dsputil: Move thirdpel-related bits into their own context

12 files changed, 489 insertions, 370 deletions

diff --git a/configure b/configure
index fced12f9bd..f26374a3d3 100755
--- a/configure
+++ b/configure
@@ -1537,6 +1537,7 @@ CONFIG_EXTRA="
     rtpdec
     rtpenc_chain
     sinewin
+    tpeldsp
     videodsp
     vp3dsp
 "
@@ -1820,7 +1821,7 @@ sipr_decoder_select="lsp"
 sp5x_decoder_select="mjpeg_decoder"
 svq1_decoder_select="hpeldsp"
 svq1_encoder_select="aandcttables dsputil hpeldsp mpegvideoenc"
-svq3_decoder_select="h264_decoder hpeldsp"
+svq3_decoder_select="h264_decoder hpeldsp tpeldsp"
 svq3_decoder_suggest="zlib"
 tak_decoder_select="dsputil"
 theora_decoder_select="vp3_decoder"
diff --git a/doc/optimization.txt b/doc/optimization.txt
index 42ad15ed28..b51183fa34 100644
--- a/doc/optimization.txt
+++ b/doc/optimization.txt
@@ -79,9 +79,6 @@ qpel{8,16}_mc??_old_c / *pixels{8,16}_l4
     Just used to work around a bug in an old libavcodec encoder version.
     Don't optimize them.
 
-tpel_mc_func {put,avg}_tpel_pixels_tab
-    Used only for SVQ3, so only optimize them if you need fast SVQ3 decoding.
-
 add_bytes/diff_bytes
     For huffyuv only, optimize if you want a faster ffhuffyuv codec.
 
diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index c04b3f1294..3d178a1387 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -65,6 +65,7 @@ OBJS-$(CONFIG_RANGECODER)              += rangecoder.o
 RDFT-OBJS-$(CONFIG_HARDCODED_TABLES)   += sin_tables.o
 OBJS-$(CONFIG_RDFT)                    += rdft.o $(RDFT-OBJS-yes)
 OBJS-$(CONFIG_SINEWIN)                 += sinewin.o
+OBJS-$(CONFIG_TPELDSP)                 += tpeldsp.o
 OBJS-$(CONFIG_VAAPI)                   += vaapi.o
 OBJS-$(CONFIG_VDPAU)                   += vdpau.o
 OBJS-$(CONFIG_VIDEODSP)                += videodsp.o
diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c
index 9fe6f0b757..b81ba47521 100644
--- a/libavcodec/dsputil.c
+++ b/libavcodec/dsputil.c
@@ -48,6 +48,7 @@ uint32_t ff_square_tab[512] = { 0, };
 #undef BIT_DEPTH
 
 #define BIT_DEPTH 8
+#include "tpel_template.c"
 #include "dsputil_template.c"
 
 // 0x7f7f7f7f or 0x7f7f7f7f7f7f7f7f or whatever, depending on the cpu's native arithmetic size
@@ -540,284 +541,6 @@ void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
     }
 }
 
-static inline void put_tpel_pixels_mc00_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    switch (width) {
-    case 2:
-        put_pixels2_8_c(dst, src, stride, height);
-        break;
-    case 4:
-        put_pixels4_8_c(dst, src, stride, height);
-        break;
-    case 8:
-        put_pixels8_8_c(dst, src, stride, height);
-        break;
-    case 16:
-        put_pixels16_8_c(dst, src, stride, height);
-        break;
-    }
-}
-
-static inline void put_tpel_pixels_mc10_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = ((2 * src[j] + src[j + 1] + 1) *
-                      683) >> 11;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void put_tpel_pixels_mc20_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = ((src[j] + 2 * src[j + 1] + 1) *
-                      683) >> 11;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void put_tpel_pixels_mc01_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = ((2 * src[j] + src[j + stride] + 1) *
-                      683) >> 11;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void put_tpel_pixels_mc11_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = ((4 * src[j]          + 3 * src[j + 1] +
-                       3 * src[j + stride] + 2 * src[j + stride + 1] + 6) *
-                      2731) >> 15;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void put_tpel_pixels_mc12_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = ((3 * src[j]          + 2 * src[j + 1] +
-                       4 * src[j + stride] + 3 * src[j + stride + 1] + 6) *
-                      2731) >> 15;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void put_tpel_pixels_mc02_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = ((src[j] + 2 * src[j + stride] + 1) *
-                      683) >> 11;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void put_tpel_pixels_mc21_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = ((3 * src[j]          + 4 * src[j + 1] +
-                       2 * src[j + stride] + 3 * src[j + stride + 1] + 6) *
-                      2731) >> 15;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void put_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = ((2 * src[j]          + 3 * src[j + 1] +
-                       3 * src[j + stride] + 4 * src[j + stride + 1] + 6) *
-                      2731) >> 15;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void avg_tpel_pixels_mc00_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    switch (width) {
-    case 2:
-        avg_pixels2_8_c(dst, src, stride, height);
-        break;
-    case 4:
-        avg_pixels4_8_c(dst, src, stride, height);
-        break;
-    case 8:
-        avg_pixels8_8_c(dst, src, stride, height);
-        break;
-    case 16:
-        avg_pixels16_8_c(dst, src, stride, height);
-        break;
-    }
-}
-
-static inline void avg_tpel_pixels_mc10_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = (dst[j] +
-                      (((2 * src[j] + src[j + 1] + 1) *
-                        683) >> 11) + 1) >> 1;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void avg_tpel_pixels_mc20_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = (dst[j] +
-                      (((src[j] + 2 * src[j + 1] + 1) *
-                        683) >> 11) + 1) >> 1;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void avg_tpel_pixels_mc01_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = (dst[j] +
-                      (((2 * src[j] + src[j + stride] + 1) *
-                        683) >> 11) + 1) >> 1;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void avg_tpel_pixels_mc11_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = (dst[j] +
-                      (((4 * src[j]          + 3 * src[j + 1] +
-                         3 * src[j + stride] + 2 * src[j + stride + 1] + 6) *
-                        2731) >> 15) + 1) >> 1;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void avg_tpel_pixels_mc12_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = (dst[j] +
-                      (((3 * src[j]          + 2 * src[j + 1] +
-                         4 * src[j + stride] + 3 * src[j + stride + 1] + 6) *
-                        2731) >> 15) + 1) >> 1;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void avg_tpel_pixels_mc02_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = (dst[j] +
-                      (((src[j] + 2 * src[j + stride] + 1) *
-                        683) >> 11) + 1) >> 1;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void avg_tpel_pixels_mc21_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = (dst[j] +
-                      (((3 * src[j]          + 4 * src[j + 1] +
-                         2 * src[j + stride] + 3 * src[j + stride + 1] + 6) *
-                        2731) >> 15) + 1) >> 1;
-        src += stride;
-        dst += stride;
-    }
-}
-
-static inline void avg_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src,
-                                          int stride, int width, int height)
-{
-    int i, j;
-
-    for (i = 0; i < height; i++) {
-        for (j = 0; j < width; j++)
-            dst[j] = (dst[j] +
-                      (((2 * src[j]          + 3 * src[j + 1] +
-                         3 * src[j + stride] + 4 * src[j + stride + 1] + 6) *
-                        2731) >> 15) + 1) >> 1;
-        src += stride;
-        dst += stride;
-    }
-}
-
 #define QPEL_MC(r, OPNAME, RND, OP)                                           \
 static void OPNAME ## mpeg4_qpel8_h_lowpass(uint8_t *dst, uint8_t *src,       \
                                             int dstStride, int srcStride,     \
@@ -2781,26 +2504,6 @@ av_cold void ff_dsputil_init(DSPContext *c, AVCodecContext *avctx)
     c->pix_abs[1][2] = pix_abs8_y2_c;
     c->pix_abs[1][3] = pix_abs8_xy2_c;
 
-    c->put_tpel_pixels_tab[0]  = put_tpel_pixels_mc00_c;
-    c->put_tpel_pixels_tab[1]  = put_tpel_pixels_mc10_c;
-    c->put_tpel_pixels_tab[2]  = put_tpel_pixels_mc20_c;
-    c->put_tpel_pixels_tab[4]  = put_tpel_pixels_mc01_c;
-    c->put_tpel_pixels_tab[5]  = put_tpel_pixels_mc11_c;
-    c->put_tpel_pixels_tab[6]  = put_tpel_pixels_mc21_c;
-    c->put_tpel_pixels_tab[8]  = put_tpel_pixels_mc02_c;
-    c->put_tpel_pixels_tab[9]  = put_tpel_pixels_mc12_c;
-    c->put_tpel_pixels_tab[10] = put_tpel_pixels_mc22_c;
-
-    c->avg_tpel_pixels_tab[0]  = avg_tpel_pixels_mc00_c;
-    c->avg_tpel_pixels_tab[1]  = avg_tpel_pixels_mc10_c;
-    c->avg_tpel_pixels_tab[2]  = avg_tpel_pixels_mc20_c;
-    c->avg_tpel_pixels_tab[4]  = avg_tpel_pixels_mc01_c;
-    c->avg_tpel_pixels_tab[5]  = avg_tpel_pixels_mc11_c;
-    c->avg_tpel_pixels_tab[6]  = avg_tpel_pixels_mc21_c;
-    c->avg_tpel_pixels_tab[8]  = avg_tpel_pixels_mc02_c;
-    c->avg_tpel_pixels_tab[9]  = avg_tpel_pixels_mc12_c;
-    c->avg_tpel_pixels_tab[10] = avg_tpel_pixels_mc22_c;
-
 #define dspfunc(PFX, IDX, NUM)                              \
     c->PFX ## _pixels_tab[IDX][0]  = PFX ## NUM ## _mc00_c; \
     c->PFX ## _pixels_tab[IDX][1]  = PFX ## NUM ## _mc10_c; \
diff --git a/libavcodec/dsputil.h b/libavcodec/dsputil.h
index d596e29099..052ac50694 100644
--- a/libavcodec/dsputil.h
+++ b/libavcodec/dsputil.h
@@ -71,9 +71,6 @@ void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
  * Block sizes for op_pixels_func are 8x4,8x8 16x8 16x16.
  * h for op_pixels_func is limited to { width / 2, width },
  * but never larger than 16 and never smaller than 4. */
-typedef void (*tpel_mc_func)(uint8_t *block /* align width (8 or 16) */,
-                             const uint8_t *pixels /* align 1 */,
-                             int line_size, int w, int h);
 typedef void (*qpel_mc_func)(uint8_t *dst /* align width (8 or 16) */,
                              uint8_t *src /* align 1 */, ptrdiff_t stride);
 
@@ -188,19 +185,6 @@ typedef struct DSPContext {
     int (*ssd_int8_vs_int16)(const int8_t *pix1, const int16_t *pix2,
                              int size);
 
-    /**
-     * Thirdpel motion compensation with rounding (a + b + 1) >> 1.
-     * this is an array[12] of motion compensation functions for the
-     * 9 thirdpel positions<br>
-     * *pixels_tab[xthirdpel + 4 * ythirdpel]
-     * @param block destination where the result is stored
-     * @param pixels source
-     * @param line_size number of bytes in a horizontal line of block
-     * @param h height
-     */
-    tpel_mc_func put_tpel_pixels_tab[11]; // FIXME individual func ptr per width?
-    tpel_mc_func avg_tpel_pixels_tab[11]; // FIXME individual func ptr per width?
-
     qpel_mc_func put_qpel_pixels_tab[2][16];
     qpel_mc_func avg_qpel_pixels_tab[2][16];
     qpel_mc_func put_no_rnd_qpel_pixels_tab[2][16];
diff --git a/libavcodec/h264qpel_template.c b/libavcodec/h264qpel_template.c
index 71821798a4..d03b0dc443 100644
--- a/libavcodec/h264qpel_template.c
+++ b/libavcodec/h264qpel_template.c
@@ -24,6 +24,7 @@
 
 #include "bit_depth_template.c"
 #include "hpel_template.c"
+#include "tpel_template.c"
 
 static inline void FUNC(copy_block2)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
 {
diff --git a/libavcodec/hpel_template.c b/libavcodec/hpel_template.c
index 1bc18ccad0..0a8550a738 100644
--- a/libavcodec/hpel_template.c
+++ b/libavcodec/hpel_template.c
@@ -22,47 +22,6 @@
 #include "pixels.h"
 
 #define DEF_HPEL(OPNAME, OP)                                            \
-static inline void FUNCC(OPNAME ## _pixels2)(uint8_t *block,            \
-                                             const uint8_t *pixels,     \
-                                             ptrdiff_t line_size,       \
-                                             int h)                     \
-{                                                                       \
-    int i;                                                              \
-    for (i = 0; i < h; i++) {                                           \
-        OP(*((pixel2 *) block), AV_RN2P(pixels));                       \
-        pixels += line_size;                                            \
-        block  += line_size;                                            \
-    }                                                                   \
-}                                                                       \
-                                                                        \
-static inline void FUNCC(OPNAME ## _pixels4)(uint8_t *block,            \
-                                             const uint8_t *pixels,     \
-                                             ptrdiff_t line_size,       \
-                                             int h)                     \
-{                                                                       \
-    int i;                                                              \
-    for (i = 0; i < h; i++) {                                           \
-        OP(*((pixel4 *) block), AV_RN4P(pixels));                       \
-        pixels += line_size;                                            \
-        block  += line_size;                                            \
-    }                                                                   \
-}                                                                       \
-                                                                        \
-static inline void FUNCC(OPNAME ## _pixels8)(uint8_t *block,            \
-                                             const uint8_t *pixels,     \
-                                             ptrdiff_t line_size,       \
-                                             int h)                     \
-{                                                                       \
-    int i;                                                              \
-    for (i = 0; i < h; i++) {                                           \
-        OP(*((pixel4 *) block), AV_RN4P(pixels));                       \
-        OP(*((pixel4 *) (block + 4 * sizeof(pixel))),                   \
-           AV_RN4P(pixels + 4 * sizeof(pixel)));                        \
-        pixels += line_size;                                            \
-        block  += line_size;                                            \
-    }                                                                   \
-}                                                                       \
-                                                                        \
 static inline void FUNC(OPNAME ## _pixels8_l2)(uint8_t *dst,            \
                                                const uint8_t *src1,     \
                                                const uint8_t *src2,     \
@@ -134,10 +93,6 @@ static inline void FUNC(OPNAME ## _pixels16_l2)(uint8_t *dst,           \
                                 dst_stride, src_stride1,                \
                                 src_stride2, h);                        \
 }                                                                       \
-                                                                        \
-CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16),                              \
-               FUNCC(OPNAME ## _pixels8),                               \
-               8 * sizeof(pixel))
 
 #define op_avg(a, b) a = rnd_avg_pixel4(a, b)
 #define op_put(a, b) a = b
diff --git a/libavcodec/hpeldsp_template.c b/libavcodec/hpeldsp_template.c
index f190457b96..3039bfa9d6 100644
--- a/libavcodec/hpeldsp_template.c
+++ b/libavcodec/hpeldsp_template.c
@@ -33,6 +33,7 @@
 #include "bit_depth_template.c"
 
 #include "hpel_template.c"
+#include "tpel_template.c"
 
 #define PIXOP2(OPNAME, OP)                                              \
 static inline void FUNC(OPNAME ## _no_rnd_pixels8_l2)(uint8_t *dst,     \
diff --git a/libavcodec/svq3.c b/libavcodec/svq3.c
index 4916314c08..fc2120b2cb 100644
--- a/libavcodec/svq3.c
+++ b/libavcodec/svq3.c
@@ -54,6 +54,7 @@
 #include "golomb.h"
 #include "hpeldsp.h"
 #include "rectangle.h"
+#include "tpeldsp.h"
 
 #if CONFIG_ZLIB
 #include <zlib.h>
@@ -70,6 +71,7 @@
 typedef struct {
     H264Context h;
     HpelDSPContext hdsp;
+    TpelDSPContext tdsp;
     H264Picture *cur_pic;
     H264Picture *next_pic;
     H264Picture *last_pic;
@@ -321,9 +323,9 @@ static inline void svq3_mc_dir_part(SVQ3Context *s,
         src = h->edge_emu_buffer;
     }
     if (thirdpel)
-        (avg ? h->dsp.avg_tpel_pixels_tab
-             : h->dsp.put_tpel_pixels_tab)[dxy](dest, src, h->linesize,
-                                                width, height);
+        (avg ? s->tdsp.avg_tpel_pixels_tab
+             : s->tdsp.put_tpel_pixels_tab)[dxy](dest, src, h->linesize,
+                                                 width, height);
     else
         (avg ? s->hdsp.avg_pixels_tab
              : s->hdsp.put_pixels_tab)[blocksize][dxy](dest, src, h->linesize,
@@ -349,10 +351,10 @@ static inline void svq3_mc_dir_part(SVQ3Context *s,
                 src = h->edge_emu_buffer;
             }
             if (thirdpel)
-                (avg ? h->dsp.avg_tpel_pixels_tab
-                     : h->dsp.put_tpel_pixels_tab)[dxy](dest, src,
-                                                        h->uvlinesize,
-                                                        width, height);
+                (avg ? s->tdsp.avg_tpel_pixels_tab
+                     : s->tdsp.put_tpel_pixels_tab)[dxy](dest, src,
+                                                         h->uvlinesize,
+                                                         width, height);
             else
                 (avg ? s->hdsp.avg_pixels_tab
                      : s->hdsp.put_pixels_tab)[blocksize][dxy](dest, src,
@@ -881,6 +883,8 @@ static av_cold int svq3_decode_init(AVCodecContext *avctx)
         return -1;
 
     ff_hpeldsp_init(&s->hdsp, avctx->flags);
+    ff_tpeldsp_init(&s->tdsp);
+
     h->flags           = avctx->flags;
     h->is_complex      = 1;
     h->picture_structure = PICT_FRAME;
diff --git a/libavcodec/tpel_template.c b/libavcodec/tpel_template.c
new file mode 100644
index 0000000000..f07679a88f
--- /dev/null
+++ b/libavcodec/tpel_template.c
@@ -0,0 +1,80 @@
+/*
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include <stddef.h>
+#include <stdint.h>
+
+#include "libavutil/intreadwrite.h"
+#include "pixels.h"
+#include "rnd_avg.h"
+
+#include "bit_depth_template.c"
+
+#define DEF_TPEL(OPNAME, OP)                                            \
+static inline void FUNCC(OPNAME ## _pixels2)(uint8_t *block,            \
+                                             const uint8_t *pixels,     \
+                                             ptrdiff_t line_size,       \
+                                             int h)                     \
+{                                                                       \
+    int i;                                                              \
+    for (i = 0; i < h; i++) {                                           \
+        OP(*((pixel2 *) block), AV_RN2P(pixels));                       \
+        pixels += line_size;                                            \
+        block  += line_size;                                            \
+    }                                                                   \
+}                                                                       \
+                                                                        \
+static inline void FUNCC(OPNAME ## _pixels4)(uint8_t *block,            \
+                                             const uint8_t *pixels,     \
+                                             ptrdiff_t line_size,       \
+                                             int h)                     \
+{                                                                       \
+    int i;                                                              \
+    for (i = 0; i < h; i++) {                                           \
+        OP(*((pixel4 *) block), AV_RN4P(pixels));                       \
+        pixels += line_size;                                            \
+        block  += line_size;                                            \
+    }                                                                   \
+}                                                                       \
+                                                                        \
+static inline void FUNCC(OPNAME ## _pixels8)(uint8_t *block,            \
+                                             const uint8_t *pixels,     \
+                                             ptrdiff_t line_size,       \
+                                             int h)                     \
+{                                                                       \
+    int i;                                                              \
+    for (i = 0; i < h; i++) {                                           \
+        OP(*((pixel4 *) block), AV_RN4P(pixels));                       \
+        OP(*((pixel4 *) (block + 4 * sizeof(pixel))),                   \
+           AV_RN4P(pixels + 4 * sizeof(pixel)));                        \
+        pixels += line_size;                                            \
+        block  += line_size;                                            \
+    }                                                                   \
+}                                                                       \
+                                                                        \
+CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16),                              \
+               FUNCC(OPNAME ## _pixels8),                               \
+               8 * sizeof(pixel))
+
+#define op_avg(a, b) a = rnd_avg_pixel4(a, b)
+#define op_put(a, b) a = b
+
+DEF_TPEL(avg, op_avg)
+DEF_TPEL(put, op_put)
+#undef op_avg
+#undef op_put
diff --git a/libavcodec/tpeldsp.c b/libavcodec/tpeldsp.c
new file mode 100644
index 0000000000..6a1681311a
--- /dev/null
+++ b/libavcodec/tpeldsp.c
@@ -0,0 +1,333 @@
+/*
+ * thirdpel DSP functions
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/**
+ * @file
+ * thirdpel DSP functions
+ */
+
+#include <stdint.h>
+
+#include "libavutil/attributes.h"
+#include "tpeldsp.h"
+
+#define BIT_DEPTH 8
+#include "tpel_template.c"
+
+static inline void put_tpel_pixels_mc00_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    switch (width) {
+    case 2:
+        put_pixels2_8_c(dst, src, stride, height);
+        break;
+    case 4:
+        put_pixels4_8_c(dst, src, stride, height);
+        break;
+    case 8:
+        put_pixels8_8_c(dst, src, stride, height);
+        break;
+    case 16:
+        put_pixels16_8_c(dst, src, stride, height);
+        break;
+    }
+}
+
+static inline void put_tpel_pixels_mc10_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = ((2 * src[j] + src[j + 1] + 1) *
+                      683) >> 11;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void put_tpel_pixels_mc20_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = ((src[j] + 2 * src[j + 1] + 1) *
+                      683) >> 11;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void put_tpel_pixels_mc01_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = ((2 * src[j] + src[j + stride] + 1) *
+                      683) >> 11;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void put_tpel_pixels_mc11_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = ((4 * src[j]          + 3 * src[j + 1] +
+                       3 * src[j + stride] + 2 * src[j + stride + 1] + 6) *
+                      2731) >> 15;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void put_tpel_pixels_mc12_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = ((3 * src[j]          + 2 * src[j + 1] +
+                       4 * src[j + stride] + 3 * src[j + stride + 1] + 6) *
+                      2731) >> 15;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void put_tpel_pixels_mc02_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = ((src[j] + 2 * src[j + stride] + 1) *
+                      683) >> 11;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void put_tpel_pixels_mc21_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = ((3 * src[j]          + 4 * src[j + 1] +
+                       2 * src[j + stride] + 3 * src[j + stride + 1] + 6) *
+                      2731) >> 15;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void put_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = ((2 * src[j]          + 3 * src[j + 1] +
+                       3 * src[j + stride] + 4 * src[j + stride + 1] + 6) *
+                      2731) >> 15;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void avg_tpel_pixels_mc00_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    switch (width) {
+    case 2:
+        avg_pixels2_8_c(dst, src, stride, height);
+        break;
+    case 4:
+        avg_pixels4_8_c(dst, src, stride, height);
+        break;
+    case 8:
+        avg_pixels8_8_c(dst, src, stride, height);
+        break;
+    case 16:
+        avg_pixels16_8_c(dst, src, stride, height);
+        break;
+    }
+}
+
+static inline void avg_tpel_pixels_mc10_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = (dst[j] +
+                      (((2 * src[j] + src[j + 1] + 1) *
+                        683) >> 11) + 1) >> 1;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void avg_tpel_pixels_mc20_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = (dst[j] +
+                      (((src[j] + 2 * src[j + 1] + 1) *
+                        683) >> 11) + 1) >> 1;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void avg_tpel_pixels_mc01_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = (dst[j] +
+                      (((2 * src[j] + src[j + stride] + 1) *
+                        683) >> 11) + 1) >> 1;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void avg_tpel_pixels_mc11_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = (dst[j] +
+                      (((4 * src[j]          + 3 * src[j + 1] +
+                         3 * src[j + stride] + 2 * src[j + stride + 1] + 6) *
+                        2731) >> 15) + 1) >> 1;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void avg_tpel_pixels_mc12_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = (dst[j] +
+                      (((3 * src[j]          + 2 * src[j + 1] +
+                         4 * src[j + stride] + 3 * src[j + stride + 1] + 6) *
+                        2731) >> 15) + 1) >> 1;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void avg_tpel_pixels_mc02_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = (dst[j] +
+                      (((src[j] + 2 * src[j + stride] + 1) *
+                        683) >> 11) + 1) >> 1;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void avg_tpel_pixels_mc21_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = (dst[j] +
+                      (((3 * src[j]          + 4 * src[j + 1] +
+                         2 * src[j + stride] + 3 * src[j + stride + 1] + 6) *
+                        2731) >> 15) + 1) >> 1;
+        src += stride;
+        dst += stride;
+    }
+}
+
+static inline void avg_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src,
+                                          int stride, int width, int height)
+{
+    int i, j;
+
+    for (i = 0; i < height; i++) {
+        for (j = 0; j < width; j++)
+            dst[j] = (dst[j] +
+                      (((2 * src[j]          + 3 * src[j + 1] +
+                         3 * src[j + stride] + 4 * src[j + stride + 1] + 6) *
+                        2731) >> 15) + 1) >> 1;
+        src += stride;
+        dst += stride;
+    }
+}
+
+av_cold void ff_tpeldsp_init(TpelDSPContext *c)
+{
+    c->put_tpel_pixels_tab[ 0] = put_tpel_pixels_mc00_c;
+    c->put_tpel_pixels_tab[ 1] = put_tpel_pixels_mc10_c;
+    c->put_tpel_pixels_tab[ 2] = put_tpel_pixels_mc20_c;
+    c->put_tpel_pixels_tab[ 4] = put_tpel_pixels_mc01_c;
+    c->put_tpel_pixels_tab[ 5] = put_tpel_pixels_mc11_c;
+    c->put_tpel_pixels_tab[ 6] = put_tpel_pixels_mc21_c;
+    c->put_tpel_pixels_tab[ 8] = put_tpel_pixels_mc02_c;
+    c->put_tpel_pixels_tab[ 9] = put_tpel_pixels_mc12_c;
+    c->put_tpel_pixels_tab[10] = put_tpel_pixels_mc22_c;
+
+    c->avg_tpel_pixels_tab[ 0] = avg_tpel_pixels_mc00_c;
+    c->avg_tpel_pixels_tab[ 1] = avg_tpel_pixels_mc10_c;
+    c->avg_tpel_pixels_tab[ 2] = avg_tpel_pixels_mc20_c;
+    c->avg_tpel_pixels_tab[ 4] = avg_tpel_pixels_mc01_c;
+    c->avg_tpel_pixels_tab[ 5] = avg_tpel_pixels_mc11_c;
+    c->avg_tpel_pixels_tab[ 6] = avg_tpel_pixels_mc21_c;
+    c->avg_tpel_pixels_tab[ 8] = avg_tpel_pixels_mc02_c;
+    c->avg_tpel_pixels_tab[ 9] = avg_tpel_pixels_mc12_c;
+    c->avg_tpel_pixels_tab[10] = avg_tpel_pixels_mc22_c;
+}
diff --git a/libavcodec/tpeldsp.h b/libavcodec/tpeldsp.h
new file mode 100644
index 0000000000..9c67d60850
--- /dev/null
+++ b/libavcodec/tpeldsp.h
@@ -0,0 +1,59 @@
+/*
+ * thirdpel DSP functions
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/**
+ * @file
+ * thirdpel DSP functions
+ */
+
+#ifndef AVCODEC_TPELDSP_H
+#define AVCODEC_TPELDSP_H
+
+#include <stdint.h>
+
+/* add and put pixel (decoding) */
+// blocksizes for hpel_pixels_func are 8x4,8x8 16x8 16x16
+// h for hpel_pixels_func is limited to {width/2, width} but never larger
+// than 16 and never smaller than 4
+typedef void (*tpel_mc_func)(uint8_t *block /* align width (8 or 16) */,
+                             const uint8_t *pixels /* align 1 */,
+                             int line_size, int w, int h);
+
+/**
+ * thirdpel DSP context
+ */
+typedef struct TpelDSPContext {
+    /**
+     * Thirdpel motion compensation with rounding (a + b + 1) >> 1.
+     * this is an array[12] of motion compensation functions for the
+     * 9 thirdpel positions<br>
+     * *pixels_tab[xthirdpel + 4 * ythirdpel]
+     * @param block destination where the result is stored
+     * @param pixels source
+     * @param line_size number of bytes in a horizontal line of block
+     * @param h height
+     */
+    tpel_mc_func put_tpel_pixels_tab[11]; // FIXME individual func ptr per width?
+    tpel_mc_func avg_tpel_pixels_tab[11]; // FIXME individual func ptr per width?
+} TpelDSPContext;
+
+void ff_tpeldsp_init(TpelDSPContext *c);
+
+#endif /* AVCODEC_TPELDSP_H */