
|
|
| 37 |
+ uint32_t a, b; \ |
|
| 38 |
+ a = AV_RN32(&src1[i * src_stride1]); \ |
|
| 39 |
+ b = AV_RN32(&src2[i * src_stride2]); \ |
|
| 40 |
+ OP(*((uint32_t *) &dst[i * dst_stride]), \ |
|
| 41 |
+ no_rnd_avg32(a, b)); \ |
|
| 42 |
+ a = AV_RN32(&src1[i * src_stride1 + 4]); \ |
|
| 43 |
+ b = AV_RN32(&src2[i * src_stride2 + 4]); \ |
|
| 44 |
+ OP(*((uint32_t *) &dst[i * dst_stride + 4]), \ |
|
| 45 |
+ no_rnd_avg32(a, b)); \ |
|
| 46 |
+ } \ |
|
| 47 |
+} \ |
|
| 48 |
+ \ |
|
| 49 |
+static inline void OPNAME ## _no_rnd_pixels16_l2_8(uint8_t *dst, \ |
|
| 50 |
+ const uint8_t *src1, \ |
|
| 51 |
+ const uint8_t *src2, \ |
|
| 52 |
+ int dst_stride, \ |
|
| 53 |
+ int src_stride1, \ |
|
| 54 |
+ int src_stride2, \ |
|
| 55 |
+ int h) \ |
|
| 56 |
+{ \
|
|
| 57 |
+ OPNAME ## _no_rnd_pixels8_l2_8(dst, src1, src2, dst_stride, \ |
|
| 58 |
+ src_stride1, src_stride2, h); \ |
|
| 59 |
+ OPNAME ## _no_rnd_pixels8_l2_8(dst + 8, \ |
|
| 60 |
+ src1 + 8, \ |
|
| 61 |
+ src2 + 8, \ |
|
| 62 |
+ dst_stride, src_stride1, \ |
|
| 63 |
+ src_stride2, h); \ |
|
| 64 |
+} \ |
|
| 65 |
+ \ |
|
| 66 |
+static inline void OPNAME ## _pixels8_l4_8(uint8_t *dst, \ |
|
| 67 |
+ const uint8_t *src1, \ |
|
| 68 |
+ const uint8_t *src2, \ |
|
| 69 |
+ const uint8_t *src3, \ |
|
| 70 |
+ const uint8_t *src4, \ |
|
| 71 |
+ int dst_stride, \ |
|
| 72 |
+ int src_stride1, \ |
|
| 73 |
+ int src_stride2, \ |
|
| 74 |
+ int src_stride3, \ |
|
| 75 |
+ int src_stride4, \ |
|
| 76 |
+ int h) \ |
|
| 77 |
+{ \
|
|
| 78 |
+ /* FIXME HIGH BIT DEPTH */ \ |
|
| 79 |
+ int i; \ |
|
| 80 |
+ \ |
|
| 81 |
+ for (i = 0; i < h; i++) { \
|
|
| 82 |
+ uint32_t a, b, c, d, l0, l1, h0, h1; \ |
|
| 83 |
+ a = AV_RN32(&src1[i * src_stride1]); \ |
|
| 84 |
+ b = AV_RN32(&src2[i * src_stride2]); \ |
|
| 85 |
+ c = AV_RN32(&src3[i * src_stride3]); \ |
|
| 86 |
+ d = AV_RN32(&src4[i * src_stride4]); \ |
|
| 87 |
+ l0 = (a & 0x03030303UL) + \ |
|
| 88 |
+ (b & 0x03030303UL) + \ |
|
| 89 |
+ 0x02020202UL; \ |
|
| 90 |
+ h0 = ((a & 0xFCFCFCFCUL) >> 2) + \ |
|
| 91 |
+ ((b & 0xFCFCFCFCUL) >> 2); \ |
|
| 92 |
+ l1 = (c & 0x03030303UL) + \ |
|
| 93 |
+ (d & 0x03030303UL); \ |
|
| 94 |
+ h1 = ((c & 0xFCFCFCFCUL) >> 2) + \ |
|
| 95 |
+ ((d & 0xFCFCFCFCUL) >> 2); \ |
|
| 96 |
+ OP(*((uint32_t *) &dst[i * dst_stride]), \ |
|
| 97 |
+ h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL)); \ |
|
| 98 |
+ a = AV_RN32(&src1[i * src_stride1 + 4]); \ |
|
| 99 |
+ b = AV_RN32(&src2[i * src_stride2 + 4]); \ |
|
| 100 |
+ c = AV_RN32(&src3[i * src_stride3 + 4]); \ |
|
| 101 |
+ d = AV_RN32(&src4[i * src_stride4 + 4]); \ |
|
| 102 |
+ l0 = (a & 0x03030303UL) + \ |
|
| 103 |
+ (b & 0x03030303UL) + \ |
|
| 104 |
+ 0x02020202UL; \ |
|
| 105 |
+ h0 = ((a & 0xFCFCFCFCUL) >> 2) + \ |
|
| 106 |
+ ((b & 0xFCFCFCFCUL) >> 2); \ |
|
| 107 |
+ l1 = (c & 0x03030303UL) + \ |
|
| 108 |
+ (d & 0x03030303UL); \ |
|
| 109 |
+ h1 = ((c & 0xFCFCFCFCUL) >> 2) + \ |
|
| 110 |
+ ((d & 0xFCFCFCFCUL) >> 2); \ |
|
| 111 |
+ OP(*((uint32_t *) &dst[i * dst_stride + 4]), \ |
|
| 112 |
+ h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL)); \ |
|
| 113 |
+ } \ |
|
| 114 |
+} \ |
|
| 115 |
+ \ |
|
| 116 |
+static inline void OPNAME ## _no_rnd_pixels8_l4_8(uint8_t *dst, \ |
|
| 117 |
+ const uint8_t *src1, \ |
|
| 118 |
+ const uint8_t *src2, \ |
|
| 119 |
+ const uint8_t *src3, \ |
|
| 120 |
+ const uint8_t *src4, \ |
|
| 121 |
+ int dst_stride, \ |
|
| 122 |
+ int src_stride1, \ |
|
| 123 |
+ int src_stride2, \ |
|
| 124 |
+ int src_stride3, \ |
|
| 125 |
+ int src_stride4, \ |
|
| 126 |
+ int h) \ |
|
| 127 |
+{ \
|
|
| 128 |
+ /* FIXME HIGH BIT DEPTH */ \ |
|
| 129 |
+ int i; \ |
|
| 130 |
+ \ |
|
| 131 |
+ for (i = 0; i < h; i++) { \
|
|
| 132 |
+ uint32_t a, b, c, d, l0, l1, h0, h1; \ |
|
| 133 |
+ a = AV_RN32(&src1[i * src_stride1]); \ |
|
| 134 |
+ b = AV_RN32(&src2[i * src_stride2]); \ |
|
| 135 |
+ c = AV_RN32(&src3[i * src_stride3]); \ |
|
| 136 |
+ d = AV_RN32(&src4[i * src_stride4]); \ |
|
| 137 |
+ l0 = (a & 0x03030303UL) + \ |
|
| 138 |
+ (b & 0x03030303UL) + \ |
|
| 139 |
+ 0x01010101UL; \ |
|
| 140 |
+ h0 = ((a & 0xFCFCFCFCUL) >> 2) + \ |
|
| 141 |
+ ((b & 0xFCFCFCFCUL) >> 2); \ |
|
| 142 |
+ l1 = (c & 0x03030303UL) + \ |
|
| 143 |
+ (d & 0x03030303UL); \ |
|
| 144 |
+ h1 = ((c & 0xFCFCFCFCUL) >> 2) + \ |
|
| 145 |
+ ((d & 0xFCFCFCFCUL) >> 2); \ |
|
| 146 |
+ OP(*((uint32_t *) &dst[i * dst_stride]), \ |
|
| 147 |
+ h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL)); \ |
|
| 148 |
+ a = AV_RN32(&src1[i * src_stride1 + 4]); \ |
|
| 149 |
+ b = AV_RN32(&src2[i * src_stride2 + 4]); \ |
|
| 150 |
+ c = AV_RN32(&src3[i * src_stride3 + 4]); \ |
|
| 151 |
+ d = AV_RN32(&src4[i * src_stride4 + 4]); \ |
|
| 152 |
+ l0 = (a & 0x03030303UL) + \ |
|
| 153 |
+ (b & 0x03030303UL) + \ |
|
| 154 |
+ 0x01010101UL; \ |
|
| 155 |
+ h0 = ((a & 0xFCFCFCFCUL) >> 2) + \ |
|
| 156 |
+ ((b & 0xFCFCFCFCUL) >> 2); \ |
|
| 157 |
+ l1 = (c & 0x03030303UL) + \ |
|
| 158 |
+ (d & 0x03030303UL); \ |
|
| 159 |
+ h1 = ((c & 0xFCFCFCFCUL) >> 2) + \ |
|
| 160 |
+ ((d & 0xFCFCFCFCUL) >> 2); \ |
|
| 161 |
+ OP(*((uint32_t *) &dst[i * dst_stride + 4]), \ |
|
| 162 |
+ h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL)); \ |
|
| 163 |
+ } \ |
|
| 164 |
+} \ |
|
| 165 |
+ \ |
|
| 166 |
+static inline void OPNAME ## _pixels16_l4_8(uint8_t *dst, \ |
|
| 167 |
+ const uint8_t *src1, \ |
|
| 168 |
+ const uint8_t *src2, \ |
|
| 169 |
+ const uint8_t *src3, \ |
|
| 170 |
+ const uint8_t *src4, \ |
|
| 171 |
+ int dst_stride, \ |
|
| 172 |
+ int src_stride1, \ |
|
| 173 |
+ int src_stride2, \ |
|
| 174 |
+ int src_stride3, \ |
|
| 175 |
+ int src_stride4, \ |
|
| 176 |
+ int h) \ |
|
| 177 |
+{ \
|
|
| 178 |
+ OPNAME ## _pixels8_l4_8(dst, src1, src2, src3, src4, dst_stride, \ |
|
| 179 |
+ src_stride1, src_stride2, src_stride3, \ |
|
| 180 |
+ src_stride4, h); \ |
|
| 181 |
+ OPNAME ## _pixels8_l4_8(dst + 8, \ |
|
| 182 |
+ src1 + 8, src2 + 8, \ |
|
| 183 |
+ src3 + 8, src4 + 8, \ |
|
| 184 |
+ dst_stride, src_stride1, src_stride2, \ |
|
| 185 |
+ src_stride3, src_stride4, h); \ |
|
| 186 |
+} \ |
|
| 187 |
+ \ |
|
| 188 |
+static inline void OPNAME ## _no_rnd_pixels16_l4_8(uint8_t *dst, \ |
|
| 189 |
+ const uint8_t *src1, \ |
|
| 190 |
+ const uint8_t *src2, \ |
|
| 191 |
+ const uint8_t *src3, \ |
|
| 192 |
+ const uint8_t *src4, \ |
|
| 193 |
+ int dst_stride, \ |
|
| 194 |
+ int src_stride1, \ |
|
| 195 |
+ int src_stride2, \ |
|
| 196 |
+ int src_stride3, \ |
|
| 197 |
+ int src_stride4, \ |
|
| 198 |
+ int h) \ |
|
| 199 |
+{ \
|
|
| 200 |
+ OPNAME ## _no_rnd_pixels8_l4_8(dst, src1, src2, src3, src4, \ |
|
| 201 |
+ dst_stride, src_stride1, \ |
|
| 202 |
+ src_stride2, src_stride3, \ |
|
| 203 |
+ src_stride4, h); \ |
|
| 204 |
+ OPNAME ## _no_rnd_pixels8_l4_8(dst + 8, \ |
|
| 205 |
+ src1 + 8, src2 + 8, \ |
|
| 206 |
+ src3 + 8, src4 + 8, \ |
|
| 207 |
+ dst_stride, src_stride1, \ |
|
| 208 |
+ src_stride2, src_stride3, \ |
|
| 209 |
+ src_stride4, h); \ |
|
| 210 |
+} \ |
|
| 211 |
+ |
|
| 212 |
+#define op_avg(a, b) a = rnd_avg32(a, b) |
|
| 213 |
+#define op_put(a, b) a = b |
|
| 214 |
+#define put_no_rnd_pixels8_8_c put_pixels8_8_c |
|
| 215 |
+PIXOP2(avg, op_avg) |
|
| 216 |
+PIXOP2(put, op_put) |
|
| 217 |
+#undef op_avg |
|
| 218 |
+#undef op_put |
| 0 | 219 |
new file mode 100644 |
| ... | ... |
@@ -0,0 +1,764 @@ |
| 0 |
+/* |
|
| 1 |
+ * quarterpel DSP functions |
|
| 2 |
+ * |
|
| 3 |
+ * This file is part of Libav. |
|
| 4 |
+ * |
|
| 5 |
+ * Libav is free software; you can redistribute it and/or |
|
| 6 |
+ * modify it under the terms of the GNU Lesser General Public |
|
| 7 |
+ * License as published by the Free Software Foundation; either |
|
| 8 |
+ * version 2.1 of the License, or (at your option) any later version. |
|
| 9 |
+ * |
|
| 10 |
+ * Libav is distributed in the hope that it will be useful, |
|
| 11 |
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
| 12 |
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
| 13 |
+ * Lesser General Public License for more details. |
|
| 14 |
+ * |
|
| 15 |
+ * You should have received a copy of the GNU Lesser General Public |
|
| 16 |
+ * License along with Libav; if not, write to the Free Software |
|
| 17 |
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
| 18 |
+ */ |
|
| 19 |
+ |
|
| 20 |
+/** |
|
| 21 |
+ * @file |
|
| 22 |
+ * quarterpel DSP functions |
|
| 23 |
+ */ |
|
| 24 |
+ |
|
| 25 |
+#include <stddef.h> |
|
| 26 |
+#include <stdint.h> |
|
| 27 |
+ |
|
| 28 |
+#include "config.h" |
|
| 29 |
+#include "libavutil/attributes.h" |
|
| 30 |
+#include "copy_block.h" |
|
| 31 |
+#include "qpeldsp.h" |
|
| 32 |
+ |
|
| 33 |
+#define BIT_DEPTH 8 |
|
| 34 |
+#include "hpel_template.c" |
|
| 35 |
+#include "tpel_template.c" |
|
| 36 |
+#include "qpel_template.c" |
|
| 37 |
+ |
|
| 38 |
+#define QPEL_MC(r, OPNAME, RND, OP) \ |
|
| 39 |
+static void OPNAME ## mpeg4_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, \ |
|
| 40 |
+ int dstStride, int srcStride, \ |
|
| 41 |
+ int h) \ |
|
| 42 |
+{ \
|
|
| 43 |
+ const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ |
|
| 44 |
+ int i; \ |
|
| 45 |
+ \ |
|
| 46 |
+ for (i = 0; i < h; i++) { \
|
|
| 47 |
+ OP(dst[0], (src[0] + src[1]) * 20 - (src[0] + src[2]) * 6 + (src[1] + src[3]) * 3 - (src[2] + src[4])); \ |
|
| 48 |
+ OP(dst[1], (src[1] + src[2]) * 20 - (src[0] + src[3]) * 6 + (src[0] + src[4]) * 3 - (src[1] + src[5])); \ |
|
| 49 |
+ OP(dst[2], (src[2] + src[3]) * 20 - (src[1] + src[4]) * 6 + (src[0] + src[5]) * 3 - (src[0] + src[6])); \ |
|
| 50 |
+ OP(dst[3], (src[3] + src[4]) * 20 - (src[2] + src[5]) * 6 + (src[1] + src[6]) * 3 - (src[0] + src[7])); \ |
|
| 51 |
+ OP(dst[4], (src[4] + src[5]) * 20 - (src[3] + src[6]) * 6 + (src[2] + src[7]) * 3 - (src[1] + src[8])); \ |
|
| 52 |
+ OP(dst[5], (src[5] + src[6]) * 20 - (src[4] + src[7]) * 6 + (src[3] + src[8]) * 3 - (src[2] + src[8])); \ |
|
| 53 |
+ OP(dst[6], (src[6] + src[7]) * 20 - (src[5] + src[8]) * 6 + (src[4] + src[8]) * 3 - (src[3] + src[7])); \ |
|
| 54 |
+ OP(dst[7], (src[7] + src[8]) * 20 - (src[6] + src[8]) * 6 + (src[5] + src[7]) * 3 - (src[4] + src[6])); \ |
|
| 55 |
+ dst += dstStride; \ |
|
| 56 |
+ src += srcStride; \ |
|
| 57 |
+ } \ |
|
| 58 |
+} \ |
|
| 59 |
+ \ |
|
| 60 |
+static void OPNAME ## mpeg4_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, \ |
|
| 61 |
+ int dstStride, int srcStride) \ |
|
| 62 |
+{ \
|
|
| 63 |
+ const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ |
|
| 64 |
+ const int w = 8; \ |
|
| 65 |
+ int i; \ |
|
| 66 |
+ \ |
|
| 67 |
+ for (i = 0; i < w; i++) { \
|
|
| 68 |
+ const int src0 = src[0 * srcStride]; \ |
|
| 69 |
+ const int src1 = src[1 * srcStride]; \ |
|
| 70 |
+ const int src2 = src[2 * srcStride]; \ |
|
| 71 |
+ const int src3 = src[3 * srcStride]; \ |
|
| 72 |
+ const int src4 = src[4 * srcStride]; \ |
|
| 73 |
+ const int src5 = src[5 * srcStride]; \ |
|
| 74 |
+ const int src6 = src[6 * srcStride]; \ |
|
| 75 |
+ const int src7 = src[7 * srcStride]; \ |
|
| 76 |
+ const int src8 = src[8 * srcStride]; \ |
|
| 77 |
+ OP(dst[0 * dstStride], (src0 + src1) * 20 - (src0 + src2) * 6 + (src1 + src3) * 3 - (src2 + src4)); \ |
|
| 78 |
+ OP(dst[1 * dstStride], (src1 + src2) * 20 - (src0 + src3) * 6 + (src0 + src4) * 3 - (src1 + src5)); \ |
|
| 79 |
+ OP(dst[2 * dstStride], (src2 + src3) * 20 - (src1 + src4) * 6 + (src0 + src5) * 3 - (src0 + src6)); \ |
|
| 80 |
+ OP(dst[3 * dstStride], (src3 + src4) * 20 - (src2 + src5) * 6 + (src1 + src6) * 3 - (src0 + src7)); \ |
|
| 81 |
+ OP(dst[4 * dstStride], (src4 + src5) * 20 - (src3 + src6) * 6 + (src2 + src7) * 3 - (src1 + src8)); \ |
|
| 82 |
+ OP(dst[5 * dstStride], (src5 + src6) * 20 - (src4 + src7) * 6 + (src3 + src8) * 3 - (src2 + src8)); \ |
|
| 83 |
+ OP(dst[6 * dstStride], (src6 + src7) * 20 - (src5 + src8) * 6 + (src4 + src8) * 3 - (src3 + src7)); \ |
|
| 84 |
+ OP(dst[7 * dstStride], (src7 + src8) * 20 - (src6 + src8) * 6 + (src5 + src7) * 3 - (src4 + src6)); \ |
|
| 85 |
+ dst++; \ |
|
| 86 |
+ src++; \ |
|
| 87 |
+ } \ |
|
| 88 |
+} \ |
|
| 89 |
+ \ |
|
| 90 |
+static void OPNAME ## mpeg4_qpel16_h_lowpass(uint8_t *dst, uint8_t *src, \ |
|
| 91 |
+ int dstStride, int srcStride, \ |
|
| 92 |
+ int h) \ |
|
| 93 |
+{ \
|
|
| 94 |
+ const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ |
|
| 95 |
+ int i; \ |
|
| 96 |
+ \ |
|
| 97 |
+ for (i = 0; i < h; i++) { \
|
|
| 98 |
+ OP(dst[0], (src[0] + src[1]) * 20 - (src[0] + src[2]) * 6 + (src[1] + src[3]) * 3 - (src[2] + src[4])); \ |
|
| 99 |
+ OP(dst[1], (src[1] + src[2]) * 20 - (src[0] + src[3]) * 6 + (src[0] + src[4]) * 3 - (src[1] + src[5])); \ |
|
| 100 |
+ OP(dst[2], (src[2] + src[3]) * 20 - (src[1] + src[4]) * 6 + (src[0] + src[5]) * 3 - (src[0] + src[6])); \ |
|
| 101 |
+ OP(dst[3], (src[3] + src[4]) * 20 - (src[2] + src[5]) * 6 + (src[1] + src[6]) * 3 - (src[0] + src[7])); \ |
|
| 102 |
+ OP(dst[4], (src[4] + src[5]) * 20 - (src[3] + src[6]) * 6 + (src[2] + src[7]) * 3 - (src[1] + src[8])); \ |
|
| 103 |
+ OP(dst[5], (src[5] + src[6]) * 20 - (src[4] + src[7]) * 6 + (src[3] + src[8]) * 3 - (src[2] + src[9])); \ |
|
| 104 |
+ OP(dst[6], (src[6] + src[7]) * 20 - (src[5] + src[8]) * 6 + (src[4] + src[9]) * 3 - (src[3] + src[10])); \ |
|
| 105 |
+ OP(dst[7], (src[7] + src[8]) * 20 - (src[6] + src[9]) * 6 + (src[5] + src[10]) * 3 - (src[4] + src[11])); \ |
|
| 106 |
+ OP(dst[8], (src[8] + src[9]) * 20 - (src[7] + src[10]) * 6 + (src[6] + src[11]) * 3 - (src[5] + src[12])); \ |
|
| 107 |
+ OP(dst[9], (src[9] + src[10]) * 20 - (src[8] + src[11]) * 6 + (src[7] + src[12]) * 3 - (src[6] + src[13])); \ |
|
| 108 |
+ OP(dst[10], (src[10] + src[11]) * 20 - (src[9] + src[12]) * 6 + (src[8] + src[13]) * 3 - (src[7] + src[14])); \ |
|
| 109 |
+ OP(dst[11], (src[11] + src[12]) * 20 - (src[10] + src[13]) * 6 + (src[9] + src[14]) * 3 - (src[8] + src[15])); \ |
|
| 110 |
+ OP(dst[12], (src[12] + src[13]) * 20 - (src[11] + src[14]) * 6 + (src[10] + src[15]) * 3 - (src[9] + src[16])); \ |
|
| 111 |
+ OP(dst[13], (src[13] + src[14]) * 20 - (src[12] + src[15]) * 6 + (src[11] + src[16]) * 3 - (src[10] + src[16])); \ |
|
| 112 |
+ OP(dst[14], (src[14] + src[15]) * 20 - (src[13] + src[16]) * 6 + (src[12] + src[16]) * 3 - (src[11] + src[15])); \ |
|
| 113 |
+ OP(dst[15], (src[15] + src[16]) * 20 - (src[14] + src[16]) * 6 + (src[13] + src[15]) * 3 - (src[12] + src[14])); \ |
|
| 114 |
+ dst += dstStride; \ |
|
| 115 |
+ src += srcStride; \ |
|
| 116 |
+ } \ |
|
| 117 |
+} \ |
|
| 118 |
+ \ |
|
| 119 |
+static void OPNAME ## mpeg4_qpel16_v_lowpass(uint8_t *dst, uint8_t *src, \ |
|
| 120 |
+ int dstStride, int srcStride) \ |
|
| 121 |
+{ \
|
|
| 122 |
+ const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ |
|
| 123 |
+ const int w = 16; \ |
|
| 124 |
+ int i; \ |
|
| 125 |
+ \ |
|
| 126 |
+ for (i = 0; i < w; i++) { \
|
|
| 127 |
+ const int src0 = src[0 * srcStride]; \ |
|
| 128 |
+ const int src1 = src[1 * srcStride]; \ |
|
| 129 |
+ const int src2 = src[2 * srcStride]; \ |
|
| 130 |
+ const int src3 = src[3 * srcStride]; \ |
|
| 131 |
+ const int src4 = src[4 * srcStride]; \ |
|
| 132 |
+ const int src5 = src[5 * srcStride]; \ |
|
| 133 |
+ const int src6 = src[6 * srcStride]; \ |
|
| 134 |
+ const int src7 = src[7 * srcStride]; \ |
|
| 135 |
+ const int src8 = src[8 * srcStride]; \ |
|
| 136 |
+ const int src9 = src[9 * srcStride]; \ |
|
| 137 |
+ const int src10 = src[10 * srcStride]; \ |
|
| 138 |
+ const int src11 = src[11 * srcStride]; \ |
|
| 139 |
+ const int src12 = src[12 * srcStride]; \ |
|
| 140 |
+ const int src13 = src[13 * srcStride]; \ |
|
| 141 |
+ const int src14 = src[14 * srcStride]; \ |
|
| 142 |
+ const int src15 = src[15 * srcStride]; \ |
|
| 143 |
+ const int src16 = src[16 * srcStride]; \ |
|
| 144 |
+ OP(dst[0 * dstStride], (src0 + src1) * 20 - (src0 + src2) * 6 + (src1 + src3) * 3 - (src2 + src4)); \ |
|
| 145 |
+ OP(dst[1 * dstStride], (src1 + src2) * 20 - (src0 + src3) * 6 + (src0 + src4) * 3 - (src1 + src5)); \ |
|
| 146 |
+ OP(dst[2 * dstStride], (src2 + src3) * 20 - (src1 + src4) * 6 + (src0 + src5) * 3 - (src0 + src6)); \ |
|
| 147 |
+ OP(dst[3 * dstStride], (src3 + src4) * 20 - (src2 + src5) * 6 + (src1 + src6) * 3 - (src0 + src7)); \ |
|
| 148 |
+ OP(dst[4 * dstStride], (src4 + src5) * 20 - (src3 + src6) * 6 + (src2 + src7) * 3 - (src1 + src8)); \ |
|
| 149 |
+ OP(dst[5 * dstStride], (src5 + src6) * 20 - (src4 + src7) * 6 + (src3 + src8) * 3 - (src2 + src9)); \ |
|
| 150 |
+ OP(dst[6 * dstStride], (src6 + src7) * 20 - (src5 + src8) * 6 + (src4 + src9) * 3 - (src3 + src10)); \ |
|
| 151 |
+ OP(dst[7 * dstStride], (src7 + src8) * 20 - (src6 + src9) * 6 + (src5 + src10) * 3 - (src4 + src11)); \ |
|
| 152 |
+ OP(dst[8 * dstStride], (src8 + src9) * 20 - (src7 + src10) * 6 + (src6 + src11) * 3 - (src5 + src12)); \ |
|
| 153 |
+ OP(dst[9 * dstStride], (src9 + src10) * 20 - (src8 + src11) * 6 + (src7 + src12) * 3 - (src6 + src13)); \ |
|
| 154 |
+ OP(dst[10 * dstStride], (src10 + src11) * 20 - (src9 + src12) * 6 + (src8 + src13) * 3 - (src7 + src14)); \ |
|
| 155 |
+ OP(dst[11 * dstStride], (src11 + src12) * 20 - (src10 + src13) * 6 + (src9 + src14) * 3 - (src8 + src15)); \ |
|
| 156 |
+ OP(dst[12 * dstStride], (src12 + src13) * 20 - (src11 + src14) * 6 + (src10 + src15) * 3 - (src9 + src16)); \ |
|
| 157 |
+ OP(dst[13 * dstStride], (src13 + src14) * 20 - (src12 + src15) * 6 + (src11 + src16) * 3 - (src10 + src16)); \ |
|
| 158 |
+ OP(dst[14 * dstStride], (src14 + src15) * 20 - (src13 + src16) * 6 + (src12 + src16) * 3 - (src11 + src15)); \ |
|
| 159 |
+ OP(dst[15 * dstStride], (src15 + src16) * 20 - (src14 + src16) * 6 + (src13 + src15) * 3 - (src12 + src14)); \ |
|
| 160 |
+ dst++; \ |
|
| 161 |
+ src++; \ |
|
| 162 |
+ } \ |
|
| 163 |
+} \ |
|
| 164 |
+ \ |
|
| 165 |
+static void OPNAME ## qpel8_mc10_c(uint8_t *dst, uint8_t *src, \ |
|
| 166 |
+ ptrdiff_t stride) \ |
|
| 167 |
+{ \
|
|
| 168 |
+ uint8_t half[64]; \ |
|
| 169 |
+ \ |
|
| 170 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(half, src, 8, stride, 8); \ |
|
| 171 |
+ OPNAME ## pixels8_l2_8(dst, src, half, stride, stride, 8, 8); \ |
|
| 172 |
+} \ |
|
| 173 |
+ \ |
|
| 174 |
+static void OPNAME ## qpel8_mc20_c(uint8_t *dst, uint8_t *src, \ |
|
| 175 |
+ ptrdiff_t stride) \ |
|
| 176 |
+{ \
|
|
| 177 |
+ OPNAME ## mpeg4_qpel8_h_lowpass(dst, src, stride, stride, 8); \ |
|
| 178 |
+} \ |
|
| 179 |
+ \ |
|
| 180 |
+static void OPNAME ## qpel8_mc30_c(uint8_t *dst, uint8_t *src, \ |
|
| 181 |
+ ptrdiff_t stride) \ |
|
| 182 |
+{ \
|
|
| 183 |
+ uint8_t half[64]; \ |
|
| 184 |
+ \ |
|
| 185 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(half, src, 8, stride, 8); \ |
|
| 186 |
+ OPNAME ## pixels8_l2_8(dst, src + 1, half, stride, stride, 8, 8); \ |
|
| 187 |
+} \ |
|
| 188 |
+ \ |
|
| 189 |
+static void OPNAME ## qpel8_mc01_c(uint8_t *dst, uint8_t *src, \ |
|
| 190 |
+ ptrdiff_t stride) \ |
|
| 191 |
+{ \
|
|
| 192 |
+ uint8_t full[16 * 9]; \ |
|
| 193 |
+ uint8_t half[64]; \ |
|
| 194 |
+ \ |
|
| 195 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 196 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(half, full, 8, 16); \ |
|
| 197 |
+ OPNAME ## pixels8_l2_8(dst, full, half, stride, 16, 8, 8); \ |
|
| 198 |
+} \ |
|
| 199 |
+ \ |
|
| 200 |
+static void OPNAME ## qpel8_mc02_c(uint8_t *dst, uint8_t *src, \ |
|
| 201 |
+ ptrdiff_t stride) \ |
|
| 202 |
+{ \
|
|
| 203 |
+ uint8_t full[16 * 9]; \ |
|
| 204 |
+ \ |
|
| 205 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 206 |
+ OPNAME ## mpeg4_qpel8_v_lowpass(dst, full, stride, 16); \ |
|
| 207 |
+} \ |
|
| 208 |
+ \ |
|
| 209 |
+static void OPNAME ## qpel8_mc03_c(uint8_t *dst, uint8_t *src, \ |
|
| 210 |
+ ptrdiff_t stride) \ |
|
| 211 |
+{ \
|
|
| 212 |
+ uint8_t full[16 * 9]; \ |
|
| 213 |
+ uint8_t half[64]; \ |
|
| 214 |
+ \ |
|
| 215 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 216 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(half, full, 8, 16); \ |
|
| 217 |
+ OPNAME ## pixels8_l2_8(dst, full + 16, half, stride, 16, 8, 8); \ |
|
| 218 |
+} \ |
|
| 219 |
+ \ |
|
| 220 |
+void ff_ ## OPNAME ## qpel8_mc11_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 221 |
+ ptrdiff_t stride) \ |
|
| 222 |
+{ \
|
|
| 223 |
+ uint8_t full[16 * 9]; \ |
|
| 224 |
+ uint8_t halfH[72]; \ |
|
| 225 |
+ uint8_t halfV[64]; \ |
|
| 226 |
+ uint8_t halfHV[64]; \ |
|
| 227 |
+ \ |
|
| 228 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 229 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 230 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full, 8, 16); \ |
|
| 231 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 232 |
+ OPNAME ## pixels8_l4_8(dst, full, halfH, halfV, halfHV, \ |
|
| 233 |
+ stride, 16, 8, 8, 8, 8); \ |
|
| 234 |
+} \ |
|
| 235 |
+ \ |
|
| 236 |
+static void OPNAME ## qpel8_mc11_c(uint8_t *dst, uint8_t *src, \ |
|
| 237 |
+ ptrdiff_t stride) \ |
|
| 238 |
+{ \
|
|
| 239 |
+ uint8_t full[16 * 9]; \ |
|
| 240 |
+ uint8_t halfH[72]; \ |
|
| 241 |
+ uint8_t halfHV[64]; \ |
|
| 242 |
+ \ |
|
| 243 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 244 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 245 |
+ put ## RND ## pixels8_l2_8(halfH, halfH, full, 8, 8, 16, 9); \ |
|
| 246 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 247 |
+ OPNAME ## pixels8_l2_8(dst, halfH, halfHV, stride, 8, 8, 8); \ |
|
| 248 |
+} \ |
|
| 249 |
+ \ |
|
| 250 |
+void ff_ ## OPNAME ## qpel8_mc31_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 251 |
+ ptrdiff_t stride) \ |
|
| 252 |
+{ \
|
|
| 253 |
+ uint8_t full[16 * 9]; \ |
|
| 254 |
+ uint8_t halfH[72]; \ |
|
| 255 |
+ uint8_t halfV[64]; \ |
|
| 256 |
+ uint8_t halfHV[64]; \ |
|
| 257 |
+ \ |
|
| 258 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 259 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 260 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full + 1, 8, 16); \ |
|
| 261 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 262 |
+ OPNAME ## pixels8_l4_8(dst, full + 1, halfH, halfV, halfHV, \ |
|
| 263 |
+ stride, 16, 8, 8, 8, 8); \ |
|
| 264 |
+} \ |
|
| 265 |
+ \ |
|
| 266 |
+static void OPNAME ## qpel8_mc31_c(uint8_t *dst, uint8_t *src, \ |
|
| 267 |
+ ptrdiff_t stride) \ |
|
| 268 |
+{ \
|
|
| 269 |
+ uint8_t full[16 * 9]; \ |
|
| 270 |
+ uint8_t halfH[72]; \ |
|
| 271 |
+ uint8_t halfHV[64]; \ |
|
| 272 |
+ \ |
|
| 273 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 274 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 275 |
+ put ## RND ## pixels8_l2_8(halfH, halfH, full + 1, 8, 8, 16, 9); \ |
|
| 276 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 277 |
+ OPNAME ## pixels8_l2_8(dst, halfH, halfHV, stride, 8, 8, 8); \ |
|
| 278 |
+} \ |
|
| 279 |
+ \ |
|
| 280 |
+void ff_ ## OPNAME ## qpel8_mc13_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 281 |
+ ptrdiff_t stride) \ |
|
| 282 |
+{ \
|
|
| 283 |
+ uint8_t full[16 * 9]; \ |
|
| 284 |
+ uint8_t halfH[72]; \ |
|
| 285 |
+ uint8_t halfV[64]; \ |
|
| 286 |
+ uint8_t halfHV[64]; \ |
|
| 287 |
+ \ |
|
| 288 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 289 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 290 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full, 8, 16); \ |
|
| 291 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 292 |
+ OPNAME ## pixels8_l4_8(dst, full + 16, halfH + 8, halfV, halfHV, \ |
|
| 293 |
+ stride, 16, 8, 8, 8, 8); \ |
|
| 294 |
+} \ |
|
| 295 |
+ \ |
|
| 296 |
+static void OPNAME ## qpel8_mc13_c(uint8_t *dst, uint8_t *src, \ |
|
| 297 |
+ ptrdiff_t stride) \ |
|
| 298 |
+{ \
|
|
| 299 |
+ uint8_t full[16 * 9]; \ |
|
| 300 |
+ uint8_t halfH[72]; \ |
|
| 301 |
+ uint8_t halfHV[64]; \ |
|
| 302 |
+ \ |
|
| 303 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 304 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 305 |
+ put ## RND ## pixels8_l2_8(halfH, halfH, full, 8, 8, 16, 9); \ |
|
| 306 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 307 |
+ OPNAME ## pixels8_l2_8(dst, halfH + 8, halfHV, stride, 8, 8, 8); \ |
|
| 308 |
+} \ |
|
| 309 |
+ \ |
|
| 310 |
+void ff_ ## OPNAME ## qpel8_mc33_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 311 |
+ ptrdiff_t stride) \ |
|
| 312 |
+{ \
|
|
| 313 |
+ uint8_t full[16 * 9]; \ |
|
| 314 |
+ uint8_t halfH[72]; \ |
|
| 315 |
+ uint8_t halfV[64]; \ |
|
| 316 |
+ uint8_t halfHV[64]; \ |
|
| 317 |
+ \ |
|
| 318 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 319 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 320 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full + 1, 8, 16); \ |
|
| 321 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 322 |
+ OPNAME ## pixels8_l4_8(dst, full + 17, halfH + 8, halfV, halfHV, \ |
|
| 323 |
+ stride, 16, 8, 8, 8, 8); \ |
|
| 324 |
+} \ |
|
| 325 |
+ \ |
|
| 326 |
+static void OPNAME ## qpel8_mc33_c(uint8_t *dst, uint8_t *src, \ |
|
| 327 |
+ ptrdiff_t stride) \ |
|
| 328 |
+{ \
|
|
| 329 |
+ uint8_t full[16 * 9]; \ |
|
| 330 |
+ uint8_t halfH[72]; \ |
|
| 331 |
+ uint8_t halfHV[64]; \ |
|
| 332 |
+ \ |
|
| 333 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 334 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 335 |
+ put ## RND ## pixels8_l2_8(halfH, halfH, full + 1, 8, 8, 16, 9); \ |
|
| 336 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 337 |
+ OPNAME ## pixels8_l2_8(dst, halfH + 8, halfHV, stride, 8, 8, 8); \ |
|
| 338 |
+} \ |
|
| 339 |
+ \ |
|
| 340 |
+static void OPNAME ## qpel8_mc21_c(uint8_t *dst, uint8_t *src, \ |
|
| 341 |
+ ptrdiff_t stride) \ |
|
| 342 |
+{ \
|
|
| 343 |
+ uint8_t halfH[72]; \ |
|
| 344 |
+ uint8_t halfHV[64]; \ |
|
| 345 |
+ \ |
|
| 346 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9); \ |
|
| 347 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 348 |
+ OPNAME ## pixels8_l2_8(dst, halfH, halfHV, stride, 8, 8, 8); \ |
|
| 349 |
+} \ |
|
| 350 |
+ \ |
|
| 351 |
+static void OPNAME ## qpel8_mc23_c(uint8_t *dst, uint8_t *src, \ |
|
| 352 |
+ ptrdiff_t stride) \ |
|
| 353 |
+{ \
|
|
| 354 |
+ uint8_t halfH[72]; \ |
|
| 355 |
+ uint8_t halfHV[64]; \ |
|
| 356 |
+ \ |
|
| 357 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9); \ |
|
| 358 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 359 |
+ OPNAME ## pixels8_l2_8(dst, halfH + 8, halfHV, stride, 8, 8, 8); \ |
|
| 360 |
+} \ |
|
| 361 |
+ \ |
|
| 362 |
+void ff_ ## OPNAME ## qpel8_mc12_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 363 |
+ ptrdiff_t stride) \ |
|
| 364 |
+{ \
|
|
| 365 |
+ uint8_t full[16 * 9]; \ |
|
| 366 |
+ uint8_t halfH[72]; \ |
|
| 367 |
+ uint8_t halfV[64]; \ |
|
| 368 |
+ uint8_t halfHV[64]; \ |
|
| 369 |
+ \ |
|
| 370 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 371 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 372 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full, 8, 16); \ |
|
| 373 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 374 |
+ OPNAME ## pixels8_l2_8(dst, halfV, halfHV, stride, 8, 8, 8); \ |
|
| 375 |
+} \ |
|
| 376 |
+ \ |
|
| 377 |
+static void OPNAME ## qpel8_mc12_c(uint8_t *dst, uint8_t *src, \ |
|
| 378 |
+ ptrdiff_t stride) \ |
|
| 379 |
+{ \
|
|
| 380 |
+ uint8_t full[16 * 9]; \ |
|
| 381 |
+ uint8_t halfH[72]; \ |
|
| 382 |
+ \ |
|
| 383 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 384 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 385 |
+ put ## RND ## pixels8_l2_8(halfH, halfH, full, 8, 8, 16, 9); \ |
|
| 386 |
+ OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8); \ |
|
| 387 |
+} \ |
|
| 388 |
+ \ |
|
| 389 |
+void ff_ ## OPNAME ## qpel8_mc32_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 390 |
+ ptrdiff_t stride) \ |
|
| 391 |
+{ \
|
|
| 392 |
+ uint8_t full[16 * 9]; \ |
|
| 393 |
+ uint8_t halfH[72]; \ |
|
| 394 |
+ uint8_t halfV[64]; \ |
|
| 395 |
+ uint8_t halfHV[64]; \ |
|
| 396 |
+ \ |
|
| 397 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 398 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 399 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full + 1, 8, 16); \ |
|
| 400 |
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8); \ |
|
| 401 |
+ OPNAME ## pixels8_l2_8(dst, halfV, halfHV, stride, 8, 8, 8); \ |
|
| 402 |
+} \ |
|
| 403 |
+ \ |
|
| 404 |
+static void OPNAME ## qpel8_mc32_c(uint8_t *dst, uint8_t *src, \ |
|
| 405 |
+ ptrdiff_t stride) \ |
|
| 406 |
+{ \
|
|
| 407 |
+ uint8_t full[16 * 9]; \ |
|
| 408 |
+ uint8_t halfH[72]; \ |
|
| 409 |
+ \ |
|
| 410 |
+ copy_block9(full, src, 16, stride, 9); \ |
|
| 411 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9); \ |
|
| 412 |
+ put ## RND ## pixels8_l2_8(halfH, halfH, full + 1, 8, 8, 16, 9); \ |
|
| 413 |
+ OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8); \ |
|
| 414 |
+} \ |
|
| 415 |
+ \ |
|
| 416 |
+static void OPNAME ## qpel8_mc22_c(uint8_t *dst, uint8_t *src, \ |
|
| 417 |
+ ptrdiff_t stride) \ |
|
| 418 |
+{ \
|
|
| 419 |
+ uint8_t halfH[72]; \ |
|
| 420 |
+ \ |
|
| 421 |
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9); \ |
|
| 422 |
+ OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8); \ |
|
| 423 |
+} \ |
|
| 424 |
+ \ |
|
| 425 |
+static void OPNAME ## qpel16_mc10_c(uint8_t *dst, uint8_t *src, \ |
|
| 426 |
+ ptrdiff_t stride) \ |
|
| 427 |
+{ \
|
|
| 428 |
+ uint8_t half[256]; \ |
|
| 429 |
+ \ |
|
| 430 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(half, src, 16, stride, 16); \ |
|
| 431 |
+ OPNAME ## pixels16_l2_8(dst, src, half, stride, stride, 16, 16); \ |
|
| 432 |
+} \ |
|
| 433 |
+ \ |
|
| 434 |
+static void OPNAME ## qpel16_mc20_c(uint8_t *dst, uint8_t *src, \ |
|
| 435 |
+ ptrdiff_t stride) \ |
|
| 436 |
+{ \
|
|
| 437 |
+ OPNAME ## mpeg4_qpel16_h_lowpass(dst, src, stride, stride, 16); \ |
|
| 438 |
+} \ |
|
| 439 |
+ \ |
|
| 440 |
+static void OPNAME ## qpel16_mc30_c(uint8_t *dst, uint8_t *src, \ |
|
| 441 |
+ ptrdiff_t stride) \ |
|
| 442 |
+{ \
|
|
| 443 |
+ uint8_t half[256]; \ |
|
| 444 |
+ \ |
|
| 445 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(half, src, 16, stride, 16); \ |
|
| 446 |
+ OPNAME ## pixels16_l2_8(dst, src + 1, half, stride, stride, 16, 16); \ |
|
| 447 |
+} \ |
|
| 448 |
+ \ |
|
| 449 |
+static void OPNAME ## qpel16_mc01_c(uint8_t *dst, uint8_t *src, \ |
|
| 450 |
+ ptrdiff_t stride) \ |
|
| 451 |
+{ \
|
|
| 452 |
+ uint8_t full[24 * 17]; \ |
|
| 453 |
+ uint8_t half[256]; \ |
|
| 454 |
+ \ |
|
| 455 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 456 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(half, full, 16, 24); \ |
|
| 457 |
+ OPNAME ## pixels16_l2_8(dst, full, half, stride, 24, 16, 16); \ |
|
| 458 |
+} \ |
|
| 459 |
+ \ |
|
| 460 |
+static void OPNAME ## qpel16_mc02_c(uint8_t *dst, uint8_t *src, \ |
|
| 461 |
+ ptrdiff_t stride) \ |
|
| 462 |
+{ \
|
|
| 463 |
+ uint8_t full[24 * 17]; \ |
|
| 464 |
+ \ |
|
| 465 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 466 |
+ OPNAME ## mpeg4_qpel16_v_lowpass(dst, full, stride, 24); \ |
|
| 467 |
+} \ |
|
| 468 |
+ \ |
|
| 469 |
+static void OPNAME ## qpel16_mc03_c(uint8_t *dst, uint8_t *src, \ |
|
| 470 |
+ ptrdiff_t stride) \ |
|
| 471 |
+{ \
|
|
| 472 |
+ uint8_t full[24 * 17]; \ |
|
| 473 |
+ uint8_t half[256]; \ |
|
| 474 |
+ \ |
|
| 475 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 476 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(half, full, 16, 24); \ |
|
| 477 |
+ OPNAME ## pixels16_l2_8(dst, full + 24, half, stride, 24, 16, 16); \ |
|
| 478 |
+} \ |
|
| 479 |
+ \ |
|
| 480 |
+void ff_ ## OPNAME ## qpel16_mc11_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 481 |
+ ptrdiff_t stride) \ |
|
| 482 |
+{ \
|
|
| 483 |
+ uint8_t full[24 * 17]; \ |
|
| 484 |
+ uint8_t halfH[272]; \ |
|
| 485 |
+ uint8_t halfV[256]; \ |
|
| 486 |
+ uint8_t halfHV[256]; \ |
|
| 487 |
+ \ |
|
| 488 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 489 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 490 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full, 16, 24); \ |
|
| 491 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 492 |
+ OPNAME ## pixels16_l4_8(dst, full, halfH, halfV, halfHV, \ |
|
| 493 |
+ stride, 24, 16, 16, 16, 16); \ |
|
| 494 |
+} \ |
|
| 495 |
+ \ |
|
| 496 |
+static void OPNAME ## qpel16_mc11_c(uint8_t *dst, uint8_t *src, \ |
|
| 497 |
+ ptrdiff_t stride) \ |
|
| 498 |
+{ \
|
|
| 499 |
+ uint8_t full[24 * 17]; \ |
|
| 500 |
+ uint8_t halfH[272]; \ |
|
| 501 |
+ uint8_t halfHV[256]; \ |
|
| 502 |
+ \ |
|
| 503 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 504 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 505 |
+ put ## RND ## pixels16_l2_8(halfH, halfH, full, 16, 16, 24, 17); \ |
|
| 506 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 507 |
+ OPNAME ## pixels16_l2_8(dst, halfH, halfHV, stride, 16, 16, 16); \ |
|
| 508 |
+} \ |
|
| 509 |
+ \ |
|
| 510 |
+void ff_ ## OPNAME ## qpel16_mc31_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 511 |
+ ptrdiff_t stride) \ |
|
| 512 |
+{ \
|
|
| 513 |
+ uint8_t full[24 * 17]; \ |
|
| 514 |
+ uint8_t halfH[272]; \ |
|
| 515 |
+ uint8_t halfV[256]; \ |
|
| 516 |
+ uint8_t halfHV[256]; \ |
|
| 517 |
+ \ |
|
| 518 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 519 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 520 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full + 1, 16, 24); \ |
|
| 521 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 522 |
+ OPNAME ## pixels16_l4_8(dst, full + 1, halfH, halfV, halfHV, \ |
|
| 523 |
+ stride, 24, 16, 16, 16, 16); \ |
|
| 524 |
+} \ |
|
| 525 |
+ \ |
|
| 526 |
+static void OPNAME ## qpel16_mc31_c(uint8_t *dst, uint8_t *src, \ |
|
| 527 |
+ ptrdiff_t stride) \ |
|
| 528 |
+{ \
|
|
| 529 |
+ uint8_t full[24 * 17]; \ |
|
| 530 |
+ uint8_t halfH[272]; \ |
|
| 531 |
+ uint8_t halfHV[256]; \ |
|
| 532 |
+ \ |
|
| 533 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 534 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 535 |
+ put ## RND ## pixels16_l2_8(halfH, halfH, full + 1, 16, 16, 24, 17); \ |
|
| 536 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 537 |
+ OPNAME ## pixels16_l2_8(dst, halfH, halfHV, stride, 16, 16, 16); \ |
|
| 538 |
+} \ |
|
| 539 |
+ \ |
|
| 540 |
+void ff_ ## OPNAME ## qpel16_mc13_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 541 |
+ ptrdiff_t stride) \ |
|
| 542 |
+{ \
|
|
| 543 |
+ uint8_t full[24 * 17]; \ |
|
| 544 |
+ uint8_t halfH[272]; \ |
|
| 545 |
+ uint8_t halfV[256]; \ |
|
| 546 |
+ uint8_t halfHV[256]; \ |
|
| 547 |
+ \ |
|
| 548 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 549 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 550 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full, 16, 24); \ |
|
| 551 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 552 |
+ OPNAME ## pixels16_l4_8(dst, full + 24, halfH + 16, halfV, halfHV, \ |
|
| 553 |
+ stride, 24, 16, 16, 16, 16); \ |
|
| 554 |
+} \ |
|
| 555 |
+ \ |
|
| 556 |
+static void OPNAME ## qpel16_mc13_c(uint8_t *dst, uint8_t *src, \ |
|
| 557 |
+ ptrdiff_t stride) \ |
|
| 558 |
+{ \
|
|
| 559 |
+ uint8_t full[24 * 17]; \ |
|
| 560 |
+ uint8_t halfH[272]; \ |
|
| 561 |
+ uint8_t halfHV[256]; \ |
|
| 562 |
+ \ |
|
| 563 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 564 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 565 |
+ put ## RND ## pixels16_l2_8(halfH, halfH, full, 16, 16, 24, 17); \ |
|
| 566 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 567 |
+ OPNAME ## pixels16_l2_8(dst, halfH + 16, halfHV, stride, 16, 16, 16); \ |
|
| 568 |
+} \ |
|
| 569 |
+ \ |
|
| 570 |
+void ff_ ## OPNAME ## qpel16_mc33_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 571 |
+ ptrdiff_t stride) \ |
|
| 572 |
+{ \
|
|
| 573 |
+ uint8_t full[24 * 17]; \ |
|
| 574 |
+ uint8_t halfH[272]; \ |
|
| 575 |
+ uint8_t halfV[256]; \ |
|
| 576 |
+ uint8_t halfHV[256]; \ |
|
| 577 |
+ \ |
|
| 578 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 579 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 580 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full + 1, 16, 24); \ |
|
| 581 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 582 |
+ OPNAME ## pixels16_l4_8(dst, full + 25, halfH + 16, halfV, halfHV, \ |
|
| 583 |
+ stride, 24, 16, 16, 16, 16); \ |
|
| 584 |
+} \ |
|
| 585 |
+ \ |
|
| 586 |
+static void OPNAME ## qpel16_mc33_c(uint8_t *dst, uint8_t *src, \ |
|
| 587 |
+ ptrdiff_t stride) \ |
|
| 588 |
+{ \
|
|
| 589 |
+ uint8_t full[24 * 17]; \ |
|
| 590 |
+ uint8_t halfH[272]; \ |
|
| 591 |
+ uint8_t halfHV[256]; \ |
|
| 592 |
+ \ |
|
| 593 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 594 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 595 |
+ put ## RND ## pixels16_l2_8(halfH, halfH, full + 1, 16, 16, 24, 17); \ |
|
| 596 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 597 |
+ OPNAME ## pixels16_l2_8(dst, halfH + 16, halfHV, stride, 16, 16, 16); \ |
|
| 598 |
+} \ |
|
| 599 |
+ \ |
|
| 600 |
+static void OPNAME ## qpel16_mc21_c(uint8_t *dst, uint8_t *src, \ |
|
| 601 |
+ ptrdiff_t stride) \ |
|
| 602 |
+{ \
|
|
| 603 |
+ uint8_t halfH[272]; \ |
|
| 604 |
+ uint8_t halfHV[256]; \ |
|
| 605 |
+ \ |
|
| 606 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17); \ |
|
| 607 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 608 |
+ OPNAME ## pixels16_l2_8(dst, halfH, halfHV, stride, 16, 16, 16); \ |
|
| 609 |
+} \ |
|
| 610 |
+ \ |
|
| 611 |
+static void OPNAME ## qpel16_mc23_c(uint8_t *dst, uint8_t *src, \ |
|
| 612 |
+ ptrdiff_t stride) \ |
|
| 613 |
+{ \
|
|
| 614 |
+ uint8_t halfH[272]; \ |
|
| 615 |
+ uint8_t halfHV[256]; \ |
|
| 616 |
+ \ |
|
| 617 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17); \ |
|
| 618 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 619 |
+ OPNAME ## pixels16_l2_8(dst, halfH + 16, halfHV, stride, 16, 16, 16); \ |
|
| 620 |
+} \ |
|
| 621 |
+ \ |
|
| 622 |
+void ff_ ## OPNAME ## qpel16_mc12_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 623 |
+ ptrdiff_t stride) \ |
|
| 624 |
+{ \
|
|
| 625 |
+ uint8_t full[24 * 17]; \ |
|
| 626 |
+ uint8_t halfH[272]; \ |
|
| 627 |
+ uint8_t halfV[256]; \ |
|
| 628 |
+ uint8_t halfHV[256]; \ |
|
| 629 |
+ \ |
|
| 630 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 631 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 632 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full, 16, 24); \ |
|
| 633 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 634 |
+ OPNAME ## pixels16_l2_8(dst, halfV, halfHV, stride, 16, 16, 16); \ |
|
| 635 |
+} \ |
|
| 636 |
+ \ |
|
| 637 |
+static void OPNAME ## qpel16_mc12_c(uint8_t *dst, uint8_t *src, \ |
|
| 638 |
+ ptrdiff_t stride) \ |
|
| 639 |
+{ \
|
|
| 640 |
+ uint8_t full[24 * 17]; \ |
|
| 641 |
+ uint8_t halfH[272]; \ |
|
| 642 |
+ \ |
|
| 643 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 644 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 645 |
+ put ## RND ## pixels16_l2_8(halfH, halfH, full, 16, 16, 24, 17); \ |
|
| 646 |
+ OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16); \ |
|
| 647 |
+} \ |
|
| 648 |
+ \ |
|
| 649 |
+void ff_ ## OPNAME ## qpel16_mc32_old_c(uint8_t *dst, uint8_t *src, \ |
|
| 650 |
+ ptrdiff_t stride) \ |
|
| 651 |
+{ \
|
|
| 652 |
+ uint8_t full[24 * 17]; \ |
|
| 653 |
+ uint8_t halfH[272]; \ |
|
| 654 |
+ uint8_t halfV[256]; \ |
|
| 655 |
+ uint8_t halfHV[256]; \ |
|
| 656 |
+ \ |
|
| 657 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 658 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 659 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full + 1, 16, 24); \ |
|
| 660 |
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16); \ |
|
| 661 |
+ OPNAME ## pixels16_l2_8(dst, halfV, halfHV, stride, 16, 16, 16); \ |
|
| 662 |
+} \ |
|
| 663 |
+ \ |
|
| 664 |
+static void OPNAME ## qpel16_mc32_c(uint8_t *dst, uint8_t *src, \ |
|
| 665 |
+ ptrdiff_t stride) \ |
|
| 666 |
+{ \
|
|
| 667 |
+ uint8_t full[24 * 17]; \ |
|
| 668 |
+ uint8_t halfH[272]; \ |
|
| 669 |
+ \ |
|
| 670 |
+ copy_block17(full, src, 24, stride, 17); \ |
|
| 671 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17); \ |
|
| 672 |
+ put ## RND ## pixels16_l2_8(halfH, halfH, full + 1, 16, 16, 24, 17); \ |
|
| 673 |
+ OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16); \ |
|
| 674 |
+} \ |
|
| 675 |
+ \ |
|
| 676 |
+static void OPNAME ## qpel16_mc22_c(uint8_t *dst, uint8_t *src, \ |
|
| 677 |
+ ptrdiff_t stride) \ |
|
| 678 |
+{ \
|
|
| 679 |
+ uint8_t halfH[272]; \ |
|
| 680 |
+ \ |
|
| 681 |
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17); \ |
|
| 682 |
+ OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16); \ |
|
| 683 |
+} |
|
| 684 |
+ |
|
| 685 |
+#define op_avg(a, b) a = (((a) + cm[((b) + 16) >> 5] + 1) >> 1) |
|
| 686 |
+#define op_put(a, b) a = cm[((b) + 16) >> 5] |
|
| 687 |
+#define op_put_no_rnd(a, b) a = cm[((b) + 15) >> 5] |
|
| 688 |
+ |
|
| 689 |
+QPEL_MC(0, put_, _, op_put) |
|
| 690 |
+QPEL_MC(1, put_no_rnd_, _no_rnd_, op_put_no_rnd) |
|
| 691 |
+QPEL_MC(0, avg_, _, op_avg) |
|
| 692 |
+ |
|
| 693 |
+#undef op_avg |
|
| 694 |
+#undef op_put |
|
| 695 |
+#undef op_put_no_rnd |
|
| 696 |
+ |
|
| 697 |
+void ff_put_pixels8x8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) |
|
| 698 |
+{
|
|
| 699 |
+ put_pixels8_8_c(dst, src, stride, 8); |
|
| 700 |
+} |
|
| 701 |
+ |
|
| 702 |
+void ff_avg_pixels8x8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) |
|
| 703 |
+{
|
|
| 704 |
+ avg_pixels8_8_c(dst, src, stride, 8); |
|
| 705 |
+} |
|
| 706 |
+ |
|
| 707 |
+void ff_put_pixels16x16_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) |
|
| 708 |
+{
|
|
| 709 |
+ put_pixels16_8_c(dst, src, stride, 16); |
|
| 710 |
+} |
|
| 711 |
+ |
|
| 712 |
+void ff_avg_pixels16x16_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) |
|
| 713 |
+{
|
|
| 714 |
+ avg_pixels16_8_c(dst, src, stride, 16); |
|
| 715 |
+} |
|
| 716 |
+ |
|
| 717 |
+#define put_qpel8_mc00_c ff_put_pixels8x8_c |
|
| 718 |
+#define avg_qpel8_mc00_c ff_avg_pixels8x8_c |
|
| 719 |
+#define put_qpel16_mc00_c ff_put_pixels16x16_c |
|
| 720 |
+#define avg_qpel16_mc00_c ff_avg_pixels16x16_c |
|
| 721 |
+#define put_no_rnd_qpel8_mc00_c ff_put_pixels8x8_c |
|
| 722 |
+#define put_no_rnd_qpel16_mc00_c ff_put_pixels16x16_c |
|
| 723 |
+ |
|
| 724 |
+void ff_put_pixels8_l2_8(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, |
|
| 725 |
+ int dst_stride, int src_stride1, int src_stride2, |
|
| 726 |
+ int h) |
|
| 727 |
+{
|
|
| 728 |
+ put_pixels8_l2_8(dst, src1, src2, dst_stride, src_stride1, src_stride2, h); |
|
| 729 |
+ |
|
| 730 |
+} |
|
| 731 |
+ |
|
| 732 |
+av_cold void ff_qpeldsp_init(QpelDSPContext *c) |
|
| 733 |
+{
|
|
| 734 |
+#define dspfunc(PFX, IDX, NUM) \ |
|
| 735 |
+ c->PFX ## _pixels_tab[IDX][0] = PFX ## NUM ## _mc00_c; \ |
|
| 736 |
+ c->PFX ## _pixels_tab[IDX][1] = PFX ## NUM ## _mc10_c; \ |
|
| 737 |
+ c->PFX ## _pixels_tab[IDX][2] = PFX ## NUM ## _mc20_c; \ |
|
| 738 |
+ c->PFX ## _pixels_tab[IDX][3] = PFX ## NUM ## _mc30_c; \ |
|
| 739 |
+ c->PFX ## _pixels_tab[IDX][4] = PFX ## NUM ## _mc01_c; \ |
|
| 740 |
+ c->PFX ## _pixels_tab[IDX][5] = PFX ## NUM ## _mc11_c; \ |
|
| 741 |
+ c->PFX ## _pixels_tab[IDX][6] = PFX ## NUM ## _mc21_c; \ |
|
| 742 |
+ c->PFX ## _pixels_tab[IDX][7] = PFX ## NUM ## _mc31_c; \ |
|
| 743 |
+ c->PFX ## _pixels_tab[IDX][8] = PFX ## NUM ## _mc02_c; \ |
|
| 744 |
+ c->PFX ## _pixels_tab[IDX][9] = PFX ## NUM ## _mc12_c; \ |
|
| 745 |
+ c->PFX ## _pixels_tab[IDX][10] = PFX ## NUM ## _mc22_c; \ |
|
| 746 |
+ c->PFX ## _pixels_tab[IDX][11] = PFX ## NUM ## _mc32_c; \ |
|
| 747 |
+ c->PFX ## _pixels_tab[IDX][12] = PFX ## NUM ## _mc03_c; \ |
|
| 748 |
+ c->PFX ## _pixels_tab[IDX][13] = PFX ## NUM ## _mc13_c; \ |
|
| 749 |
+ c->PFX ## _pixels_tab[IDX][14] = PFX ## NUM ## _mc23_c; \ |
|
| 750 |
+ c->PFX ## _pixels_tab[IDX][15] = PFX ## NUM ## _mc33_c |
|
| 751 |
+ |
|
| 752 |
+ dspfunc(put_qpel, 0, 16); |
|
| 753 |
+ dspfunc(put_qpel, 1, 8); |
|
| 754 |
+ |
|
| 755 |
+ dspfunc(put_no_rnd_qpel, 0, 16); |
|
| 756 |
+ dspfunc(put_no_rnd_qpel, 1, 8); |
|
| 757 |
+ |
|
| 758 |
+ dspfunc(avg_qpel, 0, 16); |
|
| 759 |
+ dspfunc(avg_qpel, 1, 8); |
|
| 760 |
+ |
|
| 761 |
+ if (ARCH_X86) |
|
| 762 |
+ ff_qpeldsp_init_x86(c); |
|
| 763 |
+} |
| 0 | 764 |
new file mode 100644 |
| ... | ... |
@@ -0,0 +1,78 @@ |
| 0 |
+/* |
|
| 1 |
+ * quarterpel DSP functions |
|
| 2 |
+ * |
|
| 3 |
+ * This file is part of Libav. |
|
| 4 |
+ * |
|
| 5 |
+ * Libav is free software; you can redistribute it and/or |
|
| 6 |
+ * modify it under the terms of the GNU Lesser General Public |
|
| 7 |
+ * License as published by the Free Software Foundation; either |
|
| 8 |
+ * version 2.1 of the License, or (at your option) any later version. |
|
| 9 |
+ * |
|
| 10 |
+ * Libav is distributed in the hope that it will be useful, |
|
| 11 |
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
| 12 |
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
| 13 |
+ * Lesser General Public License for more details. |
|
| 14 |
+ * |
|
| 15 |
+ * You should have received a copy of the GNU Lesser General Public |
|
| 16 |
+ * License along with Libav; if not, write to the Free Software |
|
| 17 |
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
| 18 |
+ */ |
|
| 19 |
+ |
|
| 20 |
+/** |
|
| 21 |
+ * @file |
|
| 22 |
+ * quarterpel DSP functions |
|
| 23 |
+ */ |
|
| 24 |
+ |
|
| 25 |
+#ifndef AVCODEC_QPELDSP_H |
|
| 26 |
+#define AVCODEC_QPELDSP_H |
|
| 27 |
+ |
|
| 28 |
+#include <stddef.h> |
|
| 29 |
+#include <stdint.h> |
|
| 30 |
+ |
|
| 31 |
+void ff_put_pixels8x8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); |
|
| 32 |
+void ff_avg_pixels8x8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); |
|
| 33 |
+void ff_put_pixels16x16_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); |
|
| 34 |
+void ff_avg_pixels16x16_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); |
|
| 35 |
+ |
|
| 36 |
+void ff_put_pixels8_l2_8(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, |
|
| 37 |
+ int dst_stride, int src_stride1, int src_stride2, |
|
| 38 |
+ int h); |
|
| 39 |
+ |
|
| 40 |
+#define DEF_OLD_QPEL(name) \ |
|
| 41 |
+ void ff_put_ ## name(uint8_t *dst /* align width (8 or 16) */, \ |
|
| 42 |
+ uint8_t *src /* align 1 */, ptrdiff_t stride); \ |
|
| 43 |
+ void ff_put_no_rnd_ ## name(uint8_t *dst /* align width (8 or 16) */, \ |
|
| 44 |
+ uint8_t *src /* align 1 */, ptrdiff_t stride); \ |
|
| 45 |
+ void ff_avg_ ## name(uint8_t *dst /* align width (8 or 16) */, \ |
|
| 46 |
+ uint8_t *src /* align 1 */, ptrdiff_t stride); |
|
| 47 |
+ |
|
| 48 |
+DEF_OLD_QPEL(qpel16_mc11_old_c) |
|
| 49 |
+DEF_OLD_QPEL(qpel16_mc31_old_c) |
|
| 50 |
+DEF_OLD_QPEL(qpel16_mc12_old_c) |
|
| 51 |
+DEF_OLD_QPEL(qpel16_mc32_old_c) |
|
| 52 |
+DEF_OLD_QPEL(qpel16_mc13_old_c) |
|
| 53 |
+DEF_OLD_QPEL(qpel16_mc33_old_c) |
|
| 54 |
+DEF_OLD_QPEL(qpel8_mc11_old_c) |
|
| 55 |
+DEF_OLD_QPEL(qpel8_mc31_old_c) |
|
| 56 |
+DEF_OLD_QPEL(qpel8_mc12_old_c) |
|
| 57 |
+DEF_OLD_QPEL(qpel8_mc32_old_c) |
|
| 58 |
+DEF_OLD_QPEL(qpel8_mc13_old_c) |
|
| 59 |
+DEF_OLD_QPEL(qpel8_mc33_old_c) |
|
| 60 |
+ |
|
| 61 |
+typedef void (*qpel_mc_func)(uint8_t *dst /* align width (8 or 16) */, |
|
| 62 |
+ uint8_t *src /* align 1 */, ptrdiff_t stride); |
|
| 63 |
+ |
|
| 64 |
+/** |
|
| 65 |
+ * quarterpel DSP context |
|
| 66 |
+ */ |
|
| 67 |
+typedef struct QpelDSPContext {
|
|
| 68 |
+ qpel_mc_func put_qpel_pixels_tab[2][16]; |
|
| 69 |
+ qpel_mc_func avg_qpel_pixels_tab[2][16]; |
|
| 70 |
+ qpel_mc_func put_no_rnd_qpel_pixels_tab[2][16]; |
|
| 71 |
+} QpelDSPContext; |
|
| 72 |
+ |
|
| 73 |
+void ff_qpeldsp_init(QpelDSPContext *c); |
|
| 74 |
+ |
|
| 75 |
+void ff_qpeldsp_init_x86(QpelDSPContext *c); |
|
| 76 |
+ |
|
| 77 |
+#endif /* AVCODEC_QPELDSP_H */ |
| ... | ... |
@@ -27,8 +27,8 @@ |
| 27 | 27 |
#ifndef AVCODEC_RV34DSP_H |
| 28 | 28 |
#define AVCODEC_RV34DSP_H |
| 29 | 29 |
|
| 30 |
-#include "dsputil.h" |
|
| 31 | 30 |
#include "h264chroma.h" |
| 31 |
+#include "qpeldsp.h" |
|
| 32 | 32 |
|
| 33 | 33 |
typedef void (*rv40_weight_func)(uint8_t *dst/*align width (8 or 16)*/, |
| 34 | 34 |
uint8_t *src1/*align width (8 or 16)*/, |
| ... | ... |
@@ -33,6 +33,7 @@ |
| 33 | 33 |
#include "mpegvideo.h" |
| 34 | 34 |
#include "h263.h" |
| 35 | 35 |
#include "h264chroma.h" |
| 36 |
+#include "qpeldsp.h" |
|
| 36 | 37 |
#include "vc1.h" |
| 37 | 38 |
#include "vc1data.h" |
| 38 | 39 |
#include "vc1acdata.h" |
| ... | ... |
@@ -5603,6 +5604,7 @@ static av_cold int vc1_decode_init(AVCodecContext *avctx) |
| 5603 | 5603 |
if (ff_vc1_init_common(v) < 0) |
| 5604 | 5604 |
return -1; |
| 5605 | 5605 |
ff_h264chroma_init(&v->h264chroma, 8); |
| 5606 |
+ ff_qpeldsp_init(&s->qdsp); |
|
| 5606 | 5607 |
ff_vc1dsp_init(&v->vc1dsp); |
| 5607 | 5608 |
|
| 5608 | 5609 |
if (avctx->codec_id == AV_CODEC_ID_WMV3 || avctx->codec_id == AV_CODEC_ID_WMV3IMAGE) {
|
| ... | ... |
@@ -5971,8 +5973,8 @@ static int vc1_decode_frame(AVCodecContext *avctx, void *data, |
| 5971 | 5971 |
s->current_picture_ptr->f->repeat_pict = v->rptfrm * 2; |
| 5972 | 5972 |
} |
| 5973 | 5973 |
|
| 5974 |
- s->me.qpel_put = s->dsp.put_qpel_pixels_tab; |
|
| 5975 |
- s->me.qpel_avg = s->dsp.avg_qpel_pixels_tab; |
|
| 5974 |
+ s->me.qpel_put = s->qdsp.put_qpel_pixels_tab; |
|
| 5975 |
+ s->me.qpel_avg = s->qdsp.avg_qpel_pixels_tab; |
|
| 5976 | 5976 |
|
| 5977 | 5977 |
if (avctx->hwaccel) {
|
| 5978 | 5978 |
if (avctx->hwaccel->start_frame(avctx, buf, buf_size) < 0) |
| ... | ... |
@@ -20,6 +20,7 @@ OBJS-$(CONFIG_LPC) += x86/lpc.o |
| 20 | 20 |
OBJS-$(CONFIG_MPEGAUDIODSP) += x86/mpegaudiodsp.o |
| 21 | 21 |
OBJS-$(CONFIG_MPEGVIDEO) += x86/mpegvideo.o |
| 22 | 22 |
OBJS-$(CONFIG_MPEGVIDEOENC) += x86/mpegvideoenc.o |
| 23 |
+OBJS-$(CONFIG_QPELDSP) += x86/qpeldsp_init.o |
|
| 23 | 24 |
OBJS-$(CONFIG_VIDEODSP) += x86/videodsp_init.o |
| 24 | 25 |
OBJS-$(CONFIG_VP3DSP) += x86/vp3dsp_init.o |
| 25 | 26 |
OBJS-$(CONFIG_XMM_CLOBBER_TEST) += x86/w64xmmtest.o |
| ... | ... |
@@ -44,13 +45,13 @@ OBJS-$(CONFIG_VP8_DECODER) += x86/vp8dsp_init.o |
| 44 | 44 |
OBJS-$(CONFIG_VP9_DECODER) += x86/vp9dsp_init.o |
| 45 | 45 |
|
| 46 | 46 |
MMX-OBJS-$(CONFIG_DSPUTIL) += x86/dsputil_mmx.o \ |
| 47 |
- x86/fpel_mmx.o \ |
|
| 48 | 47 |
x86/idct_mmx_xvid.o \ |
| 49 | 48 |
x86/idct_sse2_xvid.o \ |
| 50 | 49 |
x86/simple_idct.o |
| 51 | 50 |
MMX-OBJS-$(CONFIG_HPELDSP) += x86/fpel_mmx.o \ |
| 52 | 51 |
x86/hpeldsp_mmx.o |
| 53 | 52 |
MMX-OBJS-$(CONFIG_HUFFYUVDSP) += x86/huffyuvdsp_mmx.o |
| 53 |
+MMX-OBJS-$(CONFIG_QPELDSP) += x86/fpel_mmx.o |
|
| 54 | 54 |
|
| 55 | 55 |
MMX-OBJS-$(CONFIG_SVQ1_ENCODER) += x86/svq1enc_mmx.o |
| 56 | 56 |
MMX-OBJS-$(CONFIG_VC1_DECODER) += x86/vc1dsp_mmx.o |
| ... | ... |
@@ -61,10 +62,7 @@ YASM-OBJS += x86/deinterlace.o \ |
| 61 | 61 |
YASM-OBJS-$(CONFIG_AC3DSP) += x86/ac3dsp.o |
| 62 | 62 |
YASM-OBJS-$(CONFIG_DCT) += x86/dct32.o |
| 63 | 63 |
YASM-OBJS-$(CONFIG_DNXHD_ENCODER) += x86/dnxhdenc.o |
| 64 |
-YASM-OBJS-$(CONFIG_DSPUTIL) += x86/dsputil.o \ |
|
| 65 |
- x86/fpel.o \ |
|
| 66 |
- x86/mpeg4qpel.o \ |
|
| 67 |
- x86/qpel.o |
|
| 64 |
+YASM-OBJS-$(CONFIG_DSPUTIL) += x86/dsputil.o |
|
| 68 | 65 |
YASM-OBJS-$(CONFIG_ENCODERS) += x86/dsputilenc.o |
| 69 | 66 |
YASM-OBJS-$(CONFIG_FFT) += x86/fft.o |
| 70 | 67 |
YASM-OBJS-$(CONFIG_H263DSP) += x86/h263_loopfilter.o |
| ... | ... |
@@ -86,6 +84,9 @@ YASM-OBJS-$(CONFIG_HPELDSP) += x86/fpel.o \ |
| 86 | 86 |
x86/hpeldsp.o |
| 87 | 87 |
YASM-OBJS-$(CONFIG_HUFFYUVDSP) += x86/huffyuvdsp.o |
| 88 | 88 |
YASM-OBJS-$(CONFIG_MPEGAUDIODSP) += x86/imdct36.o |
| 89 |
+YASM-OBJS-$(CONFIG_QPELDSP) += x86/qpeldsp.o \ |
|
| 90 |
+ x86/fpel.o \ |
|
| 91 |
+ x86/qpel.o |
|
| 89 | 92 |
YASM-OBJS-$(CONFIG_VIDEODSP) += x86/videodsp.o |
| 90 | 93 |
YASM-OBJS-$(CONFIG_VP3DSP) += x86/vp3dsp.o |
| 91 | 94 |
|
| ... | ... |
@@ -23,55 +23,11 @@ |
| 23 | 23 |
#include "libavutil/x86/cpu.h" |
| 24 | 24 |
#include "libavcodec/avcodec.h" |
| 25 | 25 |
#include "libavcodec/dsputil.h" |
| 26 |
-#include "libavcodec/pixels.h" |
|
| 27 | 26 |
#include "libavcodec/simple_idct.h" |
| 28 | 27 |
#include "libavcodec/version.h" |
| 29 | 28 |
#include "dsputil_x86.h" |
| 30 |
-#include "fpel.h" |
|
| 31 | 29 |
#include "idct_xvid.h" |
| 32 | 30 |
|
| 33 |
-void ff_put_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, |
|
| 34 |
- int dstStride, int src1Stride, int h); |
|
| 35 |
-void ff_put_no_rnd_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, |
|
| 36 |
- uint8_t *src2, int dstStride, |
|
| 37 |
- int src1Stride, int h); |
|
| 38 |
-void ff_avg_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, |
|
| 39 |
- int dstStride, int src1Stride, int h); |
|
| 40 |
-void ff_put_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, |
|
| 41 |
- int dstStride, int src1Stride, int h); |
|
| 42 |
-void ff_avg_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, |
|
| 43 |
- int dstStride, int src1Stride, int h); |
|
| 44 |
-void ff_put_no_rnd_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, |
|
| 45 |
- int dstStride, int src1Stride, int h); |
|
| 46 |
-void ff_put_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 47 |
- int dstStride, int srcStride, int h); |
|
| 48 |
-void ff_avg_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 49 |
- int dstStride, int srcStride, int h); |
|
| 50 |
-void ff_put_no_rnd_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 51 |
- int dstStride, int srcStride, |
|
| 52 |
- int h); |
|
| 53 |
-void ff_put_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 54 |
- int dstStride, int srcStride, int h); |
|
| 55 |
-void ff_avg_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 56 |
- int dstStride, int srcStride, int h); |
|
| 57 |
-void ff_put_no_rnd_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 58 |
- int dstStride, int srcStride, |
|
| 59 |
- int h); |
|
| 60 |
-void ff_put_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 61 |
- int dstStride, int srcStride); |
|
| 62 |
-void ff_avg_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 63 |
- int dstStride, int srcStride); |
|
| 64 |
-void ff_put_no_rnd_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 65 |
- int dstStride, int srcStride); |
|
| 66 |
-void ff_put_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 67 |
- int dstStride, int srcStride); |
|
| 68 |
-void ff_avg_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 69 |
- int dstStride, int srcStride); |
|
| 70 |
-void ff_put_no_rnd_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 71 |
- int dstStride, int srcStride); |
|
| 72 |
-#define ff_put_no_rnd_pixels16_mmxext ff_put_pixels16_mmxext |
|
| 73 |
-#define ff_put_no_rnd_pixels8_mmxext ff_put_pixels8_mmxext |
|
| 74 |
- |
|
| 75 | 31 |
int32_t ff_scalarproduct_int16_mmxext(const int16_t *v1, const int16_t *v2, |
| 76 | 32 |
int order); |
| 77 | 33 |
int32_t ff_scalarproduct_int16_sse2(const int16_t *v1, const int16_t *v2, |
| ... | ... |
@@ -89,418 +45,6 @@ void ff_vector_clip_int32_int_sse2(int32_t *dst, const int32_t *src, |
| 89 | 89 |
void ff_vector_clip_int32_sse4(int32_t *dst, const int32_t *src, |
| 90 | 90 |
int32_t min, int32_t max, unsigned int len); |
| 91 | 91 |
|
| 92 |
-#if HAVE_YASM |
|
| 93 |
- |
|
| 94 |
-CALL_2X_PIXELS(ff_avg_pixels16_mmxext, ff_avg_pixels8_mmxext, 8) |
|
| 95 |
-CALL_2X_PIXELS(ff_put_pixels16_mmxext, ff_put_pixels8_mmxext, 8) |
|
| 96 |
- |
|
| 97 |
-#define QPEL_OP(OPNAME, RND, MMX) \ |
|
| 98 |
-static void OPNAME ## qpel8_mc00_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 99 |
- ptrdiff_t stride) \ |
|
| 100 |
-{ \
|
|
| 101 |
- ff_ ## OPNAME ## pixels8_ ## MMX(dst, src, stride, 8); \ |
|
| 102 |
-} \ |
|
| 103 |
- \ |
|
| 104 |
-static void OPNAME ## qpel8_mc10_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 105 |
- ptrdiff_t stride) \ |
|
| 106 |
-{ \
|
|
| 107 |
- uint64_t temp[8]; \ |
|
| 108 |
- uint8_t *const half = (uint8_t *) temp; \ |
|
| 109 |
- ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(half, src, 8, \ |
|
| 110 |
- stride, 8); \ |
|
| 111 |
- ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src, half, \ |
|
| 112 |
- stride, stride, 8); \ |
|
| 113 |
-} \ |
|
| 114 |
- \ |
|
| 115 |
-static void OPNAME ## qpel8_mc20_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 116 |
- ptrdiff_t stride) \ |
|
| 117 |
-{ \
|
|
| 118 |
- ff_ ## OPNAME ## mpeg4_qpel8_h_lowpass_ ## MMX(dst, src, stride, \ |
|
| 119 |
- stride, 8); \ |
|
| 120 |
-} \ |
|
| 121 |
- \ |
|
| 122 |
-static void OPNAME ## qpel8_mc30_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 123 |
- ptrdiff_t stride) \ |
|
| 124 |
-{ \
|
|
| 125 |
- uint64_t temp[8]; \ |
|
| 126 |
- uint8_t *const half = (uint8_t *) temp; \ |
|
| 127 |
- ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(half, src, 8, \ |
|
| 128 |
- stride, 8); \ |
|
| 129 |
- ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src + 1, half, stride, \ |
|
| 130 |
- stride, 8); \ |
|
| 131 |
-} \ |
|
| 132 |
- \ |
|
| 133 |
-static void OPNAME ## qpel8_mc01_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 134 |
- ptrdiff_t stride) \ |
|
| 135 |
-{ \
|
|
| 136 |
- uint64_t temp[8]; \ |
|
| 137 |
- uint8_t *const half = (uint8_t *) temp; \ |
|
| 138 |
- ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(half, src, \ |
|
| 139 |
- 8, stride); \ |
|
| 140 |
- ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src, half, \ |
|
| 141 |
- stride, stride, 8); \ |
|
| 142 |
-} \ |
|
| 143 |
- \ |
|
| 144 |
-static void OPNAME ## qpel8_mc02_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 145 |
- ptrdiff_t stride) \ |
|
| 146 |
-{ \
|
|
| 147 |
- ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, src, \ |
|
| 148 |
- stride, stride); \ |
|
| 149 |
-} \ |
|
| 150 |
- \ |
|
| 151 |
-static void OPNAME ## qpel8_mc03_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 152 |
- ptrdiff_t stride) \ |
|
| 153 |
-{ \
|
|
| 154 |
- uint64_t temp[8]; \ |
|
| 155 |
- uint8_t *const half = (uint8_t *) temp; \ |
|
| 156 |
- ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(half, src, \ |
|
| 157 |
- 8, stride); \ |
|
| 158 |
- ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src + stride, half, stride,\ |
|
| 159 |
- stride, 8); \ |
|
| 160 |
-} \ |
|
| 161 |
- \ |
|
| 162 |
-static void OPNAME ## qpel8_mc11_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 163 |
- ptrdiff_t stride) \ |
|
| 164 |
-{ \
|
|
| 165 |
- uint64_t half[8 + 9]; \ |
|
| 166 |
- uint8_t *const halfH = (uint8_t *) half + 64; \ |
|
| 167 |
- uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 168 |
- ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 169 |
- stride, 9); \ |
|
| 170 |
- ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, 8, \ |
|
| 171 |
- stride, 9); \ |
|
| 172 |
- ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\ |
|
| 173 |
- ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV, \ |
|
| 174 |
- stride, 8, 8); \ |
|
| 175 |
-} \ |
|
| 176 |
- \ |
|
| 177 |
-static void OPNAME ## qpel8_mc31_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 178 |
- ptrdiff_t stride) \ |
|
| 179 |
-{ \
|
|
| 180 |
- uint64_t half[8 + 9]; \ |
|
| 181 |
- uint8_t *const halfH = (uint8_t *) half + 64; \ |
|
| 182 |
- uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 183 |
- ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 184 |
- stride, 9); \ |
|
| 185 |
- ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8, \ |
|
| 186 |
- stride, 9); \ |
|
| 187 |
- ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\ |
|
| 188 |
- ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV, \ |
|
| 189 |
- stride, 8, 8); \ |
|
| 190 |
-} \ |
|
| 191 |
- \ |
|
| 192 |
-static void OPNAME ## qpel8_mc13_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 193 |
- ptrdiff_t stride) \ |
|
| 194 |
-{ \
|
|
| 195 |
- uint64_t half[8 + 9]; \ |
|
| 196 |
- uint8_t *const halfH = (uint8_t *) half + 64; \ |
|
| 197 |
- uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 198 |
- ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 199 |
- stride, 9); \ |
|
| 200 |
- ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, 8, \ |
|
| 201 |
- stride, 9); \ |
|
| 202 |
- ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\ |
|
| 203 |
- ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV, \ |
|
| 204 |
- stride, 8, 8); \ |
|
| 205 |
-} \ |
|
| 206 |
- \ |
|
| 207 |
-static void OPNAME ## qpel8_mc33_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 208 |
- ptrdiff_t stride) \ |
|
| 209 |
-{ \
|
|
| 210 |
- uint64_t half[8 + 9]; \ |
|
| 211 |
- uint8_t *const halfH = (uint8_t *) half + 64; \ |
|
| 212 |
- uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 213 |
- ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 214 |
- stride, 9); \ |
|
| 215 |
- ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8, \ |
|
| 216 |
- stride, 9); \ |
|
| 217 |
- ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\ |
|
| 218 |
- ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV, \ |
|
| 219 |
- stride, 8, 8); \ |
|
| 220 |
-} \ |
|
| 221 |
- \ |
|
| 222 |
-static void OPNAME ## qpel8_mc21_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 223 |
- ptrdiff_t stride) \ |
|
| 224 |
-{ \
|
|
| 225 |
- uint64_t half[8 + 9]; \ |
|
| 226 |
- uint8_t *const halfH = (uint8_t *) half + 64; \ |
|
| 227 |
- uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 228 |
- ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 229 |
- stride, 9); \ |
|
| 230 |
- ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\ |
|
| 231 |
- ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV, \ |
|
| 232 |
- stride, 8, 8); \ |
|
| 233 |
-} \ |
|
| 234 |
- \ |
|
| 235 |
-static void OPNAME ## qpel8_mc23_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 236 |
- ptrdiff_t stride) \ |
|
| 237 |
-{ \
|
|
| 238 |
- uint64_t half[8 + 9]; \ |
|
| 239 |
- uint8_t *const halfH = (uint8_t *) half + 64; \ |
|
| 240 |
- uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 241 |
- ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 242 |
- stride, 9); \ |
|
| 243 |
- ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\ |
|
| 244 |
- ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV, \ |
|
| 245 |
- stride, 8, 8); \ |
|
| 246 |
-} \ |
|
| 247 |
- \ |
|
| 248 |
-static void OPNAME ## qpel8_mc12_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 249 |
- ptrdiff_t stride) \ |
|
| 250 |
-{ \
|
|
| 251 |
- uint64_t half[8 + 9]; \ |
|
| 252 |
- uint8_t *const halfH = (uint8_t *) half; \ |
|
| 253 |
- ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 254 |
- stride, 9); \ |
|
| 255 |
- ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, \ |
|
| 256 |
- 8, stride, 9); \ |
|
| 257 |
- ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH, \ |
|
| 258 |
- stride, 8); \ |
|
| 259 |
-} \ |
|
| 260 |
- \ |
|
| 261 |
-static void OPNAME ## qpel8_mc32_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 262 |
- ptrdiff_t stride) \ |
|
| 263 |
-{ \
|
|
| 264 |
- uint64_t half[8 + 9]; \ |
|
| 265 |
- uint8_t *const halfH = (uint8_t *) half; \ |
|
| 266 |
- ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 267 |
- stride, 9); \ |
|
| 268 |
- ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8, \ |
|
| 269 |
- stride, 9); \ |
|
| 270 |
- ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH, \ |
|
| 271 |
- stride, 8); \ |
|
| 272 |
-} \ |
|
| 273 |
- \ |
|
| 274 |
-static void OPNAME ## qpel8_mc22_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 275 |
- ptrdiff_t stride) \ |
|
| 276 |
-{ \
|
|
| 277 |
- uint64_t half[9]; \ |
|
| 278 |
- uint8_t *const halfH = (uint8_t *) half; \ |
|
| 279 |
- ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 280 |
- stride, 9); \ |
|
| 281 |
- ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH, \ |
|
| 282 |
- stride, 8); \ |
|
| 283 |
-} \ |
|
| 284 |
- \ |
|
| 285 |
-static void OPNAME ## qpel16_mc00_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 286 |
- ptrdiff_t stride) \ |
|
| 287 |
-{ \
|
|
| 288 |
- ff_ ## OPNAME ## pixels16_ ## MMX(dst, src, stride, 16); \ |
|
| 289 |
-} \ |
|
| 290 |
- \ |
|
| 291 |
-static void OPNAME ## qpel16_mc10_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 292 |
- ptrdiff_t stride) \ |
|
| 293 |
-{ \
|
|
| 294 |
- uint64_t temp[32]; \ |
|
| 295 |
- uint8_t *const half = (uint8_t *) temp; \ |
|
| 296 |
- ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(half, src, 16, \ |
|
| 297 |
- stride, 16); \ |
|
| 298 |
- ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src, half, stride, \ |
|
| 299 |
- stride, 16); \ |
|
| 300 |
-} \ |
|
| 301 |
- \ |
|
| 302 |
-static void OPNAME ## qpel16_mc20_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 303 |
- ptrdiff_t stride) \ |
|
| 304 |
-{ \
|
|
| 305 |
- ff_ ## OPNAME ## mpeg4_qpel16_h_lowpass_ ## MMX(dst, src, \ |
|
| 306 |
- stride, stride, 16);\ |
|
| 307 |
-} \ |
|
| 308 |
- \ |
|
| 309 |
-static void OPNAME ## qpel16_mc30_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 310 |
- ptrdiff_t stride) \ |
|
| 311 |
-{ \
|
|
| 312 |
- uint64_t temp[32]; \ |
|
| 313 |
- uint8_t *const half = (uint8_t*) temp; \ |
|
| 314 |
- ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(half, src, 16, \ |
|
| 315 |
- stride, 16); \ |
|
| 316 |
- ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src + 1, half, \ |
|
| 317 |
- stride, stride, 16); \ |
|
| 318 |
-} \ |
|
| 319 |
- \ |
|
| 320 |
-static void OPNAME ## qpel16_mc01_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 321 |
- ptrdiff_t stride) \ |
|
| 322 |
-{ \
|
|
| 323 |
- uint64_t temp[32]; \ |
|
| 324 |
- uint8_t *const half = (uint8_t *) temp; \ |
|
| 325 |
- ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(half, src, 16, \ |
|
| 326 |
- stride); \ |
|
| 327 |
- ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src, half, stride, \ |
|
| 328 |
- stride, 16); \ |
|
| 329 |
-} \ |
|
| 330 |
- \ |
|
| 331 |
-static void OPNAME ## qpel16_mc02_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 332 |
- ptrdiff_t stride) \ |
|
| 333 |
-{ \
|
|
| 334 |
- ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, src, \ |
|
| 335 |
- stride, stride); \ |
|
| 336 |
-} \ |
|
| 337 |
- \ |
|
| 338 |
-static void OPNAME ## qpel16_mc03_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 339 |
- ptrdiff_t stride) \ |
|
| 340 |
-{ \
|
|
| 341 |
- uint64_t temp[32]; \ |
|
| 342 |
- uint8_t *const half = (uint8_t *) temp; \ |
|
| 343 |
- ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(half, src, 16, \ |
|
| 344 |
- stride); \ |
|
| 345 |
- ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src+stride, half, \ |
|
| 346 |
- stride, stride, 16); \ |
|
| 347 |
-} \ |
|
| 348 |
- \ |
|
| 349 |
-static void OPNAME ## qpel16_mc11_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 350 |
- ptrdiff_t stride) \ |
|
| 351 |
-{ \
|
|
| 352 |
- uint64_t half[16 * 2 + 17 * 2]; \ |
|
| 353 |
- uint8_t *const halfH = (uint8_t *) half + 256; \ |
|
| 354 |
- uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 355 |
- ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 356 |
- stride, 17); \ |
|
| 357 |
- ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16, \ |
|
| 358 |
- stride, 17); \ |
|
| 359 |
- ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \ |
|
| 360 |
- 16, 16); \ |
|
| 361 |
- ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV, \ |
|
| 362 |
- stride, 16, 16); \ |
|
| 363 |
-} \ |
|
| 364 |
- \ |
|
| 365 |
-static void OPNAME ## qpel16_mc31_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 366 |
- ptrdiff_t stride) \ |
|
| 367 |
-{ \
|
|
| 368 |
- uint64_t half[16 * 2 + 17 * 2]; \ |
|
| 369 |
- uint8_t *const halfH = (uint8_t *) half + 256; \ |
|
| 370 |
- uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 371 |
- ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 372 |
- stride, 17); \ |
|
| 373 |
- ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16, \ |
|
| 374 |
- stride, 17); \ |
|
| 375 |
- ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \ |
|
| 376 |
- 16, 16); \ |
|
| 377 |
- ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV, \ |
|
| 378 |
- stride, 16, 16); \ |
|
| 379 |
-} \ |
|
| 380 |
- \ |
|
| 381 |
-static void OPNAME ## qpel16_mc13_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 382 |
- ptrdiff_t stride) \ |
|
| 383 |
-{ \
|
|
| 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, \ |
|
| 388 |
- stride, 17); \ |
|
| 389 |
- ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16, \ |
|
| 390 |
- stride, 17); \ |
|
| 391 |
- ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \ |
|
| 392 |
- 16, 16); \ |
|
| 393 |
- ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV, \ |
|
| 394 |
- stride, 16, 16); \ |
|
| 395 |
-} \ |
|
| 396 |
- \ |
|
| 397 |
-static void OPNAME ## qpel16_mc33_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 398 |
- ptrdiff_t stride) \ |
|
| 399 |
-{ \
|
|
| 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, \ |
|
| 404 |
- stride, 17); \ |
|
| 405 |
- ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16, \ |
|
| 406 |
- stride, 17); \ |
|
| 407 |
- ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \ |
|
| 408 |
- 16, 16); \ |
|
| 409 |
- ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV, \ |
|
| 410 |
- stride, 16, 16); \ |
|
| 411 |
-} \ |
|
| 412 |
- \ |
|
| 413 |
-static void OPNAME ## qpel16_mc21_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 414 |
- ptrdiff_t stride) \ |
|
| 415 |
-{ \
|
|
| 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, \ |
|
| 420 |
- stride, 17); \ |
|
| 421 |
- ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \ |
|
| 422 |
- 16, 16); \ |
|
| 423 |
- ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV, \ |
|
| 424 |
- stride, 16, 16); \ |
|
| 425 |
-} \ |
|
| 426 |
- \ |
|
| 427 |
-static void OPNAME ## qpel16_mc23_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 428 |
- ptrdiff_t stride) \ |
|
| 429 |
-{ \
|
|
| 430 |
- uint64_t half[16 * 2 + 17 * 2]; \ |
|
| 431 |
- uint8_t *const halfH = (uint8_t *) half + 256; \ |
|
| 432 |
- uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 433 |
- ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 434 |
- stride, 17); \ |
|
| 435 |
- ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \ |
|
| 436 |
- 16, 16); \ |
|
| 437 |
- ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV, \ |
|
| 438 |
- stride, 16, 16); \ |
|
| 439 |
-} \ |
|
| 440 |
- \ |
|
| 441 |
-static void OPNAME ## qpel16_mc12_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 442 |
- ptrdiff_t stride) \ |
|
| 443 |
-{ \
|
|
| 444 |
- uint64_t half[17 * 2]; \ |
|
| 445 |
- uint8_t *const halfH = (uint8_t *) half; \ |
|
| 446 |
- ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 447 |
- stride, 17); \ |
|
| 448 |
- ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16, \ |
|
| 449 |
- stride, 17); \ |
|
| 450 |
- ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH, \ |
|
| 451 |
- stride, 16); \ |
|
| 452 |
-} \ |
|
| 453 |
- \ |
|
| 454 |
-static void OPNAME ## qpel16_mc32_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 455 |
- ptrdiff_t stride) \ |
|
| 456 |
-{ \
|
|
| 457 |
- uint64_t half[17 * 2]; \ |
|
| 458 |
- uint8_t *const halfH = (uint8_t *) half; \ |
|
| 459 |
- ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 460 |
- stride, 17); \ |
|
| 461 |
- ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16, \ |
|
| 462 |
- stride, 17); \ |
|
| 463 |
- ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH, \ |
|
| 464 |
- stride, 16); \ |
|
| 465 |
-} \ |
|
| 466 |
- \ |
|
| 467 |
-static void OPNAME ## qpel16_mc22_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 468 |
- ptrdiff_t stride) \ |
|
| 469 |
-{ \
|
|
| 470 |
- uint64_t half[17 * 2]; \ |
|
| 471 |
- uint8_t *const halfH = (uint8_t *) half; \ |
|
| 472 |
- ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 473 |
- stride, 17); \ |
|
| 474 |
- ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH, \ |
|
| 475 |
- stride, 16); \ |
|
| 476 |
-} |
|
| 477 |
- |
|
| 478 |
-QPEL_OP(put_, _, mmxext) |
|
| 479 |
-QPEL_OP(avg_, _, mmxext) |
|
| 480 |
-QPEL_OP(put_no_rnd_, _no_rnd_, mmxext) |
|
| 481 |
- |
|
| 482 |
-#endif /* HAVE_YASM */ |
|
| 483 |
- |
|
| 484 |
-#define SET_QPEL_FUNCS(PFX, IDX, SIZE, CPU, PREFIX) \ |
|
| 485 |
-do { \
|
|
| 486 |
- c->PFX ## _pixels_tab[IDX][ 0] = PREFIX ## PFX ## SIZE ## _mc00_ ## CPU; \ |
|
| 487 |
- c->PFX ## _pixels_tab[IDX][ 1] = PREFIX ## PFX ## SIZE ## _mc10_ ## CPU; \ |
|
| 488 |
- c->PFX ## _pixels_tab[IDX][ 2] = PREFIX ## PFX ## SIZE ## _mc20_ ## CPU; \ |
|
| 489 |
- c->PFX ## _pixels_tab[IDX][ 3] = PREFIX ## PFX ## SIZE ## _mc30_ ## CPU; \ |
|
| 490 |
- c->PFX ## _pixels_tab[IDX][ 4] = PREFIX ## PFX ## SIZE ## _mc01_ ## CPU; \ |
|
| 491 |
- c->PFX ## _pixels_tab[IDX][ 5] = PREFIX ## PFX ## SIZE ## _mc11_ ## CPU; \ |
|
| 492 |
- c->PFX ## _pixels_tab[IDX][ 6] = PREFIX ## PFX ## SIZE ## _mc21_ ## CPU; \ |
|
| 493 |
- c->PFX ## _pixels_tab[IDX][ 7] = PREFIX ## PFX ## SIZE ## _mc31_ ## CPU; \ |
|
| 494 |
- c->PFX ## _pixels_tab[IDX][ 8] = PREFIX ## PFX ## SIZE ## _mc02_ ## CPU; \ |
|
| 495 |
- c->PFX ## _pixels_tab[IDX][ 9] = PREFIX ## PFX ## SIZE ## _mc12_ ## CPU; \ |
|
| 496 |
- c->PFX ## _pixels_tab[IDX][10] = PREFIX ## PFX ## SIZE ## _mc22_ ## CPU; \ |
|
| 497 |
- c->PFX ## _pixels_tab[IDX][11] = PREFIX ## PFX ## SIZE ## _mc32_ ## CPU; \ |
|
| 498 |
- c->PFX ## _pixels_tab[IDX][12] = PREFIX ## PFX ## SIZE ## _mc03_ ## CPU; \ |
|
| 499 |
- c->PFX ## _pixels_tab[IDX][13] = PREFIX ## PFX ## SIZE ## _mc13_ ## CPU; \ |
|
| 500 |
- c->PFX ## _pixels_tab[IDX][14] = PREFIX ## PFX ## SIZE ## _mc23_ ## CPU; \ |
|
| 501 |
- c->PFX ## _pixels_tab[IDX][15] = PREFIX ## PFX ## SIZE ## _mc33_ ## CPU; \ |
|
| 502 |
-} while (0) |
|
| 503 |
- |
|
| 504 | 92 |
static av_cold void dsputil_init_mmx(DSPContext *c, AVCodecContext *avctx, |
| 505 | 93 |
int cpu_flags, unsigned high_bit_depth) |
| 506 | 94 |
{
|
| ... | ... |
@@ -550,14 +94,6 @@ static av_cold void dsputil_init_mmxext(DSPContext *c, AVCodecContext *avctx, |
| 550 | 550 |
#endif /* HAVE_MMXEXT_INLINE */ |
| 551 | 551 |
|
| 552 | 552 |
#if HAVE_MMXEXT_EXTERNAL |
| 553 |
- SET_QPEL_FUNCS(avg_qpel, 0, 16, mmxext, ); |
|
| 554 |
- SET_QPEL_FUNCS(avg_qpel, 1, 8, mmxext, ); |
|
| 555 |
- |
|
| 556 |
- SET_QPEL_FUNCS(put_qpel, 0, 16, mmxext, ); |
|
| 557 |
- SET_QPEL_FUNCS(put_qpel, 1, 8, mmxext, ); |
|
| 558 |
- SET_QPEL_FUNCS(put_no_rnd_qpel, 0, 16, mmxext, ); |
|
| 559 |
- SET_QPEL_FUNCS(put_no_rnd_qpel, 1, 8, mmxext, ); |
|
| 560 |
- |
|
| 561 | 553 |
c->scalarproduct_int16 = ff_scalarproduct_int16_mmxext; |
| 562 | 554 |
#endif /* HAVE_MMXEXT_EXTERNAL */ |
| 563 | 555 |
} |
| 564 | 556 |
deleted file mode 100644 |
| ... | ... |
@@ -1,558 +0,0 @@ |
| 1 |
-;****************************************************************************** |
|
| 2 |
-;* mpeg4 qpel |
|
| 3 |
-;* Copyright (c) 2008 Loren Merritt |
|
| 4 |
-;* |
|
| 5 |
-;* This file is part of Libav. |
|
| 6 |
-;* |
|
| 7 |
-;* Libav 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 |
-;* Libav 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 Libav; if not, write to the Free Software |
|
| 19 |
-;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
| 20 |
-;****************************************************************************** |
|
| 21 |
- |
|
| 22 |
-%include "libavutil/x86/x86util.asm" |
|
| 23 |
- |
|
| 24 |
-SECTION_RODATA |
|
| 25 |
-cextern pb_1 |
|
| 26 |
-cextern pw_3 |
|
| 27 |
-cextern pw_15 |
|
| 28 |
-cextern pw_16 |
|
| 29 |
-cextern pw_20 |
|
| 30 |
- |
|
| 31 |
- |
|
| 32 |
-SECTION_TEXT |
|
| 33 |
- |
|
| 34 |
-; void ff_put_no_rnd_pixels8_l2(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h) |
|
| 35 |
-%macro PUT_NO_RND_PIXELS8_L2 0 |
|
| 36 |
-cglobal put_no_rnd_pixels8_l2, 6,6 |
|
| 37 |
- movsxdifnidn r4, r4d |
|
| 38 |
- movsxdifnidn r3, r3d |
|
| 39 |
- pcmpeqb m6, m6 |
|
| 40 |
- test r5d, 1 |
|
| 41 |
- je .loop |
|
| 42 |
- mova m0, [r1] |
|
| 43 |
- mova m1, [r2] |
|
| 44 |
- add r1, r4 |
|
| 45 |
- add r2, 8 |
|
| 46 |
- pxor m0, m6 |
|
| 47 |
- pxor m1, m6 |
|
| 48 |
- PAVGB m0, m1 |
|
| 49 |
- pxor m0, m6 |
|
| 50 |
- mova [r0], m0 |
|
| 51 |
- add r0, r3 |
|
| 52 |
- dec r5d |
|
| 53 |
-.loop: |
|
| 54 |
- mova m0, [r1] |
|
| 55 |
- add r1, r4 |
|
| 56 |
- mova m1, [r1] |
|
| 57 |
- add r1, r4 |
|
| 58 |
- mova m2, [r2] |
|
| 59 |
- mova m3, [r2+8] |
|
| 60 |
- pxor m0, m6 |
|
| 61 |
- pxor m1, m6 |
|
| 62 |
- pxor m2, m6 |
|
| 63 |
- pxor m3, m6 |
|
| 64 |
- PAVGB m0, m2 |
|
| 65 |
- PAVGB m1, m3 |
|
| 66 |
- pxor m0, m6 |
|
| 67 |
- pxor m1, m6 |
|
| 68 |
- mova [r0], m0 |
|
| 69 |
- add r0, r3 |
|
| 70 |
- mova [r0], m1 |
|
| 71 |
- add r0, r3 |
|
| 72 |
- mova m0, [r1] |
|
| 73 |
- add r1, r4 |
|
| 74 |
- mova m1, [r1] |
|
| 75 |
- add r1, r4 |
|
| 76 |
- mova m2, [r2+16] |
|
| 77 |
- mova m3, [r2+24] |
|
| 78 |
- pxor m0, m6 |
|
| 79 |
- pxor m1, m6 |
|
| 80 |
- pxor m2, m6 |
|
| 81 |
- pxor m3, m6 |
|
| 82 |
- PAVGB m0, m2 |
|
| 83 |
- PAVGB m1, m3 |
|
| 84 |
- pxor m0, m6 |
|
| 85 |
- pxor m1, m6 |
|
| 86 |
- mova [r0], m0 |
|
| 87 |
- add r0, r3 |
|
| 88 |
- mova [r0], m1 |
|
| 89 |
- add r0, r3 |
|
| 90 |
- add r2, 32 |
|
| 91 |
- sub r5d, 4 |
|
| 92 |
- jne .loop |
|
| 93 |
- REP_RET |
|
| 94 |
-%endmacro |
|
| 95 |
- |
|
| 96 |
-INIT_MMX mmxext |
|
| 97 |
-PUT_NO_RND_PIXELS8_L2 |
|
| 98 |
- |
|
| 99 |
- |
|
| 100 |
-; void ff_put_no_rnd_pixels16_l2(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h) |
|
| 101 |
-%macro PUT_NO_RND_PIXELS16_l2 0 |
|
| 102 |
-cglobal put_no_rnd_pixels16_l2, 6,6 |
|
| 103 |
- movsxdifnidn r3, r3d |
|
| 104 |
- movsxdifnidn r4, r4d |
|
| 105 |
- pcmpeqb m6, m6 |
|
| 106 |
- test r5d, 1 |
|
| 107 |
- je .loop |
|
| 108 |
- mova m0, [r1] |
|
| 109 |
- mova m1, [r1+8] |
|
| 110 |
- mova m2, [r2] |
|
| 111 |
- mova m3, [r2+8] |
|
| 112 |
- pxor m0, m6 |
|
| 113 |
- pxor m1, m6 |
|
| 114 |
- pxor m2, m6 |
|
| 115 |
- pxor m3, m6 |
|
| 116 |
- PAVGB m0, m2 |
|
| 117 |
- PAVGB m1, m3 |
|
| 118 |
- pxor m0, m6 |
|
| 119 |
- pxor m1, m6 |
|
| 120 |
- add r1, r4 |
|
| 121 |
- add r2, 16 |
|
| 122 |
- mova [r0], m0 |
|
| 123 |
- mova [r0+8], m1 |
|
| 124 |
- add r0, r3 |
|
| 125 |
- dec r5d |
|
| 126 |
-.loop: |
|
| 127 |
- mova m0, [r1] |
|
| 128 |
- mova m1, [r1+8] |
|
| 129 |
- add r1, r4 |
|
| 130 |
- mova m2, [r2] |
|
| 131 |
- mova m3, [r2+8] |
|
| 132 |
- pxor m0, m6 |
|
| 133 |
- pxor m1, m6 |
|
| 134 |
- pxor m2, m6 |
|
| 135 |
- pxor m3, m6 |
|
| 136 |
- PAVGB m0, m2 |
|
| 137 |
- PAVGB m1, m3 |
|
| 138 |
- pxor m0, m6 |
|
| 139 |
- pxor m1, m6 |
|
| 140 |
- mova [r0], m0 |
|
| 141 |
- mova [r0+8], m1 |
|
| 142 |
- add r0, r3 |
|
| 143 |
- mova m0, [r1] |
|
| 144 |
- mova m1, [r1+8] |
|
| 145 |
- add r1, r4 |
|
| 146 |
- mova m2, [r2+16] |
|
| 147 |
- mova m3, [r2+24] |
|
| 148 |
- pxor m0, m6 |
|
| 149 |
- pxor m1, m6 |
|
| 150 |
- pxor m2, m6 |
|
| 151 |
- pxor m3, m6 |
|
| 152 |
- PAVGB m0, m2 |
|
| 153 |
- PAVGB m1, m3 |
|
| 154 |
- pxor m0, m6 |
|
| 155 |
- pxor m1, m6 |
|
| 156 |
- mova [r0], m0 |
|
| 157 |
- mova [r0+8], m1 |
|
| 158 |
- add r0, r3 |
|
| 159 |
- add r2, 32 |
|
| 160 |
- sub r5d, 2 |
|
| 161 |
- jne .loop |
|
| 162 |
- REP_RET |
|
| 163 |
-%endmacro |
|
| 164 |
- |
|
| 165 |
-INIT_MMX mmxext |
|
| 166 |
-PUT_NO_RND_PIXELS16_l2 |
|
| 167 |
-INIT_MMX 3dnow |
|
| 168 |
-PUT_NO_RND_PIXELS16_l2 |
|
| 169 |
- |
|
| 170 |
-%macro MPEG4_QPEL16_H_LOWPASS 1 |
|
| 171 |
-cglobal %1_mpeg4_qpel16_h_lowpass, 5, 5, 0, 16 |
|
| 172 |
- movsxdifnidn r2, r2d |
|
| 173 |
- movsxdifnidn r3, r3d |
|
| 174 |
- pxor m7, m7 |
|
| 175 |
-.loop: |
|
| 176 |
- mova m0, [r1] |
|
| 177 |
- mova m1, m0 |
|
| 178 |
- mova m2, m0 |
|
| 179 |
- punpcklbw m0, m7 |
|
| 180 |
- punpckhbw m1, m7 |
|
| 181 |
- pshufw m5, m0, 0x90 |
|
| 182 |
- pshufw m6, m0, 0x41 |
|
| 183 |
- mova m3, m2 |
|
| 184 |
- mova m4, m2 |
|
| 185 |
- psllq m2, 8 |
|
| 186 |
- psllq m3, 16 |
|
| 187 |
- psllq m4, 24 |
|
| 188 |
- punpckhbw m2, m7 |
|
| 189 |
- punpckhbw m3, m7 |
|
| 190 |
- punpckhbw m4, m7 |
|
| 191 |
- paddw m5, m3 |
|
| 192 |
- paddw m6, m2 |
|
| 193 |
- paddw m5, m5 |
|
| 194 |
- psubw m6, m5 |
|
| 195 |
- pshufw m5, m0, 6 |
|
| 196 |
- pmullw m6, [pw_3] |
|
| 197 |
- paddw m0, m4 |
|
| 198 |
- paddw m5, m1 |
|
| 199 |
- pmullw m0, [pw_20] |
|
| 200 |
- psubw m0, m5 |
|
| 201 |
- paddw m6, [PW_ROUND] |
|
| 202 |
- paddw m0, m6 |
|
| 203 |
- psraw m0, 5 |
|
| 204 |
- mova [rsp+8], m0 |
|
| 205 |
- mova m0, [r1+5] |
|
| 206 |
- mova m5, m0 |
|
| 207 |
- mova m6, m0 |
|
| 208 |
- psrlq m0, 8 |
|
| 209 |
- psrlq m5, 16 |
|
| 210 |
- punpcklbw m0, m7 |
|
| 211 |
- punpcklbw m5, m7 |
|
| 212 |
- paddw m2, m0 |
|
| 213 |
- paddw m3, m5 |
|
| 214 |
- paddw m2, m2 |
|
| 215 |
- psubw m3, m2 |
|
| 216 |
- mova m2, m6 |
|
| 217 |
- psrlq m6, 24 |
|
| 218 |
- punpcklbw m2, m7 |
|
| 219 |
- punpcklbw m6, m7 |
|
| 220 |
- pmullw m3, [pw_3] |
|
| 221 |
- paddw m1, m2 |
|
| 222 |
- paddw m4, m6 |
|
| 223 |
- pmullw m1, [pw_20] |
|
| 224 |
- psubw m3, m4 |
|
| 225 |
- paddw m1, [PW_ROUND] |
|
| 226 |
- paddw m3, m1 |
|
| 227 |
- psraw m3, 5 |
|
| 228 |
- mova m1, [rsp+8] |
|
| 229 |
- packuswb m1, m3 |
|
| 230 |
- OP_MOV [r0], m1, m4 |
|
| 231 |
- mova m1, [r1+9] |
|
| 232 |
- mova m4, m1 |
|
| 233 |
- mova m3, m1 |
|
| 234 |
- psrlq m1, 8 |
|
| 235 |
- psrlq m4, 16 |
|
| 236 |
- punpcklbw m1, m7 |
|
| 237 |
- punpcklbw m4, m7 |
|
| 238 |
- paddw m5, m1 |
|
| 239 |
- paddw m0, m4 |
|
| 240 |
- paddw m5, m5 |
|
| 241 |
- psubw m0, m5 |
|
| 242 |
- mova m5, m3 |
|
| 243 |
- psrlq m3, 24 |
|
| 244 |
- pmullw m0, [pw_3] |
|
| 245 |
- punpcklbw m3, m7 |
|
| 246 |
- paddw m2, m3 |
|
| 247 |
- psubw m0, m2 |
|
| 248 |
- mova m2, m5 |
|
| 249 |
- punpcklbw m2, m7 |
|
| 250 |
- punpckhbw m5, m7 |
|
| 251 |
- paddw m6, m2 |
|
| 252 |
- pmullw m6, [pw_20] |
|
| 253 |
- paddw m0, [PW_ROUND] |
|
| 254 |
- paddw m0, m6 |
|
| 255 |
- psraw m0, 5 |
|
| 256 |
- paddw m3, m5 |
|
| 257 |
- pshufw m6, m5, 0xf9 |
|
| 258 |
- paddw m6, m4 |
|
| 259 |
- pshufw m4, m5, 0xbe |
|
| 260 |
- pshufw m5, m5, 0x6f |
|
| 261 |
- paddw m4, m1 |
|
| 262 |
- paddw m5, m2 |
|
| 263 |
- paddw m6, m6 |
|
| 264 |
- psubw m4, m6 |
|
| 265 |
- pmullw m3, [pw_20] |
|
| 266 |
- pmullw m4, [pw_3] |
|
| 267 |
- psubw m3, m5 |
|
| 268 |
- paddw m4, [PW_ROUND] |
|
| 269 |
- paddw m4, m3 |
|
| 270 |
- psraw m4, 5 |
|
| 271 |
- packuswb m0, m4 |
|
| 272 |
- OP_MOV [r0+8], m0, m4 |
|
| 273 |
- add r1, r3 |
|
| 274 |
- add r0, r2 |
|
| 275 |
- dec r4d |
|
| 276 |
- jne .loop |
|
| 277 |
- REP_RET |
|
| 278 |
-%endmacro |
|
| 279 |
- |
|
| 280 |
-%macro PUT_OP 2-3 |
|
| 281 |
- mova %1, %2 |
|
| 282 |
-%endmacro |
|
| 283 |
- |
|
| 284 |
-%macro AVG_OP 2-3 |
|
| 285 |
- mova %3, %1 |
|
| 286 |
- pavgb %2, %3 |
|
| 287 |
- mova %1, %2 |
|
| 288 |
-%endmacro |
|
| 289 |
- |
|
| 290 |
-INIT_MMX mmxext |
|
| 291 |
-%define PW_ROUND pw_16 |
|
| 292 |
-%define OP_MOV PUT_OP |
|
| 293 |
-MPEG4_QPEL16_H_LOWPASS put |
|
| 294 |
-%define PW_ROUND pw_16 |
|
| 295 |
-%define OP_MOV AVG_OP |
|
| 296 |
-MPEG4_QPEL16_H_LOWPASS avg |
|
| 297 |
-%define PW_ROUND pw_15 |
|
| 298 |
-%define OP_MOV PUT_OP |
|
| 299 |
-MPEG4_QPEL16_H_LOWPASS put_no_rnd |
|
| 300 |
- |
|
| 301 |
- |
|
| 302 |
- |
|
| 303 |
-%macro MPEG4_QPEL8_H_LOWPASS 1 |
|
| 304 |
-cglobal %1_mpeg4_qpel8_h_lowpass, 5, 5, 0, 8 |
|
| 305 |
- movsxdifnidn r2, r2d |
|
| 306 |
- movsxdifnidn r3, r3d |
|
| 307 |
- pxor m7, m7 |
|
| 308 |
-.loop: |
|
| 309 |
- mova m0, [r1] |
|
| 310 |
- mova m1, m0 |
|
| 311 |
- mova m2, m0 |
|
| 312 |
- punpcklbw m0, m7 |
|
| 313 |
- punpckhbw m1, m7 |
|
| 314 |
- pshufw m5, m0, 0x90 |
|
| 315 |
- pshufw m6, m0, 0x41 |
|
| 316 |
- mova m3, m2 |
|
| 317 |
- mova m4, m2 |
|
| 318 |
- psllq m2, 8 |
|
| 319 |
- psllq m3, 16 |
|
| 320 |
- psllq m4, 24 |
|
| 321 |
- punpckhbw m2, m7 |
|
| 322 |
- punpckhbw m3, m7 |
|
| 323 |
- punpckhbw m4, m7 |
|
| 324 |
- paddw m5, m3 |
|
| 325 |
- paddw m6, m2 |
|
| 326 |
- paddw m5, m5 |
|
| 327 |
- psubw m6, m5 |
|
| 328 |
- pshufw m5, m0, 0x6 |
|
| 329 |
- pmullw m6, [pw_3] |
|
| 330 |
- paddw m0, m4 |
|
| 331 |
- paddw m5, m1 |
|
| 332 |
- pmullw m0, [pw_20] |
|
| 333 |
- psubw m0, m5 |
|
| 334 |
- paddw m6, [PW_ROUND] |
|
| 335 |
- paddw m0, m6 |
|
| 336 |
- psraw m0, 5 |
|
| 337 |
- movh m5, [r1+5] |
|
| 338 |
- punpcklbw m5, m7 |
|
| 339 |
- pshufw m6, m5, 0xf9 |
|
| 340 |
- paddw m1, m5 |
|
| 341 |
- paddw m2, m6 |
|
| 342 |
- pshufw m6, m5, 0xbe |
|
| 343 |
- pshufw m5, m5, 0x6f |
|
| 344 |
- paddw m3, m6 |
|
| 345 |
- paddw m4, m5 |
|
| 346 |
- paddw m2, m2 |
|
| 347 |
- psubw m3, m2 |
|
| 348 |
- pmullw m1, [pw_20] |
|
| 349 |
- pmullw m3, [pw_3] |
|
| 350 |
- psubw m3, m4 |
|
| 351 |
- paddw m1, [PW_ROUND] |
|
| 352 |
- paddw m3, m1 |
|
| 353 |
- psraw m3, 5 |
|
| 354 |
- packuswb m0, m3 |
|
| 355 |
- OP_MOV [r0], m0, m4 |
|
| 356 |
- add r1, r3 |
|
| 357 |
- add r0, r2 |
|
| 358 |
- dec r4d |
|
| 359 |
- jne .loop |
|
| 360 |
- REP_RET |
|
| 361 |
-%endmacro |
|
| 362 |
- |
|
| 363 |
-INIT_MMX mmxext |
|
| 364 |
-%define PW_ROUND pw_16 |
|
| 365 |
-%define OP_MOV PUT_OP |
|
| 366 |
-MPEG4_QPEL8_H_LOWPASS put |
|
| 367 |
-%define PW_ROUND pw_16 |
|
| 368 |
-%define OP_MOV AVG_OP |
|
| 369 |
-MPEG4_QPEL8_H_LOWPASS avg |
|
| 370 |
-%define PW_ROUND pw_15 |
|
| 371 |
-%define OP_MOV PUT_OP |
|
| 372 |
-MPEG4_QPEL8_H_LOWPASS put_no_rnd |
|
| 373 |
- |
|
| 374 |
- |
|
| 375 |
- |
|
| 376 |
-%macro QPEL_V_LOW 5 |
|
| 377 |
- paddw m0, m1 |
|
| 378 |
- mova m4, [pw_20] |
|
| 379 |
- pmullw m4, m0 |
|
| 380 |
- mova m0, %4 |
|
| 381 |
- mova m5, %1 |
|
| 382 |
- paddw m5, m0 |
|
| 383 |
- psubw m4, m5 |
|
| 384 |
- mova m5, %2 |
|
| 385 |
- mova m6, %3 |
|
| 386 |
- paddw m5, m3 |
|
| 387 |
- paddw m6, m2 |
|
| 388 |
- paddw m6, m6 |
|
| 389 |
- psubw m5, m6 |
|
| 390 |
- pmullw m5, [pw_3] |
|
| 391 |
- paddw m4, [PW_ROUND] |
|
| 392 |
- paddw m5, m4 |
|
| 393 |
- psraw m5, 5 |
|
| 394 |
- packuswb m5, m5 |
|
| 395 |
- OP_MOV %5, m5, m7 |
|
| 396 |
- SWAP 0,1,2,3 |
|
| 397 |
-%endmacro |
|
| 398 |
- |
|
| 399 |
-%macro MPEG4_QPEL16_V_LOWPASS 1 |
|
| 400 |
-cglobal %1_mpeg4_qpel16_v_lowpass, 4, 6, 0, 544 |
|
| 401 |
- movsxdifnidn r2, r2d |
|
| 402 |
- movsxdifnidn r3, r3d |
|
| 403 |
- |
|
| 404 |
- mov r4d, 17 |
|
| 405 |
- mov r5, rsp |
|
| 406 |
- pxor m7, m7 |
|
| 407 |
-.looph: |
|
| 408 |
- mova m0, [r1] |
|
| 409 |
- mova m1, [r1] |
|
| 410 |
- mova m2, [r1+8] |
|
| 411 |
- mova m3, [r1+8] |
|
| 412 |
- punpcklbw m0, m7 |
|
| 413 |
- punpckhbw m1, m7 |
|
| 414 |
- punpcklbw m2, m7 |
|
| 415 |
- punpckhbw m3, m7 |
|
| 416 |
- mova [r5], m0 |
|
| 417 |
- mova [r5+0x88], m1 |
|
| 418 |
- mova [r5+0x110], m2 |
|
| 419 |
- mova [r5+0x198], m3 |
|
| 420 |
- add r5, 8 |
|
| 421 |
- add r1, r3 |
|
| 422 |
- dec r4d |
|
| 423 |
- jne .looph |
|
| 424 |
- |
|
| 425 |
- |
|
| 426 |
- ; NOTE: r1 CHANGES VALUES: r1 -> 4 - 14*dstStride |
|
| 427 |
- mov r4d, 4 |
|
| 428 |
- mov r1, 4 |
|
| 429 |
- neg r2 |
|
| 430 |
- lea r1, [r1+r2*8] |
|
| 431 |
- lea r1, [r1+r2*4] |
|
| 432 |
- lea r1, [r1+r2*2] |
|
| 433 |
- neg r2 |
|
| 434 |
- mov r5, rsp |
|
| 435 |
-.loopv: |
|
| 436 |
- pxor m7, m7 |
|
| 437 |
- mova m0, [r5+ 0x0] |
|
| 438 |
- mova m1, [r5+ 0x8] |
|
| 439 |
- mova m2, [r5+0x10] |
|
| 440 |
- mova m3, [r5+0x18] |
|
| 441 |
- QPEL_V_LOW [r5+0x10], [r5+ 0x8], [r5+ 0x0], [r5+0x20], [r0] |
|
| 442 |
- QPEL_V_LOW [r5+ 0x8], [r5+ 0x0], [r5+ 0x0], [r5+0x28], [r0+r2] |
|
| 443 |
- lea r0, [r0+r2*2] |
|
| 444 |
- QPEL_V_LOW [r5+ 0x0], [r5+ 0x0], [r5+ 0x8], [r5+0x30], [r0] |
|
| 445 |
- QPEL_V_LOW [r5+ 0x0], [r5+ 0x8], [r5+0x10], [r5+0x38], [r0+r2] |
|
| 446 |
- lea r0, [r0+r2*2] |
|
| 447 |
- QPEL_V_LOW [r5+ 0x8], [r5+0x10], [r5+0x18], [r5+0x40], [r0] |
|
| 448 |
- QPEL_V_LOW [r5+0x10], [r5+0x18], [r5+0x20], [r5+0x48], [r0+r2] |
|
| 449 |
- lea r0, [r0+r2*2] |
|
| 450 |
- QPEL_V_LOW [r5+0x18], [r5+0x20], [r5+0x28], [r5+0x50], [r0] |
|
| 451 |
- QPEL_V_LOW [r5+0x20], [r5+0x28], [r5+0x30], [r5+0x58], [r0+r2] |
|
| 452 |
- lea r0, [r0+r2*2] |
|
| 453 |
- QPEL_V_LOW [r5+0x28], [r5+0x30], [r5+0x38], [r5+0x60], [r0] |
|
| 454 |
- QPEL_V_LOW [r5+0x30], [r5+0x38], [r5+0x40], [r5+0x68], [r0+r2] |
|
| 455 |
- lea r0, [r0+r2*2] |
|
| 456 |
- QPEL_V_LOW [r5+0x38], [r5+0x40], [r5+0x48], [r5+0x70], [r0] |
|
| 457 |
- QPEL_V_LOW [r5+0x40], [r5+0x48], [r5+0x50], [r5+0x78], [r0+r2] |
|
| 458 |
- lea r0, [r0+r2*2] |
|
| 459 |
- QPEL_V_LOW [r5+0x48], [r5+0x50], [r5+0x58], [r5+0x80], [r0] |
|
| 460 |
- QPEL_V_LOW [r5+0x50], [r5+0x58], [r5+0x60], [r5+0x80], [r0+r2] |
|
| 461 |
- lea r0, [r0+r2*2] |
|
| 462 |
- QPEL_V_LOW [r5+0x58], [r5+0x60], [r5+0x68], [r5+0x78], [r0] |
|
| 463 |
- QPEL_V_LOW [r5+0x60], [r5+0x68], [r5+0x70], [r5+0x70], [r0+r2] |
|
| 464 |
- |
|
| 465 |
- add r5, 0x88 |
|
| 466 |
- add r0, r1 |
|
| 467 |
- dec r4d |
|
| 468 |
- jne .loopv |
|
| 469 |
- REP_RET |
|
| 470 |
-%endmacro |
|
| 471 |
- |
|
| 472 |
-%macro PUT_OPH 2-3 |
|
| 473 |
- movh %1, %2 |
|
| 474 |
-%endmacro |
|
| 475 |
- |
|
| 476 |
-%macro AVG_OPH 2-3 |
|
| 477 |
- movh %3, %1 |
|
| 478 |
- pavgb %2, %3 |
|
| 479 |
- movh %1, %2 |
|
| 480 |
-%endmacro |
|
| 481 |
- |
|
| 482 |
-INIT_MMX mmxext |
|
| 483 |
-%define PW_ROUND pw_16 |
|
| 484 |
-%define OP_MOV PUT_OPH |
|
| 485 |
-MPEG4_QPEL16_V_LOWPASS put |
|
| 486 |
-%define PW_ROUND pw_16 |
|
| 487 |
-%define OP_MOV AVG_OPH |
|
| 488 |
-MPEG4_QPEL16_V_LOWPASS avg |
|
| 489 |
-%define PW_ROUND pw_15 |
|
| 490 |
-%define OP_MOV PUT_OPH |
|
| 491 |
-MPEG4_QPEL16_V_LOWPASS put_no_rnd |
|
| 492 |
- |
|
| 493 |
- |
|
| 494 |
- |
|
| 495 |
-%macro MPEG4_QPEL8_V_LOWPASS 1 |
|
| 496 |
-cglobal %1_mpeg4_qpel8_v_lowpass, 4, 6, 0, 288 |
|
| 497 |
- movsxdifnidn r2, r2d |
|
| 498 |
- movsxdifnidn r3, r3d |
|
| 499 |
- |
|
| 500 |
- mov r4d, 9 |
|
| 501 |
- mov r5, rsp |
|
| 502 |
- pxor m7, m7 |
|
| 503 |
-.looph: |
|
| 504 |
- mova m0, [r1] |
|
| 505 |
- mova m1, [r1] |
|
| 506 |
- punpcklbw m0, m7 |
|
| 507 |
- punpckhbw m1, m7 |
|
| 508 |
- mova [r5], m0 |
|
| 509 |
- mova [r5+0x48], m1 |
|
| 510 |
- add r5, 8 |
|
| 511 |
- add r1, r3 |
|
| 512 |
- dec r4d |
|
| 513 |
- jne .looph |
|
| 514 |
- |
|
| 515 |
- |
|
| 516 |
- ; NOTE: r1 CHANGES VALUES: r1 -> 4 - 6*dstStride |
|
| 517 |
- mov r4d, 2 |
|
| 518 |
- mov r1, 4 |
|
| 519 |
- neg r2 |
|
| 520 |
- lea r1, [r1+r2*4] |
|
| 521 |
- lea r1, [r1+r2*2] |
|
| 522 |
- neg r2 |
|
| 523 |
- mov r5, rsp |
|
| 524 |
-.loopv: |
|
| 525 |
- pxor m7, m7 |
|
| 526 |
- mova m0, [r5+ 0x0] |
|
| 527 |
- mova m1, [r5+ 0x8] |
|
| 528 |
- mova m2, [r5+0x10] |
|
| 529 |
- mova m3, [r5+0x18] |
|
| 530 |
- QPEL_V_LOW [r5+0x10], [r5+ 0x8], [r5+ 0x0], [r5+0x20], [r0] |
|
| 531 |
- QPEL_V_LOW [r5+ 0x8], [r5+ 0x0], [r5+ 0x0], [r5+0x28], [r0+r2] |
|
| 532 |
- lea r0, [r0+r2*2] |
|
| 533 |
- QPEL_V_LOW [r5+ 0x0], [r5+ 0x0], [r5+ 0x8], [r5+0x30], [r0] |
|
| 534 |
- QPEL_V_LOW [r5+ 0x0], [r5+ 0x8], [r5+0x10], [r5+0x38], [r0+r2] |
|
| 535 |
- lea r0, [r0+r2*2] |
|
| 536 |
- QPEL_V_LOW [r5+ 0x8], [r5+0x10], [r5+0x18], [r5+0x40], [r0] |
|
| 537 |
- QPEL_V_LOW [r5+0x10], [r5+0x18], [r5+0x20], [r5+0x40], [r0+r2] |
|
| 538 |
- lea r0, [r0+r2*2] |
|
| 539 |
- QPEL_V_LOW [r5+0x18], [r5+0x20], [r5+0x28], [r5+0x38], [r0] |
|
| 540 |
- QPEL_V_LOW [r5+0x20], [r5+0x28], [r5+0x30], [r5+0x30], [r0+r2] |
|
| 541 |
- |
|
| 542 |
- add r5, 0x48 |
|
| 543 |
- add r0, r1 |
|
| 544 |
- dec r4d |
|
| 545 |
- jne .loopv |
|
| 546 |
- REP_RET |
|
| 547 |
-%endmacro |
|
| 548 |
- |
|
| 549 |
-INIT_MMX mmxext |
|
| 550 |
-%define PW_ROUND pw_16 |
|
| 551 |
-%define OP_MOV PUT_OPH |
|
| 552 |
-MPEG4_QPEL8_V_LOWPASS put |
|
| 553 |
-%define PW_ROUND pw_16 |
|
| 554 |
-%define OP_MOV AVG_OPH |
|
| 555 |
-MPEG4_QPEL8_V_LOWPASS avg |
|
| 556 |
-%define PW_ROUND pw_15 |
|
| 557 |
-%define OP_MOV PUT_OPH |
|
| 558 |
-MPEG4_QPEL8_V_LOWPASS put_no_rnd |
| 559 | 1 |
new file mode 100644 |
| ... | ... |
@@ -0,0 +1,559 @@ |
| 0 |
+;****************************************************************************** |
|
| 1 |
+;* quarterpel DSP functions |
|
| 2 |
+;* |
|
| 3 |
+;* Copyright (c) 2008 Loren Merritt |
|
| 4 |
+;* |
|
| 5 |
+;* This file is part of Libav. |
|
| 6 |
+;* |
|
| 7 |
+;* Libav 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 |
+;* Libav 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 Libav; if not, write to the Free Software |
|
| 19 |
+;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
| 20 |
+;****************************************************************************** |
|
| 21 |
+ |
|
| 22 |
+%include "libavutil/x86/x86util.asm" |
|
| 23 |
+ |
|
| 24 |
+SECTION_RODATA |
|
| 25 |
+cextern pb_1 |
|
| 26 |
+cextern pw_3 |
|
| 27 |
+cextern pw_15 |
|
| 28 |
+cextern pw_16 |
|
| 29 |
+cextern pw_20 |
|
| 30 |
+ |
|
| 31 |
+ |
|
| 32 |
+SECTION_TEXT |
|
| 33 |
+ |
|
| 34 |
+; void ff_put_no_rnd_pixels8_l2(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h) |
|
| 35 |
+%macro PUT_NO_RND_PIXELS8_L2 0 |
|
| 36 |
+cglobal put_no_rnd_pixels8_l2, 6,6 |
|
| 37 |
+ movsxdifnidn r4, r4d |
|
| 38 |
+ movsxdifnidn r3, r3d |
|
| 39 |
+ pcmpeqb m6, m6 |
|
| 40 |
+ test r5d, 1 |
|
| 41 |
+ je .loop |
|
| 42 |
+ mova m0, [r1] |
|
| 43 |
+ mova m1, [r2] |
|
| 44 |
+ add r1, r4 |
|
| 45 |
+ add r2, 8 |
|
| 46 |
+ pxor m0, m6 |
|
| 47 |
+ pxor m1, m6 |
|
| 48 |
+ PAVGB m0, m1 |
|
| 49 |
+ pxor m0, m6 |
|
| 50 |
+ mova [r0], m0 |
|
| 51 |
+ add r0, r3 |
|
| 52 |
+ dec r5d |
|
| 53 |
+.loop: |
|
| 54 |
+ mova m0, [r1] |
|
| 55 |
+ add r1, r4 |
|
| 56 |
+ mova m1, [r1] |
|
| 57 |
+ add r1, r4 |
|
| 58 |
+ mova m2, [r2] |
|
| 59 |
+ mova m3, [r2+8] |
|
| 60 |
+ pxor m0, m6 |
|
| 61 |
+ pxor m1, m6 |
|
| 62 |
+ pxor m2, m6 |
|
| 63 |
+ pxor m3, m6 |
|
| 64 |
+ PAVGB m0, m2 |
|
| 65 |
+ PAVGB m1, m3 |
|
| 66 |
+ pxor m0, m6 |
|
| 67 |
+ pxor m1, m6 |
|
| 68 |
+ mova [r0], m0 |
|
| 69 |
+ add r0, r3 |
|
| 70 |
+ mova [r0], m1 |
|
| 71 |
+ add r0, r3 |
|
| 72 |
+ mova m0, [r1] |
|
| 73 |
+ add r1, r4 |
|
| 74 |
+ mova m1, [r1] |
|
| 75 |
+ add r1, r4 |
|
| 76 |
+ mova m2, [r2+16] |
|
| 77 |
+ mova m3, [r2+24] |
|
| 78 |
+ pxor m0, m6 |
|
| 79 |
+ pxor m1, m6 |
|
| 80 |
+ pxor m2, m6 |
|
| 81 |
+ pxor m3, m6 |
|
| 82 |
+ PAVGB m0, m2 |
|
| 83 |
+ PAVGB m1, m3 |
|
| 84 |
+ pxor m0, m6 |
|
| 85 |
+ pxor m1, m6 |
|
| 86 |
+ mova [r0], m0 |
|
| 87 |
+ add r0, r3 |
|
| 88 |
+ mova [r0], m1 |
|
| 89 |
+ add r0, r3 |
|
| 90 |
+ add r2, 32 |
|
| 91 |
+ sub r5d, 4 |
|
| 92 |
+ jne .loop |
|
| 93 |
+ REP_RET |
|
| 94 |
+%endmacro |
|
| 95 |
+ |
|
| 96 |
+INIT_MMX mmxext |
|
| 97 |
+PUT_NO_RND_PIXELS8_L2 |
|
| 98 |
+ |
|
| 99 |
+ |
|
| 100 |
+; void ff_put_no_rnd_pixels16_l2(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h) |
|
| 101 |
+%macro PUT_NO_RND_PIXELS16_l2 0 |
|
| 102 |
+cglobal put_no_rnd_pixels16_l2, 6,6 |
|
| 103 |
+ movsxdifnidn r3, r3d |
|
| 104 |
+ movsxdifnidn r4, r4d |
|
| 105 |
+ pcmpeqb m6, m6 |
|
| 106 |
+ test r5d, 1 |
|
| 107 |
+ je .loop |
|
| 108 |
+ mova m0, [r1] |
|
| 109 |
+ mova m1, [r1+8] |
|
| 110 |
+ mova m2, [r2] |
|
| 111 |
+ mova m3, [r2+8] |
|
| 112 |
+ pxor m0, m6 |
|
| 113 |
+ pxor m1, m6 |
|
| 114 |
+ pxor m2, m6 |
|
| 115 |
+ pxor m3, m6 |
|
| 116 |
+ PAVGB m0, m2 |
|
| 117 |
+ PAVGB m1, m3 |
|
| 118 |
+ pxor m0, m6 |
|
| 119 |
+ pxor m1, m6 |
|
| 120 |
+ add r1, r4 |
|
| 121 |
+ add r2, 16 |
|
| 122 |
+ mova [r0], m0 |
|
| 123 |
+ mova [r0+8], m1 |
|
| 124 |
+ add r0, r3 |
|
| 125 |
+ dec r5d |
|
| 126 |
+.loop: |
|
| 127 |
+ mova m0, [r1] |
|
| 128 |
+ mova m1, [r1+8] |
|
| 129 |
+ add r1, r4 |
|
| 130 |
+ mova m2, [r2] |
|
| 131 |
+ mova m3, [r2+8] |
|
| 132 |
+ pxor m0, m6 |
|
| 133 |
+ pxor m1, m6 |
|
| 134 |
+ pxor m2, m6 |
|
| 135 |
+ pxor m3, m6 |
|
| 136 |
+ PAVGB m0, m2 |
|
| 137 |
+ PAVGB m1, m3 |
|
| 138 |
+ pxor m0, m6 |
|
| 139 |
+ pxor m1, m6 |
|
| 140 |
+ mova [r0], m0 |
|
| 141 |
+ mova [r0+8], m1 |
|
| 142 |
+ add r0, r3 |
|
| 143 |
+ mova m0, [r1] |
|
| 144 |
+ mova m1, [r1+8] |
|
| 145 |
+ add r1, r4 |
|
| 146 |
+ mova m2, [r2+16] |
|
| 147 |
+ mova m3, [r2+24] |
|
| 148 |
+ pxor m0, m6 |
|
| 149 |
+ pxor m1, m6 |
|
| 150 |
+ pxor m2, m6 |
|
| 151 |
+ pxor m3, m6 |
|
| 152 |
+ PAVGB m0, m2 |
|
| 153 |
+ PAVGB m1, m3 |
|
| 154 |
+ pxor m0, m6 |
|
| 155 |
+ pxor m1, m6 |
|
| 156 |
+ mova [r0], m0 |
|
| 157 |
+ mova [r0+8], m1 |
|
| 158 |
+ add r0, r3 |
|
| 159 |
+ add r2, 32 |
|
| 160 |
+ sub r5d, 2 |
|
| 161 |
+ jne .loop |
|
| 162 |
+ REP_RET |
|
| 163 |
+%endmacro |
|
| 164 |
+ |
|
| 165 |
+INIT_MMX mmxext |
|
| 166 |
+PUT_NO_RND_PIXELS16_l2 |
|
| 167 |
+INIT_MMX 3dnow |
|
| 168 |
+PUT_NO_RND_PIXELS16_l2 |
|
| 169 |
+ |
|
| 170 |
+%macro MPEG4_QPEL16_H_LOWPASS 1 |
|
| 171 |
+cglobal %1_mpeg4_qpel16_h_lowpass, 5, 5, 0, 16 |
|
| 172 |
+ movsxdifnidn r2, r2d |
|
| 173 |
+ movsxdifnidn r3, r3d |
|
| 174 |
+ pxor m7, m7 |
|
| 175 |
+.loop: |
|
| 176 |
+ mova m0, [r1] |
|
| 177 |
+ mova m1, m0 |
|
| 178 |
+ mova m2, m0 |
|
| 179 |
+ punpcklbw m0, m7 |
|
| 180 |
+ punpckhbw m1, m7 |
|
| 181 |
+ pshufw m5, m0, 0x90 |
|
| 182 |
+ pshufw m6, m0, 0x41 |
|
| 183 |
+ mova m3, m2 |
|
| 184 |
+ mova m4, m2 |
|
| 185 |
+ psllq m2, 8 |
|
| 186 |
+ psllq m3, 16 |
|
| 187 |
+ psllq m4, 24 |
|
| 188 |
+ punpckhbw m2, m7 |
|
| 189 |
+ punpckhbw m3, m7 |
|
| 190 |
+ punpckhbw m4, m7 |
|
| 191 |
+ paddw m5, m3 |
|
| 192 |
+ paddw m6, m2 |
|
| 193 |
+ paddw m5, m5 |
|
| 194 |
+ psubw m6, m5 |
|
| 195 |
+ pshufw m5, m0, 6 |
|
| 196 |
+ pmullw m6, [pw_3] |
|
| 197 |
+ paddw m0, m4 |
|
| 198 |
+ paddw m5, m1 |
|
| 199 |
+ pmullw m0, [pw_20] |
|
| 200 |
+ psubw m0, m5 |
|
| 201 |
+ paddw m6, [PW_ROUND] |
|
| 202 |
+ paddw m0, m6 |
|
| 203 |
+ psraw m0, 5 |
|
| 204 |
+ mova [rsp+8], m0 |
|
| 205 |
+ mova m0, [r1+5] |
|
| 206 |
+ mova m5, m0 |
|
| 207 |
+ mova m6, m0 |
|
| 208 |
+ psrlq m0, 8 |
|
| 209 |
+ psrlq m5, 16 |
|
| 210 |
+ punpcklbw m0, m7 |
|
| 211 |
+ punpcklbw m5, m7 |
|
| 212 |
+ paddw m2, m0 |
|
| 213 |
+ paddw m3, m5 |
|
| 214 |
+ paddw m2, m2 |
|
| 215 |
+ psubw m3, m2 |
|
| 216 |
+ mova m2, m6 |
|
| 217 |
+ psrlq m6, 24 |
|
| 218 |
+ punpcklbw m2, m7 |
|
| 219 |
+ punpcklbw m6, m7 |
|
| 220 |
+ pmullw m3, [pw_3] |
|
| 221 |
+ paddw m1, m2 |
|
| 222 |
+ paddw m4, m6 |
|
| 223 |
+ pmullw m1, [pw_20] |
|
| 224 |
+ psubw m3, m4 |
|
| 225 |
+ paddw m1, [PW_ROUND] |
|
| 226 |
+ paddw m3, m1 |
|
| 227 |
+ psraw m3, 5 |
|
| 228 |
+ mova m1, [rsp+8] |
|
| 229 |
+ packuswb m1, m3 |
|
| 230 |
+ OP_MOV [r0], m1, m4 |
|
| 231 |
+ mova m1, [r1+9] |
|
| 232 |
+ mova m4, m1 |
|
| 233 |
+ mova m3, m1 |
|
| 234 |
+ psrlq m1, 8 |
|
| 235 |
+ psrlq m4, 16 |
|
| 236 |
+ punpcklbw m1, m7 |
|
| 237 |
+ punpcklbw m4, m7 |
|
| 238 |
+ paddw m5, m1 |
|
| 239 |
+ paddw m0, m4 |
|
| 240 |
+ paddw m5, m5 |
|
| 241 |
+ psubw m0, m5 |
|
| 242 |
+ mova m5, m3 |
|
| 243 |
+ psrlq m3, 24 |
|
| 244 |
+ pmullw m0, [pw_3] |
|
| 245 |
+ punpcklbw m3, m7 |
|
| 246 |
+ paddw m2, m3 |
|
| 247 |
+ psubw m0, m2 |
|
| 248 |
+ mova m2, m5 |
|
| 249 |
+ punpcklbw m2, m7 |
|
| 250 |
+ punpckhbw m5, m7 |
|
| 251 |
+ paddw m6, m2 |
|
| 252 |
+ pmullw m6, [pw_20] |
|
| 253 |
+ paddw m0, [PW_ROUND] |
|
| 254 |
+ paddw m0, m6 |
|
| 255 |
+ psraw m0, 5 |
|
| 256 |
+ paddw m3, m5 |
|
| 257 |
+ pshufw m6, m5, 0xf9 |
|
| 258 |
+ paddw m6, m4 |
|
| 259 |
+ pshufw m4, m5, 0xbe |
|
| 260 |
+ pshufw m5, m5, 0x6f |
|
| 261 |
+ paddw m4, m1 |
|
| 262 |
+ paddw m5, m2 |
|
| 263 |
+ paddw m6, m6 |
|
| 264 |
+ psubw m4, m6 |
|
| 265 |
+ pmullw m3, [pw_20] |
|
| 266 |
+ pmullw m4, [pw_3] |
|
| 267 |
+ psubw m3, m5 |
|
| 268 |
+ paddw m4, [PW_ROUND] |
|
| 269 |
+ paddw m4, m3 |
|
| 270 |
+ psraw m4, 5 |
|
| 271 |
+ packuswb m0, m4 |
|
| 272 |
+ OP_MOV [r0+8], m0, m4 |
|
| 273 |
+ add r1, r3 |
|
| 274 |
+ add r0, r2 |
|
| 275 |
+ dec r4d |
|
| 276 |
+ jne .loop |
|
| 277 |
+ REP_RET |
|
| 278 |
+%endmacro |
|
| 279 |
+ |
|
| 280 |
+%macro PUT_OP 2-3 |
|
| 281 |
+ mova %1, %2 |
|
| 282 |
+%endmacro |
|
| 283 |
+ |
|
| 284 |
+%macro AVG_OP 2-3 |
|
| 285 |
+ mova %3, %1 |
|
| 286 |
+ pavgb %2, %3 |
|
| 287 |
+ mova %1, %2 |
|
| 288 |
+%endmacro |
|
| 289 |
+ |
|
| 290 |
+INIT_MMX mmxext |
|
| 291 |
+%define PW_ROUND pw_16 |
|
| 292 |
+%define OP_MOV PUT_OP |
|
| 293 |
+MPEG4_QPEL16_H_LOWPASS put |
|
| 294 |
+%define PW_ROUND pw_16 |
|
| 295 |
+%define OP_MOV AVG_OP |
|
| 296 |
+MPEG4_QPEL16_H_LOWPASS avg |
|
| 297 |
+%define PW_ROUND pw_15 |
|
| 298 |
+%define OP_MOV PUT_OP |
|
| 299 |
+MPEG4_QPEL16_H_LOWPASS put_no_rnd |
|
| 300 |
+ |
|
| 301 |
+ |
|
| 302 |
+ |
|
| 303 |
+%macro MPEG4_QPEL8_H_LOWPASS 1 |
|
| 304 |
+cglobal %1_mpeg4_qpel8_h_lowpass, 5, 5, 0, 8 |
|
| 305 |
+ movsxdifnidn r2, r2d |
|
| 306 |
+ movsxdifnidn r3, r3d |
|
| 307 |
+ pxor m7, m7 |
|
| 308 |
+.loop: |
|
| 309 |
+ mova m0, [r1] |
|
| 310 |
+ mova m1, m0 |
|
| 311 |
+ mova m2, m0 |
|
| 312 |
+ punpcklbw m0, m7 |
|
| 313 |
+ punpckhbw m1, m7 |
|
| 314 |
+ pshufw m5, m0, 0x90 |
|
| 315 |
+ pshufw m6, m0, 0x41 |
|
| 316 |
+ mova m3, m2 |
|
| 317 |
+ mova m4, m2 |
|
| 318 |
+ psllq m2, 8 |
|
| 319 |
+ psllq m3, 16 |
|
| 320 |
+ psllq m4, 24 |
|
| 321 |
+ punpckhbw m2, m7 |
|
| 322 |
+ punpckhbw m3, m7 |
|
| 323 |
+ punpckhbw m4, m7 |
|
| 324 |
+ paddw m5, m3 |
|
| 325 |
+ paddw m6, m2 |
|
| 326 |
+ paddw m5, m5 |
|
| 327 |
+ psubw m6, m5 |
|
| 328 |
+ pshufw m5, m0, 0x6 |
|
| 329 |
+ pmullw m6, [pw_3] |
|
| 330 |
+ paddw m0, m4 |
|
| 331 |
+ paddw m5, m1 |
|
| 332 |
+ pmullw m0, [pw_20] |
|
| 333 |
+ psubw m0, m5 |
|
| 334 |
+ paddw m6, [PW_ROUND] |
|
| 335 |
+ paddw m0, m6 |
|
| 336 |
+ psraw m0, 5 |
|
| 337 |
+ movh m5, [r1+5] |
|
| 338 |
+ punpcklbw m5, m7 |
|
| 339 |
+ pshufw m6, m5, 0xf9 |
|
| 340 |
+ paddw m1, m5 |
|
| 341 |
+ paddw m2, m6 |
|
| 342 |
+ pshufw m6, m5, 0xbe |
|
| 343 |
+ pshufw m5, m5, 0x6f |
|
| 344 |
+ paddw m3, m6 |
|
| 345 |
+ paddw m4, m5 |
|
| 346 |
+ paddw m2, m2 |
|
| 347 |
+ psubw m3, m2 |
|
| 348 |
+ pmullw m1, [pw_20] |
|
| 349 |
+ pmullw m3, [pw_3] |
|
| 350 |
+ psubw m3, m4 |
|
| 351 |
+ paddw m1, [PW_ROUND] |
|
| 352 |
+ paddw m3, m1 |
|
| 353 |
+ psraw m3, 5 |
|
| 354 |
+ packuswb m0, m3 |
|
| 355 |
+ OP_MOV [r0], m0, m4 |
|
| 356 |
+ add r1, r3 |
|
| 357 |
+ add r0, r2 |
|
| 358 |
+ dec r4d |
|
| 359 |
+ jne .loop |
|
| 360 |
+ REP_RET |
|
| 361 |
+%endmacro |
|
| 362 |
+ |
|
| 363 |
+INIT_MMX mmxext |
|
| 364 |
+%define PW_ROUND pw_16 |
|
| 365 |
+%define OP_MOV PUT_OP |
|
| 366 |
+MPEG4_QPEL8_H_LOWPASS put |
|
| 367 |
+%define PW_ROUND pw_16 |
|
| 368 |
+%define OP_MOV AVG_OP |
|
| 369 |
+MPEG4_QPEL8_H_LOWPASS avg |
|
| 370 |
+%define PW_ROUND pw_15 |
|
| 371 |
+%define OP_MOV PUT_OP |
|
| 372 |
+MPEG4_QPEL8_H_LOWPASS put_no_rnd |
|
| 373 |
+ |
|
| 374 |
+ |
|
| 375 |
+ |
|
| 376 |
+%macro QPEL_V_LOW 5 |
|
| 377 |
+ paddw m0, m1 |
|
| 378 |
+ mova m4, [pw_20] |
|
| 379 |
+ pmullw m4, m0 |
|
| 380 |
+ mova m0, %4 |
|
| 381 |
+ mova m5, %1 |
|
| 382 |
+ paddw m5, m0 |
|
| 383 |
+ psubw m4, m5 |
|
| 384 |
+ mova m5, %2 |
|
| 385 |
+ mova m6, %3 |
|
| 386 |
+ paddw m5, m3 |
|
| 387 |
+ paddw m6, m2 |
|
| 388 |
+ paddw m6, m6 |
|
| 389 |
+ psubw m5, m6 |
|
| 390 |
+ pmullw m5, [pw_3] |
|
| 391 |
+ paddw m4, [PW_ROUND] |
|
| 392 |
+ paddw m5, m4 |
|
| 393 |
+ psraw m5, 5 |
|
| 394 |
+ packuswb m5, m5 |
|
| 395 |
+ OP_MOV %5, m5, m7 |
|
| 396 |
+ SWAP 0,1,2,3 |
|
| 397 |
+%endmacro |
|
| 398 |
+ |
|
| 399 |
+%macro MPEG4_QPEL16_V_LOWPASS 1 |
|
| 400 |
+cglobal %1_mpeg4_qpel16_v_lowpass, 4, 6, 0, 544 |
|
| 401 |
+ movsxdifnidn r2, r2d |
|
| 402 |
+ movsxdifnidn r3, r3d |
|
| 403 |
+ |
|
| 404 |
+ mov r4d, 17 |
|
| 405 |
+ mov r5, rsp |
|
| 406 |
+ pxor m7, m7 |
|
| 407 |
+.looph: |
|
| 408 |
+ mova m0, [r1] |
|
| 409 |
+ mova m1, [r1] |
|
| 410 |
+ mova m2, [r1+8] |
|
| 411 |
+ mova m3, [r1+8] |
|
| 412 |
+ punpcklbw m0, m7 |
|
| 413 |
+ punpckhbw m1, m7 |
|
| 414 |
+ punpcklbw m2, m7 |
|
| 415 |
+ punpckhbw m3, m7 |
|
| 416 |
+ mova [r5], m0 |
|
| 417 |
+ mova [r5+0x88], m1 |
|
| 418 |
+ mova [r5+0x110], m2 |
|
| 419 |
+ mova [r5+0x198], m3 |
|
| 420 |
+ add r5, 8 |
|
| 421 |
+ add r1, r3 |
|
| 422 |
+ dec r4d |
|
| 423 |
+ jne .looph |
|
| 424 |
+ |
|
| 425 |
+ |
|
| 426 |
+ ; NOTE: r1 CHANGES VALUES: r1 -> 4 - 14*dstStride |
|
| 427 |
+ mov r4d, 4 |
|
| 428 |
+ mov r1, 4 |
|
| 429 |
+ neg r2 |
|
| 430 |
+ lea r1, [r1+r2*8] |
|
| 431 |
+ lea r1, [r1+r2*4] |
|
| 432 |
+ lea r1, [r1+r2*2] |
|
| 433 |
+ neg r2 |
|
| 434 |
+ mov r5, rsp |
|
| 435 |
+.loopv: |
|
| 436 |
+ pxor m7, m7 |
|
| 437 |
+ mova m0, [r5+ 0x0] |
|
| 438 |
+ mova m1, [r5+ 0x8] |
|
| 439 |
+ mova m2, [r5+0x10] |
|
| 440 |
+ mova m3, [r5+0x18] |
|
| 441 |
+ QPEL_V_LOW [r5+0x10], [r5+ 0x8], [r5+ 0x0], [r5+0x20], [r0] |
|
| 442 |
+ QPEL_V_LOW [r5+ 0x8], [r5+ 0x0], [r5+ 0x0], [r5+0x28], [r0+r2] |
|
| 443 |
+ lea r0, [r0+r2*2] |
|
| 444 |
+ QPEL_V_LOW [r5+ 0x0], [r5+ 0x0], [r5+ 0x8], [r5+0x30], [r0] |
|
| 445 |
+ QPEL_V_LOW [r5+ 0x0], [r5+ 0x8], [r5+0x10], [r5+0x38], [r0+r2] |
|
| 446 |
+ lea r0, [r0+r2*2] |
|
| 447 |
+ QPEL_V_LOW [r5+ 0x8], [r5+0x10], [r5+0x18], [r5+0x40], [r0] |
|
| 448 |
+ QPEL_V_LOW [r5+0x10], [r5+0x18], [r5+0x20], [r5+0x48], [r0+r2] |
|
| 449 |
+ lea r0, [r0+r2*2] |
|
| 450 |
+ QPEL_V_LOW [r5+0x18], [r5+0x20], [r5+0x28], [r5+0x50], [r0] |
|
| 451 |
+ QPEL_V_LOW [r5+0x20], [r5+0x28], [r5+0x30], [r5+0x58], [r0+r2] |
|
| 452 |
+ lea r0, [r0+r2*2] |
|
| 453 |
+ QPEL_V_LOW [r5+0x28], [r5+0x30], [r5+0x38], [r5+0x60], [r0] |
|
| 454 |
+ QPEL_V_LOW [r5+0x30], [r5+0x38], [r5+0x40], [r5+0x68], [r0+r2] |
|
| 455 |
+ lea r0, [r0+r2*2] |
|
| 456 |
+ QPEL_V_LOW [r5+0x38], [r5+0x40], [r5+0x48], [r5+0x70], [r0] |
|
| 457 |
+ QPEL_V_LOW [r5+0x40], [r5+0x48], [r5+0x50], [r5+0x78], [r0+r2] |
|
| 458 |
+ lea r0, [r0+r2*2] |
|
| 459 |
+ QPEL_V_LOW [r5+0x48], [r5+0x50], [r5+0x58], [r5+0x80], [r0] |
|
| 460 |
+ QPEL_V_LOW [r5+0x50], [r5+0x58], [r5+0x60], [r5+0x80], [r0+r2] |
|
| 461 |
+ lea r0, [r0+r2*2] |
|
| 462 |
+ QPEL_V_LOW [r5+0x58], [r5+0x60], [r5+0x68], [r5+0x78], [r0] |
|
| 463 |
+ QPEL_V_LOW [r5+0x60], [r5+0x68], [r5+0x70], [r5+0x70], [r0+r2] |
|
| 464 |
+ |
|
| 465 |
+ add r5, 0x88 |
|
| 466 |
+ add r0, r1 |
|
| 467 |
+ dec r4d |
|
| 468 |
+ jne .loopv |
|
| 469 |
+ REP_RET |
|
| 470 |
+%endmacro |
|
| 471 |
+ |
|
| 472 |
+%macro PUT_OPH 2-3 |
|
| 473 |
+ movh %1, %2 |
|
| 474 |
+%endmacro |
|
| 475 |
+ |
|
| 476 |
+%macro AVG_OPH 2-3 |
|
| 477 |
+ movh %3, %1 |
|
| 478 |
+ pavgb %2, %3 |
|
| 479 |
+ movh %1, %2 |
|
| 480 |
+%endmacro |
|
| 481 |
+ |
|
| 482 |
+INIT_MMX mmxext |
|
| 483 |
+%define PW_ROUND pw_16 |
|
| 484 |
+%define OP_MOV PUT_OPH |
|
| 485 |
+MPEG4_QPEL16_V_LOWPASS put |
|
| 486 |
+%define PW_ROUND pw_16 |
|
| 487 |
+%define OP_MOV AVG_OPH |
|
| 488 |
+MPEG4_QPEL16_V_LOWPASS avg |
|
| 489 |
+%define PW_ROUND pw_15 |
|
| 490 |
+%define OP_MOV PUT_OPH |
|
| 491 |
+MPEG4_QPEL16_V_LOWPASS put_no_rnd |
|
| 492 |
+ |
|
| 493 |
+ |
|
| 494 |
+ |
|
| 495 |
+%macro MPEG4_QPEL8_V_LOWPASS 1 |
|
| 496 |
+cglobal %1_mpeg4_qpel8_v_lowpass, 4, 6, 0, 288 |
|
| 497 |
+ movsxdifnidn r2, r2d |
|
| 498 |
+ movsxdifnidn r3, r3d |
|
| 499 |
+ |
|
| 500 |
+ mov r4d, 9 |
|
| 501 |
+ mov r5, rsp |
|
| 502 |
+ pxor m7, m7 |
|
| 503 |
+.looph: |
|
| 504 |
+ mova m0, [r1] |
|
| 505 |
+ mova m1, [r1] |
|
| 506 |
+ punpcklbw m0, m7 |
|
| 507 |
+ punpckhbw m1, m7 |
|
| 508 |
+ mova [r5], m0 |
|
| 509 |
+ mova [r5+0x48], m1 |
|
| 510 |
+ add r5, 8 |
|
| 511 |
+ add r1, r3 |
|
| 512 |
+ dec r4d |
|
| 513 |
+ jne .looph |
|
| 514 |
+ |
|
| 515 |
+ |
|
| 516 |
+ ; NOTE: r1 CHANGES VALUES: r1 -> 4 - 6*dstStride |
|
| 517 |
+ mov r4d, 2 |
|
| 518 |
+ mov r1, 4 |
|
| 519 |
+ neg r2 |
|
| 520 |
+ lea r1, [r1+r2*4] |
|
| 521 |
+ lea r1, [r1+r2*2] |
|
| 522 |
+ neg r2 |
|
| 523 |
+ mov r5, rsp |
|
| 524 |
+.loopv: |
|
| 525 |
+ pxor m7, m7 |
|
| 526 |
+ mova m0, [r5+ 0x0] |
|
| 527 |
+ mova m1, [r5+ 0x8] |
|
| 528 |
+ mova m2, [r5+0x10] |
|
| 529 |
+ mova m3, [r5+0x18] |
|
| 530 |
+ QPEL_V_LOW [r5+0x10], [r5+ 0x8], [r5+ 0x0], [r5+0x20], [r0] |
|
| 531 |
+ QPEL_V_LOW [r5+ 0x8], [r5+ 0x0], [r5+ 0x0], [r5+0x28], [r0+r2] |
|
| 532 |
+ lea r0, [r0+r2*2] |
|
| 533 |
+ QPEL_V_LOW [r5+ 0x0], [r5+ 0x0], [r5+ 0x8], [r5+0x30], [r0] |
|
| 534 |
+ QPEL_V_LOW [r5+ 0x0], [r5+ 0x8], [r5+0x10], [r5+0x38], [r0+r2] |
|
| 535 |
+ lea r0, [r0+r2*2] |
|
| 536 |
+ QPEL_V_LOW [r5+ 0x8], [r5+0x10], [r5+0x18], [r5+0x40], [r0] |
|
| 537 |
+ QPEL_V_LOW [r5+0x10], [r5+0x18], [r5+0x20], [r5+0x40], [r0+r2] |
|
| 538 |
+ lea r0, [r0+r2*2] |
|
| 539 |
+ QPEL_V_LOW [r5+0x18], [r5+0x20], [r5+0x28], [r5+0x38], [r0] |
|
| 540 |
+ QPEL_V_LOW [r5+0x20], [r5+0x28], [r5+0x30], [r5+0x30], [r0+r2] |
|
| 541 |
+ |
|
| 542 |
+ add r5, 0x48 |
|
| 543 |
+ add r0, r1 |
|
| 544 |
+ dec r4d |
|
| 545 |
+ jne .loopv |
|
| 546 |
+ REP_RET |
|
| 547 |
+%endmacro |
|
| 548 |
+ |
|
| 549 |
+INIT_MMX mmxext |
|
| 550 |
+%define PW_ROUND pw_16 |
|
| 551 |
+%define OP_MOV PUT_OPH |
|
| 552 |
+MPEG4_QPEL8_V_LOWPASS put |
|
| 553 |
+%define PW_ROUND pw_16 |
|
| 554 |
+%define OP_MOV AVG_OPH |
|
| 555 |
+MPEG4_QPEL8_V_LOWPASS avg |
|
| 556 |
+%define PW_ROUND pw_15 |
|
| 557 |
+%define OP_MOV PUT_OPH |
|
| 558 |
+MPEG4_QPEL8_V_LOWPASS put_no_rnd |
| 0 | 559 |
new file mode 100644 |
| ... | ... |
@@ -0,0 +1,501 @@ |
| 0 |
+/* |
|
| 1 |
+ * quarterpel DSP functions |
|
| 2 |
+ * |
|
| 3 |
+ * This file is part of Libav. |
|
| 4 |
+ * |
|
| 5 |
+ * Libav is free software; you can redistribute it and/or |
|
| 6 |
+ * modify it under the terms of the GNU Lesser General Public |
|
| 7 |
+ * License as published by the Free Software Foundation; either |
|
| 8 |
+ * version 2.1 of the License, or (at your option) any later version. |
|
| 9 |
+ * |
|
| 10 |
+ * Libav is distributed in the hope that it will be useful, |
|
| 11 |
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
| 12 |
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
| 13 |
+ * Lesser General Public License for more details. |
|
| 14 |
+ * |
|
| 15 |
+ * You should have received a copy of the GNU Lesser General Public |
|
| 16 |
+ * License along with Libav; if not, write to the Free Software |
|
| 17 |
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
| 18 |
+ */ |
|
| 19 |
+ |
|
| 20 |
+#include <stddef.h> |
|
| 21 |
+#include <stdint.h> |
|
| 22 |
+ |
|
| 23 |
+#include "config.h" |
|
| 24 |
+#include "libavutil/attributes.h" |
|
| 25 |
+#include "libavutil/cpu.h" |
|
| 26 |
+#include "libavutil/x86/cpu.h" |
|
| 27 |
+#include "libavcodec/pixels.h" |
|
| 28 |
+#include "libavcodec/qpeldsp.h" |
|
| 29 |
+#include "fpel.h" |
|
| 30 |
+ |
|
| 31 |
+void ff_put_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, |
|
| 32 |
+ int dstStride, int src1Stride, int h); |
|
| 33 |
+void ff_put_no_rnd_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, |
|
| 34 |
+ uint8_t *src2, int dstStride, |
|
| 35 |
+ int src1Stride, int h); |
|
| 36 |
+void ff_avg_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, |
|
| 37 |
+ int dstStride, int src1Stride, int h); |
|
| 38 |
+void ff_put_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, |
|
| 39 |
+ int dstStride, int src1Stride, int h); |
|
| 40 |
+void ff_avg_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, |
|
| 41 |
+ int dstStride, int src1Stride, int h); |
|
| 42 |
+void ff_put_no_rnd_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2, |
|
| 43 |
+ int dstStride, int src1Stride, int h); |
|
| 44 |
+void ff_put_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 45 |
+ int dstStride, int srcStride, int h); |
|
| 46 |
+void ff_avg_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 47 |
+ int dstStride, int srcStride, int h); |
|
| 48 |
+void ff_put_no_rnd_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 49 |
+ int dstStride, int srcStride, |
|
| 50 |
+ int h); |
|
| 51 |
+void ff_put_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 52 |
+ int dstStride, int srcStride, int h); |
|
| 53 |
+void ff_avg_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 54 |
+ int dstStride, int srcStride, int h); |
|
| 55 |
+void ff_put_no_rnd_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 56 |
+ int dstStride, int srcStride, |
|
| 57 |
+ int h); |
|
| 58 |
+void ff_put_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 59 |
+ int dstStride, int srcStride); |
|
| 60 |
+void ff_avg_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 61 |
+ int dstStride, int srcStride); |
|
| 62 |
+void ff_put_no_rnd_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 63 |
+ int dstStride, int srcStride); |
|
| 64 |
+void ff_put_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 65 |
+ int dstStride, int srcStride); |
|
| 66 |
+void ff_avg_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 67 |
+ int dstStride, int srcStride); |
|
| 68 |
+void ff_put_no_rnd_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, |
|
| 69 |
+ int dstStride, int srcStride); |
|
| 70 |
+#define ff_put_no_rnd_pixels16_mmxext ff_put_pixels16_mmxext |
|
| 71 |
+#define ff_put_no_rnd_pixels8_mmxext ff_put_pixels8_mmxext |
|
| 72 |
+ |
|
| 73 |
+#if HAVE_YASM |
|
| 74 |
+ |
|
| 75 |
+CALL_2X_PIXELS(ff_avg_pixels16_mmxext, ff_avg_pixels8_mmxext, 8) |
|
| 76 |
+CALL_2X_PIXELS(ff_put_pixels16_mmxext, ff_put_pixels8_mmxext, 8) |
|
| 77 |
+ |
|
| 78 |
+#define QPEL_OP(OPNAME, RND, MMX) \ |
|
| 79 |
+static void OPNAME ## qpel8_mc00_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 80 |
+ ptrdiff_t stride) \ |
|
| 81 |
+{ \
|
|
| 82 |
+ ff_ ## OPNAME ## pixels8_ ## MMX(dst, src, stride, 8); \ |
|
| 83 |
+} \ |
|
| 84 |
+ \ |
|
| 85 |
+static void OPNAME ## qpel8_mc10_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 86 |
+ ptrdiff_t stride) \ |
|
| 87 |
+{ \
|
|
| 88 |
+ uint64_t temp[8]; \ |
|
| 89 |
+ uint8_t *const half = (uint8_t *) temp; \ |
|
| 90 |
+ ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(half, src, 8, \ |
|
| 91 |
+ stride, 8); \ |
|
| 92 |
+ ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src, half, \ |
|
| 93 |
+ stride, stride, 8); \ |
|
| 94 |
+} \ |
|
| 95 |
+ \ |
|
| 96 |
+static void OPNAME ## qpel8_mc20_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 97 |
+ ptrdiff_t stride) \ |
|
| 98 |
+{ \
|
|
| 99 |
+ ff_ ## OPNAME ## mpeg4_qpel8_h_lowpass_ ## MMX(dst, src, stride, \ |
|
| 100 |
+ stride, 8); \ |
|
| 101 |
+} \ |
|
| 102 |
+ \ |
|
| 103 |
+static void OPNAME ## qpel8_mc30_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 104 |
+ ptrdiff_t stride) \ |
|
| 105 |
+{ \
|
|
| 106 |
+ uint64_t temp[8]; \ |
|
| 107 |
+ uint8_t *const half = (uint8_t *) temp; \ |
|
| 108 |
+ ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(half, src, 8, \ |
|
| 109 |
+ stride, 8); \ |
|
| 110 |
+ ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src + 1, half, stride, \ |
|
| 111 |
+ stride, 8); \ |
|
| 112 |
+} \ |
|
| 113 |
+ \ |
|
| 114 |
+static void OPNAME ## qpel8_mc01_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 115 |
+ ptrdiff_t stride) \ |
|
| 116 |
+{ \
|
|
| 117 |
+ uint64_t temp[8]; \ |
|
| 118 |
+ uint8_t *const half = (uint8_t *) temp; \ |
|
| 119 |
+ ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(half, src, \ |
|
| 120 |
+ 8, stride); \ |
|
| 121 |
+ ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src, half, \ |
|
| 122 |
+ stride, stride, 8); \ |
|
| 123 |
+} \ |
|
| 124 |
+ \ |
|
| 125 |
+static void OPNAME ## qpel8_mc02_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 126 |
+ ptrdiff_t stride) \ |
|
| 127 |
+{ \
|
|
| 128 |
+ ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, src, \ |
|
| 129 |
+ stride, stride); \ |
|
| 130 |
+} \ |
|
| 131 |
+ \ |
|
| 132 |
+static void OPNAME ## qpel8_mc03_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 133 |
+ ptrdiff_t stride) \ |
|
| 134 |
+{ \
|
|
| 135 |
+ uint64_t temp[8]; \ |
|
| 136 |
+ uint8_t *const half = (uint8_t *) temp; \ |
|
| 137 |
+ ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(half, src, \ |
|
| 138 |
+ 8, stride); \ |
|
| 139 |
+ ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src + stride, half, stride,\ |
|
| 140 |
+ stride, 8); \ |
|
| 141 |
+} \ |
|
| 142 |
+ \ |
|
| 143 |
+static void OPNAME ## qpel8_mc11_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 144 |
+ ptrdiff_t stride) \ |
|
| 145 |
+{ \
|
|
| 146 |
+ uint64_t half[8 + 9]; \ |
|
| 147 |
+ uint8_t *const halfH = (uint8_t *) half + 64; \ |
|
| 148 |
+ uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 149 |
+ ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 150 |
+ stride, 9); \ |
|
| 151 |
+ ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, 8, \ |
|
| 152 |
+ stride, 9); \ |
|
| 153 |
+ ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\ |
|
| 154 |
+ ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV, \ |
|
| 155 |
+ stride, 8, 8); \ |
|
| 156 |
+} \ |
|
| 157 |
+ \ |
|
| 158 |
+static void OPNAME ## qpel8_mc31_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 159 |
+ ptrdiff_t stride) \ |
|
| 160 |
+{ \
|
|
| 161 |
+ uint64_t half[8 + 9]; \ |
|
| 162 |
+ uint8_t *const halfH = (uint8_t *) half + 64; \ |
|
| 163 |
+ uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 164 |
+ ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 165 |
+ stride, 9); \ |
|
| 166 |
+ ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8, \ |
|
| 167 |
+ stride, 9); \ |
|
| 168 |
+ ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\ |
|
| 169 |
+ ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV, \ |
|
| 170 |
+ stride, 8, 8); \ |
|
| 171 |
+} \ |
|
| 172 |
+ \ |
|
| 173 |
+static void OPNAME ## qpel8_mc13_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 174 |
+ ptrdiff_t stride) \ |
|
| 175 |
+{ \
|
|
| 176 |
+ uint64_t half[8 + 9]; \ |
|
| 177 |
+ uint8_t *const halfH = (uint8_t *) half + 64; \ |
|
| 178 |
+ uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 179 |
+ ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 180 |
+ stride, 9); \ |
|
| 181 |
+ ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, 8, \ |
|
| 182 |
+ stride, 9); \ |
|
| 183 |
+ ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\ |
|
| 184 |
+ ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV, \ |
|
| 185 |
+ stride, 8, 8); \ |
|
| 186 |
+} \ |
|
| 187 |
+ \ |
|
| 188 |
+static void OPNAME ## qpel8_mc33_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 189 |
+ ptrdiff_t stride) \ |
|
| 190 |
+{ \
|
|
| 191 |
+ uint64_t half[8 + 9]; \ |
|
| 192 |
+ uint8_t *const halfH = (uint8_t *) half + 64; \ |
|
| 193 |
+ uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 194 |
+ ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 195 |
+ stride, 9); \ |
|
| 196 |
+ ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8, \ |
|
| 197 |
+ stride, 9); \ |
|
| 198 |
+ ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\ |
|
| 199 |
+ ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV, \ |
|
| 200 |
+ stride, 8, 8); \ |
|
| 201 |
+} \ |
|
| 202 |
+ \ |
|
| 203 |
+static void OPNAME ## qpel8_mc21_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 204 |
+ ptrdiff_t stride) \ |
|
| 205 |
+{ \
|
|
| 206 |
+ uint64_t half[8 + 9]; \ |
|
| 207 |
+ uint8_t *const halfH = (uint8_t *) half + 64; \ |
|
| 208 |
+ uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 209 |
+ ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 210 |
+ stride, 9); \ |
|
| 211 |
+ ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\ |
|
| 212 |
+ ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV, \ |
|
| 213 |
+ stride, 8, 8); \ |
|
| 214 |
+} \ |
|
| 215 |
+ \ |
|
| 216 |
+static void OPNAME ## qpel8_mc23_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 217 |
+ ptrdiff_t stride) \ |
|
| 218 |
+{ \
|
|
| 219 |
+ uint64_t half[8 + 9]; \ |
|
| 220 |
+ uint8_t *const halfH = (uint8_t *) half + 64; \ |
|
| 221 |
+ uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 222 |
+ ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 223 |
+ stride, 9); \ |
|
| 224 |
+ ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\ |
|
| 225 |
+ ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV, \ |
|
| 226 |
+ stride, 8, 8); \ |
|
| 227 |
+} \ |
|
| 228 |
+ \ |
|
| 229 |
+static void OPNAME ## qpel8_mc12_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 230 |
+ ptrdiff_t stride) \ |
|
| 231 |
+{ \
|
|
| 232 |
+ uint64_t half[8 + 9]; \ |
|
| 233 |
+ uint8_t *const halfH = (uint8_t *) half; \ |
|
| 234 |
+ ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 235 |
+ stride, 9); \ |
|
| 236 |
+ ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, \ |
|
| 237 |
+ 8, stride, 9); \ |
|
| 238 |
+ ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH, \ |
|
| 239 |
+ stride, 8); \ |
|
| 240 |
+} \ |
|
| 241 |
+ \ |
|
| 242 |
+static void OPNAME ## qpel8_mc32_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 243 |
+ ptrdiff_t stride) \ |
|
| 244 |
+{ \
|
|
| 245 |
+ uint64_t half[8 + 9]; \ |
|
| 246 |
+ uint8_t *const halfH = (uint8_t *) half; \ |
|
| 247 |
+ ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 248 |
+ stride, 9); \ |
|
| 249 |
+ ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8, \ |
|
| 250 |
+ stride, 9); \ |
|
| 251 |
+ ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH, \ |
|
| 252 |
+ stride, 8); \ |
|
| 253 |
+} \ |
|
| 254 |
+ \ |
|
| 255 |
+static void OPNAME ## qpel8_mc22_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 256 |
+ ptrdiff_t stride) \ |
|
| 257 |
+{ \
|
|
| 258 |
+ uint64_t half[9]; \ |
|
| 259 |
+ uint8_t *const halfH = (uint8_t *) half; \ |
|
| 260 |
+ ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, \ |
|
| 261 |
+ stride, 9); \ |
|
| 262 |
+ ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH, \ |
|
| 263 |
+ stride, 8); \ |
|
| 264 |
+} \ |
|
| 265 |
+ \ |
|
| 266 |
+static void OPNAME ## qpel16_mc00_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 267 |
+ ptrdiff_t stride) \ |
|
| 268 |
+{ \
|
|
| 269 |
+ ff_ ## OPNAME ## pixels16_ ## MMX(dst, src, stride, 16); \ |
|
| 270 |
+} \ |
|
| 271 |
+ \ |
|
| 272 |
+static void OPNAME ## qpel16_mc10_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 273 |
+ ptrdiff_t stride) \ |
|
| 274 |
+{ \
|
|
| 275 |
+ uint64_t temp[32]; \ |
|
| 276 |
+ uint8_t *const half = (uint8_t *) temp; \ |
|
| 277 |
+ ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(half, src, 16, \ |
|
| 278 |
+ stride, 16); \ |
|
| 279 |
+ ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src, half, stride, \ |
|
| 280 |
+ stride, 16); \ |
|
| 281 |
+} \ |
|
| 282 |
+ \ |
|
| 283 |
+static void OPNAME ## qpel16_mc20_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 284 |
+ ptrdiff_t stride) \ |
|
| 285 |
+{ \
|
|
| 286 |
+ ff_ ## OPNAME ## mpeg4_qpel16_h_lowpass_ ## MMX(dst, src, \ |
|
| 287 |
+ stride, stride, 16);\ |
|
| 288 |
+} \ |
|
| 289 |
+ \ |
|
| 290 |
+static void OPNAME ## qpel16_mc30_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 291 |
+ ptrdiff_t stride) \ |
|
| 292 |
+{ \
|
|
| 293 |
+ uint64_t temp[32]; \ |
|
| 294 |
+ uint8_t *const half = (uint8_t*) temp; \ |
|
| 295 |
+ ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(half, src, 16, \ |
|
| 296 |
+ stride, 16); \ |
|
| 297 |
+ ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src + 1, half, \ |
|
| 298 |
+ stride, stride, 16); \ |
|
| 299 |
+} \ |
|
| 300 |
+ \ |
|
| 301 |
+static void OPNAME ## qpel16_mc01_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 302 |
+ ptrdiff_t stride) \ |
|
| 303 |
+{ \
|
|
| 304 |
+ uint64_t temp[32]; \ |
|
| 305 |
+ uint8_t *const half = (uint8_t *) temp; \ |
|
| 306 |
+ ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(half, src, 16, \ |
|
| 307 |
+ stride); \ |
|
| 308 |
+ ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src, half, stride, \ |
|
| 309 |
+ stride, 16); \ |
|
| 310 |
+} \ |
|
| 311 |
+ \ |
|
| 312 |
+static void OPNAME ## qpel16_mc02_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 313 |
+ ptrdiff_t stride) \ |
|
| 314 |
+{ \
|
|
| 315 |
+ ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, src, \ |
|
| 316 |
+ stride, stride); \ |
|
| 317 |
+} \ |
|
| 318 |
+ \ |
|
| 319 |
+static void OPNAME ## qpel16_mc03_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 320 |
+ ptrdiff_t stride) \ |
|
| 321 |
+{ \
|
|
| 322 |
+ uint64_t temp[32]; \ |
|
| 323 |
+ uint8_t *const half = (uint8_t *) temp; \ |
|
| 324 |
+ ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(half, src, 16, \ |
|
| 325 |
+ stride); \ |
|
| 326 |
+ ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src+stride, half, \ |
|
| 327 |
+ stride, stride, 16); \ |
|
| 328 |
+} \ |
|
| 329 |
+ \ |
|
| 330 |
+static void OPNAME ## qpel16_mc11_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 331 |
+ ptrdiff_t stride) \ |
|
| 332 |
+{ \
|
|
| 333 |
+ uint64_t half[16 * 2 + 17 * 2]; \ |
|
| 334 |
+ uint8_t *const halfH = (uint8_t *) half + 256; \ |
|
| 335 |
+ uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 336 |
+ ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 337 |
+ stride, 17); \ |
|
| 338 |
+ ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16, \ |
|
| 339 |
+ stride, 17); \ |
|
| 340 |
+ ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \ |
|
| 341 |
+ 16, 16); \ |
|
| 342 |
+ ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV, \ |
|
| 343 |
+ stride, 16, 16); \ |
|
| 344 |
+} \ |
|
| 345 |
+ \ |
|
| 346 |
+static void OPNAME ## qpel16_mc31_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 347 |
+ ptrdiff_t stride) \ |
|
| 348 |
+{ \
|
|
| 349 |
+ uint64_t half[16 * 2 + 17 * 2]; \ |
|
| 350 |
+ uint8_t *const halfH = (uint8_t *) half + 256; \ |
|
| 351 |
+ uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 352 |
+ ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 353 |
+ stride, 17); \ |
|
| 354 |
+ ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16, \ |
|
| 355 |
+ stride, 17); \ |
|
| 356 |
+ ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \ |
|
| 357 |
+ 16, 16); \ |
|
| 358 |
+ ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV, \ |
|
| 359 |
+ stride, 16, 16); \ |
|
| 360 |
+} \ |
|
| 361 |
+ \ |
|
| 362 |
+static void OPNAME ## qpel16_mc13_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 363 |
+ ptrdiff_t stride) \ |
|
| 364 |
+{ \
|
|
| 365 |
+ uint64_t half[16 * 2 + 17 * 2]; \ |
|
| 366 |
+ uint8_t *const halfH = (uint8_t *) half + 256; \ |
|
| 367 |
+ uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 368 |
+ ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 369 |
+ stride, 17); \ |
|
| 370 |
+ ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16, \ |
|
| 371 |
+ stride, 17); \ |
|
| 372 |
+ ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \ |
|
| 373 |
+ 16, 16); \ |
|
| 374 |
+ ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV, \ |
|
| 375 |
+ stride, 16, 16); \ |
|
| 376 |
+} \ |
|
| 377 |
+ \ |
|
| 378 |
+static void OPNAME ## qpel16_mc33_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 379 |
+ ptrdiff_t stride) \ |
|
| 380 |
+{ \
|
|
| 381 |
+ uint64_t half[16 * 2 + 17 * 2]; \ |
|
| 382 |
+ uint8_t *const halfH = (uint8_t *) half + 256; \ |
|
| 383 |
+ uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 384 |
+ ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 385 |
+ stride, 17); \ |
|
| 386 |
+ ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16, \ |
|
| 387 |
+ stride, 17); \ |
|
| 388 |
+ ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \ |
|
| 389 |
+ 16, 16); \ |
|
| 390 |
+ ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV, \ |
|
| 391 |
+ stride, 16, 16); \ |
|
| 392 |
+} \ |
|
| 393 |
+ \ |
|
| 394 |
+static void OPNAME ## qpel16_mc21_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 395 |
+ ptrdiff_t stride) \ |
|
| 396 |
+{ \
|
|
| 397 |
+ uint64_t half[16 * 2 + 17 * 2]; \ |
|
| 398 |
+ uint8_t *const halfH = (uint8_t *) half + 256; \ |
|
| 399 |
+ uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 400 |
+ ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 401 |
+ stride, 17); \ |
|
| 402 |
+ ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \ |
|
| 403 |
+ 16, 16); \ |
|
| 404 |
+ ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV, \ |
|
| 405 |
+ stride, 16, 16); \ |
|
| 406 |
+} \ |
|
| 407 |
+ \ |
|
| 408 |
+static void OPNAME ## qpel16_mc23_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 409 |
+ ptrdiff_t stride) \ |
|
| 410 |
+{ \
|
|
| 411 |
+ uint64_t half[16 * 2 + 17 * 2]; \ |
|
| 412 |
+ uint8_t *const halfH = (uint8_t *) half + 256; \ |
|
| 413 |
+ uint8_t *const halfHV = (uint8_t *) half; \ |
|
| 414 |
+ ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 415 |
+ stride, 17); \ |
|
| 416 |
+ ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, \ |
|
| 417 |
+ 16, 16); \ |
|
| 418 |
+ ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV, \ |
|
| 419 |
+ stride, 16, 16); \ |
|
| 420 |
+} \ |
|
| 421 |
+ \ |
|
| 422 |
+static void OPNAME ## qpel16_mc12_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 423 |
+ ptrdiff_t stride) \ |
|
| 424 |
+{ \
|
|
| 425 |
+ uint64_t half[17 * 2]; \ |
|
| 426 |
+ uint8_t *const halfH = (uint8_t *) half; \ |
|
| 427 |
+ ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 428 |
+ stride, 17); \ |
|
| 429 |
+ ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16, \ |
|
| 430 |
+ stride, 17); \ |
|
| 431 |
+ ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH, \ |
|
| 432 |
+ stride, 16); \ |
|
| 433 |
+} \ |
|
| 434 |
+ \ |
|
| 435 |
+static void OPNAME ## qpel16_mc32_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 436 |
+ ptrdiff_t stride) \ |
|
| 437 |
+{ \
|
|
| 438 |
+ uint64_t half[17 * 2]; \ |
|
| 439 |
+ uint8_t *const halfH = (uint8_t *) half; \ |
|
| 440 |
+ ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 441 |
+ stride, 17); \ |
|
| 442 |
+ ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16, \ |
|
| 443 |
+ stride, 17); \ |
|
| 444 |
+ ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH, \ |
|
| 445 |
+ stride, 16); \ |
|
| 446 |
+} \ |
|
| 447 |
+ \ |
|
| 448 |
+static void OPNAME ## qpel16_mc22_ ## MMX(uint8_t *dst, uint8_t *src, \ |
|
| 449 |
+ ptrdiff_t stride) \ |
|
| 450 |
+{ \
|
|
| 451 |
+ uint64_t half[17 * 2]; \ |
|
| 452 |
+ uint8_t *const halfH = (uint8_t *) half; \ |
|
| 453 |
+ ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, \ |
|
| 454 |
+ stride, 17); \ |
|
| 455 |
+ ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH, \ |
|
| 456 |
+ stride, 16); \ |
|
| 457 |
+} |
|
| 458 |
+ |
|
| 459 |
+QPEL_OP(put_, _, mmxext) |
|
| 460 |
+QPEL_OP(avg_, _, mmxext) |
|
| 461 |
+QPEL_OP(put_no_rnd_, _no_rnd_, mmxext) |
|
| 462 |
+ |
|
| 463 |
+#endif /* HAVE_YASM */ |
|
| 464 |
+ |
|
| 465 |
+#define SET_QPEL_FUNCS(PFX, IDX, SIZE, CPU, PREFIX) \ |
|
| 466 |
+do { \
|
|
| 467 |
+ c->PFX ## _pixels_tab[IDX][ 0] = PREFIX ## PFX ## SIZE ## _mc00_ ## CPU; \ |
|
| 468 |
+ c->PFX ## _pixels_tab[IDX][ 1] = PREFIX ## PFX ## SIZE ## _mc10_ ## CPU; \ |
|
| 469 |
+ c->PFX ## _pixels_tab[IDX][ 2] = PREFIX ## PFX ## SIZE ## _mc20_ ## CPU; \ |
|
| 470 |
+ c->PFX ## _pixels_tab[IDX][ 3] = PREFIX ## PFX ## SIZE ## _mc30_ ## CPU; \ |
|
| 471 |
+ c->PFX ## _pixels_tab[IDX][ 4] = PREFIX ## PFX ## SIZE ## _mc01_ ## CPU; \ |
|
| 472 |
+ c->PFX ## _pixels_tab[IDX][ 5] = PREFIX ## PFX ## SIZE ## _mc11_ ## CPU; \ |
|
| 473 |
+ c->PFX ## _pixels_tab[IDX][ 6] = PREFIX ## PFX ## SIZE ## _mc21_ ## CPU; \ |
|
| 474 |
+ c->PFX ## _pixels_tab[IDX][ 7] = PREFIX ## PFX ## SIZE ## _mc31_ ## CPU; \ |
|
| 475 |
+ c->PFX ## _pixels_tab[IDX][ 8] = PREFIX ## PFX ## SIZE ## _mc02_ ## CPU; \ |
|
| 476 |
+ c->PFX ## _pixels_tab[IDX][ 9] = PREFIX ## PFX ## SIZE ## _mc12_ ## CPU; \ |
|
| 477 |
+ c->PFX ## _pixels_tab[IDX][10] = PREFIX ## PFX ## SIZE ## _mc22_ ## CPU; \ |
|
| 478 |
+ c->PFX ## _pixels_tab[IDX][11] = PREFIX ## PFX ## SIZE ## _mc32_ ## CPU; \ |
|
| 479 |
+ c->PFX ## _pixels_tab[IDX][12] = PREFIX ## PFX ## SIZE ## _mc03_ ## CPU; \ |
|
| 480 |
+ c->PFX ## _pixels_tab[IDX][13] = PREFIX ## PFX ## SIZE ## _mc13_ ## CPU; \ |
|
| 481 |
+ c->PFX ## _pixels_tab[IDX][14] = PREFIX ## PFX ## SIZE ## _mc23_ ## CPU; \ |
|
| 482 |
+ c->PFX ## _pixels_tab[IDX][15] = PREFIX ## PFX ## SIZE ## _mc33_ ## CPU; \ |
|
| 483 |
+} while (0) |
|
| 484 |
+ |
|
| 485 |
+av_cold void ff_qpeldsp_init_x86(QpelDSPContext *c) |
|
| 486 |
+{
|
|
| 487 |
+ int cpu_flags = av_get_cpu_flags(); |
|
| 488 |
+ |
|
| 489 |
+ if (X86_MMXEXT(cpu_flags)) {
|
|
| 490 |
+#if HAVE_MMXEXT_EXTERNAL |
|
| 491 |
+ SET_QPEL_FUNCS(avg_qpel, 0, 16, mmxext, ); |
|
| 492 |
+ SET_QPEL_FUNCS(avg_qpel, 1, 8, mmxext, ); |
|
| 493 |
+ |
|
| 494 |
+ SET_QPEL_FUNCS(put_qpel, 0, 16, mmxext, ); |
|
| 495 |
+ SET_QPEL_FUNCS(put_qpel, 1, 8, mmxext, ); |
|
| 496 |
+ SET_QPEL_FUNCS(put_no_rnd_qpel, 0, 16, mmxext, ); |
|
| 497 |
+ SET_QPEL_FUNCS(put_no_rnd_qpel, 1, 8, mmxext, ); |
|
| 498 |
+#endif /* HAVE_MMXEXT_EXTERNAL */ |
|
| 499 |
+ } |
|
| 500 |
+} |