Merge remote-tracking branch 'qatar/master'

* qatar/master:
  SBR DSP: fix SSE code to not use SSE2 instructions.
  cpu: initialize mask to -1, so that by default, optimizations are used.
  error_resilience: initialize s->block_index[].
  svq3: protect against negative quantizers.
  Don't use ff_cropTbl[] for IDCT.
  swscale: make filterPos 32bit.
  FATE: add CPUFLAGS variable, mapping to -cpuflags avconv option.
  avconv: add -cpuflags option for setting supported cpuflags.
  cpu: add av_set_cpu_flags_mask().
  libx264: Allow overriding the sliced threads option
  avconv: fix counting encoded video size.

Conflicts:
	doc/APIchanges
	doc/fate.texi
	doc/ffmpeg.texi
	ffmpeg.c
	libavcodec/h264idct_template.c
	libavcodec/svq3.c
	libavutil/avutil.h
	libavutil/cpu.c
	libavutil/cpu.h
	libswscale/swscale.c
	tests/Makefile
	tests/fate-run.sh
	tests/regression-funcs.sh

Merged-by: Michael Niedermayer <michaelni@gmx.at>

30 files changed, 297 insertions, 230 deletions

diff --git a/cmdutils.c b/cmdutils.c
index d0b3f22..7624eaf 100644
--- a/cmdutils.c
+++ b/cmdutils.c
@@ -340,11 +340,8 @@ void parse_options(void *optctx, int argc, char **argv, const OptionDef *options
     }
 }
 
-/*
- * Return index of option opt in argv or 0 if not found.
- */
-static int locate_option(int argc, char **argv, const OptionDef *options,
-                         const char *optname)
+int locate_option(int argc, char **argv, const OptionDef *options,
+                  const char *optname)
 {
     const OptionDef *po;
     int i;
@@ -537,13 +534,54 @@ int opt_max_alloc(const char *opt, const char *arg)
 
 int opt_cpuflags(const char *opt, const char *arg)
 {
-    char *tail;
-    long flags = strtol(arg, &tail, 10);
+#define CPUFLAG_MMX2     (AV_CPU_FLAG_MMX      | AV_CPU_FLAG_MMX2)
+#define CPUFLAG_3DNOW    (AV_CPU_FLAG_3DNOW    | AV_CPU_FLAG_MMX)
+#define CPUFLAG_3DNOWEXT (AV_CPU_FLAG_3DNOWEXT | CPUFLAG_3DNOW)
+#define CPUFLAG_SSE      (AV_CPU_FLAG_SSE      | CPUFLAG_MMX2)
+#define CPUFLAG_SSE2     (AV_CPU_FLAG_SSE2     | CPUFLAG_SSE)
+#define CPUFLAG_SSE2SLOW (AV_CPU_FLAG_SSE2SLOW | CPUFLAG_SSE2)
+#define CPUFLAG_SSE3     (AV_CPU_FLAG_SSE3     | CPUFLAG_SSE2)
+#define CPUFLAG_SSE3SLOW (AV_CPU_FLAG_SSE3SLOW | CPUFLAG_SSE3)
+#define CPUFLAG_SSSE3    (AV_CPU_FLAG_SSSE3    | CPUFLAG_SSE3)
+#define CPUFLAG_SSE4     (AV_CPU_FLAG_SSE4     | CPUFLAG_SSSE3)
+#define CPUFLAG_SSE42    (AV_CPU_FLAG_SSE42    | CPUFLAG_SSE4)
+#define CPUFLAG_AVX      (AV_CPU_FLAG_AVX      | CPUFLAG_SSE42)
+#define CPUFLAG_XOP      (AV_CPU_FLAG_XOP      | CPUFLAG_AVX)
+#define CPUFLAG_FMA4     (AV_CPU_FLAG_FMA4     | CPUFLAG_AVX)
+    static const AVOption cpuflags_opts[] = {
+        { "flags"   , NULL, 0, AV_OPT_TYPE_FLAGS, { 0 }, INT64_MIN, INT64_MAX, .unit = "flags" },
+        { "altivec" , NULL, 0, AV_OPT_TYPE_CONST, { AV_CPU_FLAG_ALTIVEC  },    .unit = "flags" },
+        { "mmx"     , NULL, 0, AV_OPT_TYPE_CONST, { AV_CPU_FLAG_MMX      },    .unit = "flags" },
+        { "mmx2"    , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_MMX2         },    .unit = "flags" },
+        { "sse"     , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE          },    .unit = "flags" },
+        { "sse2"    , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE2         },    .unit = "flags" },
+        { "sse2slow", NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE2SLOW     },    .unit = "flags" },
+        { "sse3"    , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE3         },    .unit = "flags" },
+        { "sse3slow", NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE3SLOW     },    .unit = "flags" },
+        { "ssse3"   , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSSE3        },    .unit = "flags" },
+        { "atom"    , NULL, 0, AV_OPT_TYPE_CONST, { AV_CPU_FLAG_ATOM     },    .unit = "flags" },
+        { "sse4.1"  , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE4         },    .unit = "flags" },
+        { "sse4.2"  , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE42        },    .unit = "flags" },
+        { "avx"     , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_AVX          },    .unit = "flags" },
+        { "xop"     , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_XOP          },    .unit = "flags" },
+        { "fma4"    , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_FMA4         },    .unit = "flags" },
+        { "3dnow"   , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_3DNOW        },    .unit = "flags" },
+        { "3dnowext", NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_3DNOWEXT     },    .unit = "flags" },
+        { NULL },
+    };
+    static const AVClass class = {
+        .class_name = "cpuflags",
+        .item_name  = av_default_item_name,
+        .option     = cpuflags_opts,
+        .version    = LIBAVUTIL_VERSION_INT,
+    };
+    int flags = av_get_cpu_flags();
+    int ret;
+    const AVClass *pclass = &class;
+
+    if ((ret = av_opt_eval_flags(&pclass, &cpuflags_opts[0], arg, &flags)) < 0)
+        return ret;
 
-    if (*tail) {
-        av_log(NULL, AV_LOG_FATAL, "Invalid cpuflags \"%s\".\n", arg);
-        exit_program(1);
-    }
     av_force_cpu_flags(flags);
     return 0;
 }
diff --git a/cmdutils.h b/cmdutils.h
index ae7a87a..26cd96a 100644
--- a/cmdutils.h
+++ b/cmdutils.h
@@ -207,6 +207,12 @@ int parse_option(void *optctx, const char *opt, const char *arg,
 void parse_loglevel(int argc, char **argv, const OptionDef *options);
 
 /**
+ * Return index of option opt in argv or 0 if not found.
+ */
+int locate_option(int argc, char **argv, const OptionDef *options,
+                  const char *optname);
+
+/**
  * Check if the given stream matches a stream specifier.
  *
  * @param s  Corresponding format context.
diff --git a/cmdutils_common_opts.h b/cmdutils_common_opts.h
index 044f63b..334dac9 100644
--- a/cmdutils_common_opts.h
+++ b/cmdutils_common_opts.h
@@ -16,4 +16,4 @@
     { "debug", HAS_ARG, {(void*)opt_codec_debug}, "set debug flags", "flags" },
     { "report", 0, {(void*)opt_report}, "generate a report" },
     { "max_alloc", HAS_ARG, {(void*)opt_max_alloc}, "set maximum size of a single allocated block", "bytes" },
-    { "cpuflags", HAS_ARG, {(void*)opt_cpuflags}, "force specific cpu flags", "flags" },
+    { "cpuflags", HAS_ARG | OPT_EXPERT, {(void*)opt_cpuflags}, "force specific cpu flags", "flags" },
diff --git a/doc/avtools-common-opts.texi b/doc/avtools-common-opts.texi
index 94d47fd..78ac92c 100644
--- a/doc/avtools-common-opts.texi
+++ b/doc/avtools-common-opts.texi
@@ -134,6 +134,10 @@ It also implies @code{-loglevel verbose}.
 Note: setting the environment variable @code{FFREPORT} to any value has the
 same effect.
 
+@item -cpuflags flags (@emph{global})
+Allows setting and clearing cpu flags. This option is intended
+for testing. Do not use it unless you know what you're doing.
+
 @end table
 
 @section AVOptions
diff --git a/doc/fate.texi b/doc/fate.texi
index 1612bdf..07d8c53 100644
--- a/doc/fate.texi
+++ b/doc/fate.texi
@@ -166,9 +166,11 @@ the synchronisation of the samples directory.
 @item THREADS
     Specify how many threads to use while running regression tests, it is
     quite useful to detect thread-related regressions.
+@item CPUFLAGS
+    Specify CPU flags.
 @end table
 
 Example:
 @example
-make V=1 SAMPLES=/var/fate/samples THREADS=2 fate
-@end example
\ No newline at end of file
+make V=1 SAMPLES=/var/fate/samples THREADS=2 CPUFLAGS=mmx fate
+@end example
diff --git a/ffmpeg.c b/ffmpeg.c
index 7afec4c..5499d5d 100644
--- a/ffmpeg.c
+++ b/ffmpeg.c
@@ -4963,6 +4963,13 @@ static int opt_deinterlace(const char *opt, const char *arg)
     return 0;
 }
 
+static void parse_cpuflags(int argc, char **argv, const OptionDef *options)
+{
+    int idx = locate_option(argc, argv, options, "cpuflags");
+    if (idx && argv[idx + 1])
+        opt_cpuflags("cpuflags", argv[idx + 1]);
+}
+
 #define OFFSET(x) offsetof(OptionsContext, x)
 static const OptionDef options[] = {
     /* main options */
@@ -5136,6 +5143,8 @@ int main(int argc, char **argv)
 
     term_init();
 
+    parse_cpuflags(argc, argv, options);
+
     /* parse options */
     parse_options(&o, argc, argv, options, opt_output_file);
 
diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c
index 9821703..7379fa2 100644
--- a/libavcodec/dsputil.c
+++ b/libavcodec/dsputil.c
@@ -367,18 +367,17 @@ void ff_put_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels,
                              int line_size)
 {
     int i;
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     /* read the pixels */
     for(i=0;i<8;i++) {
-        pixels[0] = cm[block[0]];
-        pixels[1] = cm[block[1]];
-        pixels[2] = cm[block[2]];
-        pixels[3] = cm[block[3]];
-        pixels[4] = cm[block[4]];
-        pixels[5] = cm[block[5]];
-        pixels[6] = cm[block[6]];
-        pixels[7] = cm[block[7]];
+        pixels[0] = av_clip_uint8(block[0]);
+        pixels[1] = av_clip_uint8(block[1]);
+        pixels[2] = av_clip_uint8(block[2]);
+        pixels[3] = av_clip_uint8(block[3]);
+        pixels[4] = av_clip_uint8(block[4]);
+        pixels[5] = av_clip_uint8(block[5]);
+        pixels[6] = av_clip_uint8(block[6]);
+        pixels[7] = av_clip_uint8(block[7]);
 
         pixels += line_size;
         block += 8;
@@ -389,14 +388,13 @@ static void put_pixels_clamped4_c(const DCTELEM *block, uint8_t *restrict pixels
                                  int line_size)
 {
     int i;
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     /* read the pixels */
     for(i=0;i<4;i++) {
-        pixels[0] = cm[block[0]];
-        pixels[1] = cm[block[1]];
-        pixels[2] = cm[block[2]];
-        pixels[3] = cm[block[3]];
+        pixels[0] = av_clip_uint8(block[0]);
+        pixels[1] = av_clip_uint8(block[1]);
+        pixels[2] = av_clip_uint8(block[2]);
+        pixels[3] = av_clip_uint8(block[3]);
 
         pixels += line_size;
         block += 8;
@@ -407,12 +405,11 @@ static void put_pixels_clamped2_c(const DCTELEM *block, uint8_t *restrict pixels
                                  int line_size)
 {
     int i;
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     /* read the pixels */
     for(i=0;i<2;i++) {
-        pixels[0] = cm[block[0]];
-        pixels[1] = cm[block[1]];
+        pixels[0] = av_clip_uint8(block[0]);
+        pixels[1] = av_clip_uint8(block[1]);
 
         pixels += line_size;
         block += 8;
@@ -444,18 +441,17 @@ void ff_add_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels,
                              int line_size)
 {
     int i;
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     /* read the pixels */
     for(i=0;i<8;i++) {
-        pixels[0] = cm[pixels[0] + block[0]];
-        pixels[1] = cm[pixels[1] + block[1]];
-        pixels[2] = cm[pixels[2] + block[2]];
-        pixels[3] = cm[pixels[3] + block[3]];
-        pixels[4] = cm[pixels[4] + block[4]];
-        pixels[5] = cm[pixels[5] + block[5]];
-        pixels[6] = cm[pixels[6] + block[6]];
-        pixels[7] = cm[pixels[7] + block[7]];
+        pixels[0] = av_clip_uint8(pixels[0] + block[0]);
+        pixels[1] = av_clip_uint8(pixels[1] + block[1]);
+        pixels[2] = av_clip_uint8(pixels[2] + block[2]);
+        pixels[3] = av_clip_uint8(pixels[3] + block[3]);
+        pixels[4] = av_clip_uint8(pixels[4] + block[4]);
+        pixels[5] = av_clip_uint8(pixels[5] + block[5]);
+        pixels[6] = av_clip_uint8(pixels[6] + block[6]);
+        pixels[7] = av_clip_uint8(pixels[7] + block[7]);
         pixels += line_size;
         block += 8;
     }
@@ -465,14 +461,13 @@ static void add_pixels_clamped4_c(const DCTELEM *block, uint8_t *restrict pixels
                           int line_size)
 {
     int i;
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     /* read the pixels */
     for(i=0;i<4;i++) {
-        pixels[0] = cm[pixels[0] + block[0]];
-        pixels[1] = cm[pixels[1] + block[1]];
-        pixels[2] = cm[pixels[2] + block[2]];
-        pixels[3] = cm[pixels[3] + block[3]];
+        pixels[0] = av_clip_uint8(pixels[0] + block[0]);
+        pixels[1] = av_clip_uint8(pixels[1] + block[1]);
+        pixels[2] = av_clip_uint8(pixels[2] + block[2]);
+        pixels[3] = av_clip_uint8(pixels[3] + block[3]);
         pixels += line_size;
         block += 8;
     }
@@ -482,12 +477,11 @@ static void add_pixels_clamped2_c(const DCTELEM *block, uint8_t *restrict pixels
                           int line_size)
 {
     int i;
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     /* read the pixels */
     for(i=0;i<2;i++) {
-        pixels[0] = cm[pixels[0] + block[0]];
-        pixels[1] = cm[pixels[1] + block[1]];
+        pixels[0] = av_clip_uint8(pixels[0] + block[0]);
+        pixels[1] = av_clip_uint8(pixels[1] + block[1]);
         pixels += line_size;
         block += 8;
     }
@@ -2779,15 +2773,11 @@ static void ff_jref_idct2_add(uint8_t *dest, int line_size, DCTELEM *block)
 
 static void ff_jref_idct1_put(uint8_t *dest, int line_size, DCTELEM *block)
 {
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
-
-    dest[0] = cm[(block[0] + 4)>>3];
+    dest[0] = av_clip_uint8((block[0] + 4)>>3);
 }
 static void ff_jref_idct1_add(uint8_t *dest, int line_size, DCTELEM *block)
 {
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
-
-    dest[0] = cm[dest[0] + ((block[0] + 4)>>3)];
+    dest[0] = av_clip_uint8(dest[0] + ((block[0] + 4)>>3));
 }
 
 static void just_return(void *mem av_unused, int stride av_unused, int h av_unused) { return; }
diff --git a/libavcodec/error_resilience.c b/libavcodec/error_resilience.c
index 1c1420f..3582558 100644
--- a/libavcodec/error_resilience.c
+++ b/libavcodec/error_resilience.c
@@ -440,9 +440,14 @@ static void guess_mv(MpegEncContext *s)
     if ((!(s->avctx->error_concealment&FF_EC_GUESS_MVS)) ||
         num_avail <= mb_width / 2) {
         for (mb_y = 0; mb_y < s->mb_height; mb_y++) {
+            s->mb_x = 0;
+            s->mb_y = mb_y;
+            ff_init_block_index(s);
             for (mb_x = 0; mb_x < s->mb_width; mb_x++) {
                 const int mb_xy = mb_x + mb_y * s->mb_stride;
 
+                ff_update_block_index(s);
+
                 if (IS_INTRA(s->current_picture.f.mb_type[mb_xy]))
                     continue;
                 if (!(s->error_status_table[mb_xy] & ER_MV_ERROR))
@@ -477,6 +482,9 @@ static void guess_mv(MpegEncContext *s)
 
             changed = 0;
             for (mb_y = 0; mb_y < s->mb_height; mb_y++) {
+                s->mb_x = 0;
+                s->mb_y = mb_y;
+                ff_init_block_index(s);
                 for (mb_x = 0; mb_x < s->mb_width; mb_x++) {
                     const int mb_xy        = mb_x + mb_y * s->mb_stride;
                     int mv_predictor[8][2] = { { 0 } };
@@ -488,6 +496,8 @@ static void guess_mv(MpegEncContext *s)
                     const int mot_index    = (mb_x + mb_y * mot_stride) * mot_step;
                     int prev_x, prev_y, prev_ref;
 
+                    ff_update_block_index(s);
+
                     if ((mb_x ^ mb_y ^ pass) & 1)
                         continue;
 
@@ -1098,11 +1108,16 @@ void ff_er_frame_end(MpegEncContext *s)
 
     /* handle inter blocks with damaged AC */
     for (mb_y = 0; mb_y < s->mb_height; mb_y++) {
+        s->mb_x = 0;
+        s->mb_y = mb_y;
+        ff_init_block_index(s);
         for (mb_x = 0; mb_x < s->mb_width; mb_x++) {
             const int mb_xy   = mb_x + mb_y * s->mb_stride;
             const int mb_type = s->current_picture.f.mb_type[mb_xy];
             int dir           = !s->last_picture.f.data[0];
 
+            ff_update_block_index(s);
+
             error = s->error_status_table[mb_xy];
 
             if (IS_INTRA(mb_type))
@@ -1140,11 +1155,16 @@ void ff_er_frame_end(MpegEncContext *s)
     /* guess MVs */
     if (s->pict_type == AV_PICTURE_TYPE_B) {
         for (mb_y = 0; mb_y < s->mb_height; mb_y++) {
+            s->mb_x = 0;
+            s->mb_y = mb_y;
+            ff_init_block_index(s);
             for (mb_x = 0; mb_x < s->mb_width; mb_x++) {
                 int       xy      = mb_x * 2 + mb_y * 2 * s->b8_stride;
                 const int mb_xy   = mb_x + mb_y * s->mb_stride;
                 const int mb_type = s->current_picture.f.mb_type[mb_xy];
 
+                ff_update_block_index(s);
+
                 error = s->error_status_table[mb_xy];
 
                 if (IS_INTRA(mb_type))
diff --git a/libavcodec/h264idct_template.c b/libavcodec/h264idct_template.c
index c59976a..5c730fd 100644
--- a/libavcodec/h264idct_template.c
+++ b/libavcodec/h264idct_template.c
@@ -49,7 +49,6 @@ static const uint8_t scan8[16*3]={
 void FUNCC(ff_h264_idct_add)(uint8_t *_dst, DCTELEM *_block, int stride)
 {
     int i;
-    INIT_CLIP
     pixel *dst = (pixel*)_dst;
     dctcoef *block = (dctcoef*)_block;
     stride >>= sizeof(pixel)-1;
@@ -74,16 +73,15 @@ void FUNCC(ff_h264_idct_add)(uint8_t *_dst, DCTELEM *_block, int stride)
         const int z2= (block[1 + 4*i]>>1) -  block[3 + 4*i];
         const int z3=  block[1 + 4*i]     + (block[3 + 4*i]>>1);
 
-        dst[i + 0*stride]= CLIP(dst[i + 0*stride] + ((z0 + z3) >> 6));
-        dst[i + 1*stride]= CLIP(dst[i + 1*stride] + ((z1 + z2) >> 6));
-        dst[i + 2*stride]= CLIP(dst[i + 2*stride] + ((z1 - z2) >> 6));
-        dst[i + 3*stride]= CLIP(dst[i + 3*stride] + ((z0 - z3) >> 6));
+        dst[i + 0*stride]= av_clip_pixel(dst[i + 0*stride] + ((z0 + z3) >> 6));
+        dst[i + 1*stride]= av_clip_pixel(dst[i + 1*stride] + ((z1 + z2) >> 6));
+        dst[i + 2*stride]= av_clip_pixel(dst[i + 2*stride] + ((z1 - z2) >> 6));
+        dst[i + 3*stride]= av_clip_pixel(dst[i + 3*stride] + ((z0 - z3) >> 6));
     }
 }
 
 void FUNCC(ff_h264_idct8_add)(uint8_t *_dst, DCTELEM *_block, int stride){
     int i;
-    INIT_CLIP
     pixel *dst = (pixel*)_dst;
     dctcoef *block = (dctcoef*)_block;
     stride >>= sizeof(pixel)-1;
@@ -143,14 +141,14 @@ void FUNCC(ff_h264_idct8_add)(uint8_t *_dst, DCTELEM *_block, int stride){
         const int b5 = (a3>>2) - a5;
         const int b7 =  a7 - (a1>>2);
 
-        dst[i + 0*stride] = CLIP( dst[i + 0*stride] + ((b0 + b7) >> 6) );
-        dst[i + 1*stride] = CLIP( dst[i + 1*stride] + ((b2 + b5) >> 6) );
-        dst[i + 2*stride] = CLIP( dst[i + 2*stride] + ((b4 + b3) >> 6) );
-        dst[i + 3*stride] = CLIP( dst[i + 3*stride] + ((b6 + b1) >> 6) );
-        dst[i + 4*stride] = CLIP( dst[i + 4*stride] + ((b6 - b1) >> 6) );
-        dst[i + 5*stride] = CLIP( dst[i + 5*stride] + ((b4 - b3) >> 6) );
-        dst[i + 6*stride] = CLIP( dst[i + 6*stride] + ((b2 - b5) >> 6) );
-        dst[i + 7*stride] = CLIP( dst[i + 7*stride] + ((b0 - b7) >> 6) );
+        dst[i + 0*stride] = av_clip_pixel( dst[i + 0*stride] + ((b0 + b7) >> 6) );
+        dst[i + 1*stride] = av_clip_pixel( dst[i + 1*stride] + ((b2 + b5) >> 6) );
+        dst[i + 2*stride] = av_clip_pixel( dst[i + 2*stride] + ((b4 + b3) >> 6) );
+        dst[i + 3*stride] = av_clip_pixel( dst[i + 3*stride] + ((b6 + b1) >> 6) );
+        dst[i + 4*stride] = av_clip_pixel( dst[i + 4*stride] + ((b6 - b1) >> 6) );
+        dst[i + 5*stride] = av_clip_pixel( dst[i + 5*stride] + ((b4 - b3) >> 6) );
+        dst[i + 6*stride] = av_clip_pixel( dst[i + 6*stride] + ((b2 - b5) >> 6) );
+        dst[i + 7*stride] = av_clip_pixel( dst[i + 7*stride] + ((b0 - b7) >> 6) );
     }
 }
 
@@ -158,13 +156,12 @@ void FUNCC(ff_h264_idct8_add)(uint8_t *_dst, DCTELEM *_block, int stride){
 void FUNCC(ff_h264_idct_dc_add)(uint8_t *p_dst, DCTELEM *block, int stride){
     int i, j;
     int dc = (((dctcoef*)block)[0] + 32) >> 6;
-    INIT_CLIP
     pixel *dst = (pixel*)p_dst;
     stride >>= sizeof(pixel)-1;
     for( j = 0; j < 4; j++ )
     {
         for( i = 0; i < 4; i++ )
-            dst[i] = CLIP( dst[i] + dc );
+            dst[i] = av_clip_pixel( dst[i] + dc );
         dst += stride;
     }
 }
@@ -172,13 +169,12 @@ void FUNCC(ff_h264_idct_dc_add)(uint8_t *p_dst, DCTELEM *block, int stride){
 void FUNCC(ff_h264_idct8_dc_add)(uint8_t *p_dst, DCTELEM *block, int stride){
     int i, j;
     int dc = (((dctcoef*)block)[0] + 32) >> 6;
-    INIT_CLIP
     pixel *dst = (pixel*)p_dst;
     stride >>= sizeof(pixel)-1;
     for( j = 0; j < 8; j++ )
     {
         for( i = 0; i < 8; i++ )
-            dst[i] = CLIP( dst[i] + dc );
+            dst[i] = av_clip_pixel( dst[i] + dc );
         dst += stride;
     }
 }
diff --git a/libavcodec/libx264.c b/libavcodec/libx264.c
index 5610038..3d503db 100644
--- a/libavcodec/libx264.c
+++ b/libavcodec/libx264.c
@@ -454,6 +454,8 @@ static av_cold int X264_init(AVCodecContext *avctx)
     x4->params.analyse.b_psnr = avctx->flags & CODEC_FLAG_PSNR;
 
     x4->params.i_threads      = avctx->thread_count;
+    if (avctx->thread_type)
+        x4->params.b_sliced_threads = avctx->thread_type == FF_THREAD_SLICE;
 
     x4->params.b_interlaced   = avctx->flags & CODEC_FLAG_INTERLACED_DCT;
 
@@ -631,6 +633,7 @@ static const AVCodecDefault x264_defaults[] = {
     { "coder",            "-1" },
     { "cmp",              "-1" },
     { "threads",          AV_STRINGIFY(X264_THREADS_AUTO) },
+    { "thread_type",      "0" },
     { NULL },
 };
 
diff --git a/libavcodec/rv34dsp.c b/libavcodec/rv34dsp.c
index 4347999..4145c4d 100644
--- a/libavcodec/rv34dsp.c
+++ b/libavcodec/rv34dsp.c
@@ -102,15 +102,13 @@ static void rv34_inv_transform_noround_c(DCTELEM *block){
 
 static void rv34_idct_dc_add_c(uint8_t *dst, ptrdiff_t stride, int dc)
 {
-    const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
     int i, j;
 
-    cm += (13*13*dc + 0x200) >> 10;
-
+    dc = (13*13*dc + 0x200) >> 10;
     for (i = 0; i < 4; i++)
     {
         for (j = 0; j < 4; j++)
-            dst[j] = cm[ dst[j] ];
+            dst[j] = av_clip_uint8( dst[j] + dc );
 
         dst += stride;
     }
diff --git a/libavcodec/simple_idct.c b/libavcodec/simple_idct.c
index 1bf2d56..2931850 100644
--- a/libavcodec/simple_idct.c
+++ b/libavcodec/simple_idct.c
@@ -132,7 +132,6 @@ void ff_simple_idct248_put(uint8_t *dest, int line_size, DCTELEM *block)
 static inline void idct4col_add(uint8_t *dest, int line_size, const DCTELEM *col)
 {
     int c0, c1, c2, c3, a0, a1, a2, a3;
-    const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     a0 = col[8*0];
     a1 = col[8*1];
@@ -142,13 +141,13 @@ static inline void idct4col_add(uint8_t *dest, int line_size, const DCTELEM *col
     c2 = (a0 - a2)*C3 + (1 << (C_SHIFT - 1));
     c1 = a1 * C1 + a3 * C2;
     c3 = a1 * C2 - a3 * C1;
-    dest[0] = cm[dest[0] + ((c0 + c1) >> C_SHIFT)];
+    dest[0] = av_clip_uint8(dest[0] + ((c0 + c1) >> C_SHIFT));
     dest += line_size;
-    dest[0] = cm[dest[0] + ((c2 + c3) >> C_SHIFT)];
+    dest[0] = av_clip_uint8(dest[0] + ((c2 + c3) >> C_SHIFT));
     dest += line_size;
-    dest[0] = cm[dest[0] + ((c2 - c3) >> C_SHIFT)];
+    dest[0] = av_clip_uint8(dest[0] + ((c2 - c3) >> C_SHIFT));
     dest += line_size;
-    dest[0] = cm[dest[0] + ((c0 - c1) >> C_SHIFT)];
+    dest[0] = av_clip_uint8(dest[0] + ((c0 - c1) >> C_SHIFT));
 }
 
 #define RN_SHIFT 15
@@ -160,7 +159,6 @@ static inline void idct4col_add(uint8_t *dest, int line_size, const DCTELEM *col
 static inline void idct4row(DCTELEM *row)
 {
     int c0, c1, c2, c3, a0, a1, a2, a3;
-    //const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     a0 = row[0];
     a1 = row[1];
diff --git a/libavcodec/simple_idct_template.c b/libavcodec/simple_idct_template.c
index 4bb5136..b67893c 100644
--- a/libavcodec/simple_idct_template.c
+++ b/libavcodec/simple_idct_template.c
@@ -224,50 +224,48 @@ static inline void FUNC(idctSparseColPut)(pixel *dest, int line_size,
                                           DCTELEM *col)
 {
     int a0, a1, a2, a3, b0, b1, b2, b3;
-    INIT_CLIP;
 
     IDCT_COLS;
 
-    dest[0] = CLIP((a0 + b0) >> COL_SHIFT);
+    dest[0] = av_clip_pixel((a0 + b0) >> COL_SHIFT);
     dest += line_size;
-    dest[0] = CLIP((a1 + b1) >> COL_SHIFT);
+    dest[0] = av_clip_pixel((a1 + b1) >> COL_SHIFT);
     dest += line_size;
-    dest[0] = CLIP((a2 + b2) >> COL_SHIFT);
+    dest[0] = av_clip_pixel((a2 + b2) >> COL_SHIFT);
     dest += line_size;
-    dest[0] = CLIP((a3 + b3) >> COL_SHIFT);
+    dest[0] = av_clip_pixel((a3 + b3) >> COL_SHIFT);
     dest += line_size;
-    dest[0] = CLIP((a3 - b3) >> COL_SHIFT);
+    dest[0] = av_clip_pixel((a3 - b3) >> COL_SHIFT);
     dest += line_size;
-    dest[0] = CLIP((a2 - b2) >> COL_SHIFT);
+    dest[0] = av_clip_pixel((a2 - b2) >> COL_SHIFT);
     dest += line_size;
-    dest[0] = CLIP((a1 - b1) >> COL_SHIFT);
+    dest[0] = av_clip_pixel((a1 - b1) >> COL_SHIFT);
     dest += line_size;
-    dest[0] = CLIP((a0 - b0) >> COL_SHIFT);
+    dest[0] = av_clip_pixel((a0 - b0) >> COL_SHIFT);
 }
 
 static inline void FUNC(idctSparseColAdd)(pixel *dest, int line_size,
                                           DCTELEM *col)
 {
     int a0, a1, a2, a3, b0, b1, b2, b3;
-    INIT_CLIP;
 
     IDCT_COLS;
 
-    dest[0] = CLIP(dest[0] + ((a0 + b0) >> COL_SHIFT));
+    dest[0] = av_clip_pixel(dest[0] + ((a0 + b0) >> COL_SHIFT));
     dest += line_size;
-    dest[0] = CLIP(dest[0] + ((a1 + b1) >> COL_SHIFT));
+    dest[0] = av_clip_pixel(dest[0] + ((a1 + b1) >> COL_SHIFT));
     dest += line_size;
-    dest[0] = CLIP(dest[0] + ((a2 + b2) >> COL_SHIFT));
+    dest[0] = av_clip_pixel(dest[0] + ((a2 + b2) >> COL_SHIFT));
     dest += line_size;
-    dest[0] = CLIP(dest[0] + ((a3 + b3) >> COL_SHIFT));
+    dest[0] = av_clip_pixel(dest[0] + ((a3 + b3) >> COL_SHIFT));
     dest += line_size;
-    dest[0] = CLIP(dest[0] + ((a3 - b3) >> COL_SHIFT));
+    dest[0] = av_clip_pixel(dest[0] + ((a3 - b3) >> COL_SHIFT));
     dest += line_size;
-    dest[0] = CLIP(dest[0] + ((a2 - b2) >> COL_SHIFT));
+    dest[0] = av_clip_pixel(dest[0] + ((a2 - b2) >> COL_SHIFT));
     dest += line_size;
-    dest[0] = CLIP(dest[0] + ((a1 - b1) >> COL_SHIFT));
+    dest[0] = av_clip_pixel(dest[0] + ((a1 - b1) >> COL_SHIFT));
     dest += line_size;
-    dest[0] = CLIP(dest[0] + ((a0 - b0) >> COL_SHIFT));
+    dest[0] = av_clip_pixel(dest[0] + ((a0 - b0) >> COL_SHIFT));
 }
 
 static inline void FUNC(idctSparseCol)(DCTELEM *col)
diff --git a/libavcodec/vc1dsp.c b/libavcodec/vc1dsp.c
index 91578a7..da7593f 100644
--- a/libavcodec/vc1dsp.c
+++ b/libavcodec/vc1dsp.c
@@ -139,8 +139,6 @@ static void vc1_h_s_overlap_c(DCTELEM *left, DCTELEM *right)
  * @see 8.6
  */
 static av_always_inline int vc1_filter_line(uint8_t* src, int stride, int pq){
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
-
     int a0 = (2*(src[-2*stride] - src[ 1*stride]) - 5*(src[-1*stride] - src[ 0*stride]) + 4) >> 3;
     int a0_sign = a0 >> 31;        /* Store sign */
     a0 = (a0 ^ a0_sign) - a0_sign; /* a0 = FFABS(a0); */
@@ -163,8 +161,8 @@ static av_always_inline int vc1_filter_line(uint8_t* src, int stride, int pq){
                 else{
                     d = FFMIN(d, clip);
                     d = (d ^ d_sign) - d_sign;          /* Restore sign */
-                    src[-1*stride] = cm[src[-1*stride] - d];
-                    src[ 0*stride] = cm[src[ 0*stride] + d];
+                    src[-1*stride] = av_clip_uint8(src[-1*stride] - d);
+                    src[ 0*stride] = av_clip_uint8(src[ 0*stride] + d);
                 }
                 return 1;
             }
@@ -234,19 +232,17 @@ static void vc1_inv_trans_8x8_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
 {
     int i;
     int dc = block[0];
-    const uint8_t *cm;
     dc = (3 * dc +  1) >> 1;
     dc = (3 * dc + 16) >> 5;
-    cm = ff_cropTbl + MAX_NEG_CROP + dc;
     for(i = 0; i < 8; i++){
-        dest[0] = cm[dest[0]];
-        dest[1] = cm[dest[1]];
-        dest[2] = cm[dest[2]];
-        dest[3] = cm[dest[3]];
-        dest[4] = cm[dest[4]];
-        dest[5] = cm[dest[5]];
-        dest[6] = cm[dest[6]];
-        dest[7] = cm[dest[7]];
+        dest[0] = av_clip_uint8(dest[0] + dc);
+        dest[1] = av_clip_uint8(dest[1] + dc);
+        dest[2] = av_clip_uint8(dest[2] + dc);
+        dest[3] = av_clip_uint8(dest[3] + dc);
+        dest[4] = av_clip_uint8(dest[4] + dc);
+        dest[5] = av_clip_uint8(dest[5] + dc);
+        dest[6] = av_clip_uint8(dest[6] + dc);
+        dest[7] = av_clip_uint8(dest[7] + dc);
         dest += linesize;
     }
 }
@@ -326,19 +322,17 @@ static void vc1_inv_trans_8x4_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
 {
     int i;
     int dc = block[0];
-    const uint8_t *cm;
     dc = ( 3 * dc +  1) >> 1;
     dc = (17 * dc + 64) >> 7;
-    cm = ff_cropTbl + MAX_NEG_CROP + dc;
     for(i = 0; i < 4; i++){
-        dest[0] = cm[dest[0]];
-        dest[1] = cm[dest[1]];
-        dest[2] = cm[dest[2]];
-        dest[3] = cm[dest[3]];
-        dest[4] = cm[dest[4]];
-        dest[5] = cm[dest[5]];
-        dest[6] = cm[dest[6]];
-        dest[7] = cm[dest[7]];
+        dest[0] = av_clip_uint8(dest[0] + dc);
+        dest[1] = av_clip_uint8(dest[1] + dc);
+        dest[2] = av_clip_uint8(dest[2] + dc);
+        dest[3] = av_clip_uint8(dest[3] + dc);
+        dest[4] = av_clip_uint8(dest[4] + dc);
+        dest[5] = av_clip_uint8(dest[5] + dc);
+        dest[6] = av_clip_uint8(dest[6] + dc);
+        dest[7] = av_clip_uint8(dest[7] + dc);
         dest += linesize;
     }
 }
@@ -403,15 +397,13 @@ static void vc1_inv_trans_4x8_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
 {
     int i;
     int dc = block[0];
-    const uint8_t *cm;
     dc = (17 * dc +  4) >> 3;
     dc = (12 * dc + 64) >> 7;
-    cm = ff_cropTbl + MAX_NEG_CROP + dc;
     for(i = 0; i < 8; i++){
-        dest[0] = cm[dest[0]];
-        dest[1] = cm[dest[1]];
-        dest[2] = cm[dest[2]];
-        dest[3] = cm[dest[3]];
+        dest[0] = av_clip_uint8(dest[0] + dc);
+        dest[1] = av_clip_uint8(dest[1] + dc);
+        dest[2] = av_clip_uint8(dest[2] + dc);
+        dest[3] = av_clip_uint8(dest[3] + dc);
         dest += linesize;
     }
 }
@@ -476,15 +468,13 @@ static void vc1_inv_trans_4x4_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
 {
     int i;
     int dc = block[0];
-    const uint8_t *cm;
     dc = (17 * dc +  4) >> 3;
     dc = (17 * dc + 64) >> 7;
-    cm = ff_cropTbl + MAX_NEG_CROP + dc;
     for(i = 0; i < 4; i++){
-        dest[0] = cm[dest[0]];
-        dest[1] = cm[dest[1]];
-        dest[2] = cm[dest[2]];
-        dest[3] = cm[dest[3]];
+        dest[0] = av_clip_uint8(dest[0] + dc);
+        dest[1] = av_clip_uint8(dest[1] + dc);
+        dest[2] = av_clip_uint8(dest[2] + dc);
+        dest[3] = av_clip_uint8(dest[3] + dc);
         dest += linesize;
     }
 }
diff --git a/libavcodec/vp3dsp.c b/libavcodec/vp3dsp.c
index 94efa3b..9fded0f 100644
--- a/libavcodec/vp3dsp.c
+++ b/libavcodec/vp3dsp.c
@@ -41,7 +41,6 @@
 static av_always_inline void idct(uint8_t *dst, int stride, int16_t *input, int type)
 {
     int16_t *ip = input;
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     int A, B, C, D, Ad, Bd, Cd, Dd, E, F, G, H;
     int Ed, Gd, Add, Bdd, Fd, Hd;
@@ -147,29 +146,29 @@ static av_always_inline void idct(uint8_t *dst, int stride, int16_t *input, int
                 ip[5*8] = (Fd + Bdd ) >> 4;
                 ip[6*8] = (Fd - Bdd ) >> 4;
             }else if(type==1){
-                dst[0*stride] = cm[(Gd + Cd )  >> 4];
-                dst[7*stride] = cm[(Gd - Cd )  >> 4];
+                dst[0*stride] = av_clip_uint8((Gd + Cd )  >> 4);
+                dst[7*stride] = av_clip_uint8((Gd - Cd )  >> 4);
 
-                dst[1*stride] = cm[(Add + Hd ) >> 4];
-                dst[2*stride] = cm[(Add - Hd ) >> 4];
+                dst[1*stride] = av_clip_uint8((Add + Hd ) >> 4);
+                dst[2*stride] = av_clip_uint8((Add - Hd ) >> 4);
 
-                dst[3*stride] = cm[(Ed + Dd )  >> 4];
-                dst[4*stride] = cm[(Ed - Dd )  >> 4];
+                dst[3*stride] = av_clip_uint8((Ed + Dd )  >> 4);
+                dst[4*stride] = av_clip_uint8((Ed - Dd )  >> 4);
 
-                dst[5*stride] = cm[(Fd + Bdd ) >> 4];
-                dst[6*stride] = cm[(Fd - Bdd ) >> 4];
+                dst[5*stride] = av_clip_uint8((Fd + Bdd ) >> 4);
+                dst[6*stride] = av_clip_uint8((Fd - Bdd ) >> 4);
             }else{
-                dst[0*stride] = cm[dst[0*stride] + ((Gd + Cd )  >> 4)];
-                dst[7*stride] = cm[dst[7*stride] + ((Gd - Cd )  >> 4)];
+                dst[0*stride] = av_clip_uint8(dst[0*stride] + ((Gd + Cd )  >> 4));
+                dst[7*stride] = av_clip_uint8(dst[7*stride] + ((Gd - Cd )  >> 4));
 
-                dst[1*stride] = cm[dst[1*stride] + ((Add + Hd ) >> 4)];
-                dst[2*stride] = cm[dst[2*stride] + ((Add - Hd ) >> 4)];
+                dst[1*stride] = av_clip_uint8(dst[1*stride] + ((Add + Hd ) >> 4));
+                dst[2*stride] = av_clip_uint8(dst[2*stride] + ((Add - Hd ) >> 4));
 
-                dst[3*stride] = cm[dst[3*stride] + ((Ed + Dd )  >> 4)];
-                dst[4*stride] = cm[dst[4*stride] + ((Ed - Dd )  >> 4)];
+                dst[3*stride] = av_clip_uint8(dst[3*stride] + ((Ed + Dd )  >> 4));
+                dst[4*stride] = av_clip_uint8(dst[4*stride] + ((Ed - Dd )  >> 4));
 
-                dst[5*stride] = cm[dst[5*stride] + ((Fd + Bdd ) >> 4)];
-                dst[6*stride] = cm[dst[6*stride] + ((Fd - Bdd ) >> 4)];
+                dst[5*stride] = av_clip_uint8(dst[5*stride] + ((Fd + Bdd ) >> 4));
+                dst[6*stride] = av_clip_uint8(dst[6*stride] + ((Fd - Bdd ) >> 4));
             }
 
         } else {
@@ -190,18 +189,18 @@ static av_always_inline void idct(uint8_t *dst, int stride, int16_t *input, int
                 dst[4*stride]=
                 dst[5*stride]=
                 dst[6*stride]=
-                dst[7*stride]= cm[128 + ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20)];
+                dst[7*stride]= av_clip_uint8(128 + ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20));
             }else{
                 if(ip[0*8]){
                     int v= ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
-                    dst[0*stride] = cm[dst[0*stride] + v];
-                    dst[1*stride] = cm[dst[1*stride] + v];
-                    dst[2*stride] = cm[dst[2*stride] + v];
-                    dst[3*stride] = cm[dst[3*stride] + v];
-                    dst[4*stride] = cm[dst[4*stride] + v];
-                    dst[5*stride] = cm[dst[5*stride] + v];
-                    dst[6*stride] = cm[dst[6*stride] + v];
-                    dst[7*stride] = cm[dst[7*stride] + v];
+                    dst[0*stride] = av_clip_uint8(dst[0*stride] + v);
+                    dst[1*stride] = av_clip_uint8(dst[1*stride] + v);
+                    dst[2*stride] = av_clip_uint8(dst[2*stride] + v);
+                    dst[3*stride] = av_clip_uint8(dst[3*stride] + v);
+                    dst[4*stride] = av_clip_uint8(dst[4*stride] + v);
+                    dst[5*stride] = av_clip_uint8(dst[5*stride] + v);
+                    dst[6*stride] = av_clip_uint8(dst[6*stride] + v);
+                    dst[7*stride] = av_clip_uint8(dst[7*stride] + v);
                 }
             }
         }
@@ -225,17 +224,16 @@ void ff_vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*
 
 void ff_vp3_idct_dc_add_c(uint8_t *dest/*align 8*/, int line_size, const DCTELEM *block/*align 16*/){
     int i, dc = (block[0] + 15) >> 5;
-    const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP + dc;
 
     for(i = 0; i < 8; i++){
-        dest[0] = cm[dest[0]];
-        dest[1] = cm[dest[1]];
-        dest[2] = cm[dest[2]];
-        dest[3] = cm[dest[3]];
-        dest[4] = cm[dest[4]];
-        dest[5] = cm[dest[5]];
-        dest[6] = cm[dest[6]];
-        dest[7] = cm[dest[7]];
+        dest[0] = av_clip_uint8(dest[0] + dc);
+        dest[1] = av_clip_uint8(dest[1] + dc);
+        dest[2] = av_clip_uint8(dest[2] + dc);
+        dest[3] = av_clip_uint8(dest[3] + dc);
+        dest[4] = av_clip_uint8(dest[4] + dc);
+        dest[5] = av_clip_uint8(dest[5] + dc);
+        dest[6] = av_clip_uint8(dest[6] + dc);
+        dest[7] = av_clip_uint8(dest[7] + dc);
         dest += line_size;
     }
 }
diff --git a/libavcodec/vp8dsp.c b/libavcodec/vp8dsp.c
index 12f6988..1ee070c 100644
--- a/libavcodec/vp8dsp.c
+++ b/libavcodec/vp8dsp.c
@@ -80,7 +80,6 @@ static void vp8_luma_dc_wht_dc_c(DCTELEM block[4][4][16], DCTELEM dc[16])
 static void vp8_idct_add_c(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride)
 {
     int i, t0, t1, t2, t3;
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
     DCTELEM tmp[16];
 
     for (i = 0; i < 4; i++) {
@@ -105,10 +104,10 @@ static void vp8_idct_add_c(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride)
         t2 = MUL_35468(tmp[1*4+i]) - MUL_20091(tmp[3*4+i]);
         t3 = MUL_20091(tmp[1*4+i]) + MUL_35468(tmp[3*4+i]);
 
-        dst[0] = cm[dst[0] + ((t0 + t3 + 4) >> 3)];
-        dst[1] = cm[dst[1] + ((t1 + t2 + 4) >> 3)];
-        dst[2] = cm[dst[2] + ((t1 - t2 + 4) >> 3)];
-        dst[3] = cm[dst[3] + ((t0 - t3 + 4) >> 3)];
+        dst[0] = av_clip_uint8(dst[0] + ((t0 + t3 + 4) >> 3));
+        dst[1] = av_clip_uint8(dst[1] + ((t1 + t2 + 4) >> 3));
+        dst[2] = av_clip_uint8(dst[2] + ((t1 - t2 + 4) >> 3));
+        dst[3] = av_clip_uint8(dst[3] + ((t0 - t3 + 4) >> 3));
         dst += stride;
     }
 }
@@ -116,14 +115,13 @@ static void vp8_idct_add_c(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride)
 static void vp8_idct_dc_add_c(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride)
 {
     int i, dc = (block[0] + 4) >> 3;
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP + dc;
     block[0] = 0;
 
     for (i = 0; i < 4; i++) {
-        dst[0] = cm[dst[0]];
-        dst[1] = cm[dst[1]];
-        dst[2] = cm[dst[2]];
-        dst[3] = cm[dst[3]];
+        dst[0] = av_clip_uint8(dst[0] + dc);
+        dst[1] = av_clip_uint8(dst[1] + dc);
+        dst[2] = av_clip_uint8(dst[2] + dc);
+        dst[3] = av_clip_uint8(dst[3] + dc);
         dst += stride;
     }
 }
diff --git a/libavcodec/x86/sbrdsp.asm b/libavcodec/x86/sbrdsp.asm
index c165c52..c3b559b 100644
--- a/libavcodec/x86/sbrdsp.asm
+++ b/libavcodec/x86/sbrdsp.asm
@@ -104,7 +104,7 @@ cglobal sbr_hf_g_filt, 5, 6, 5
     movq        m2, [r1]
     punpckldq   m0, m0
     mulps       m2, m0
-    movq      [r0], m2
+    movlps    [r0], m2
     add         r0, 8
     add         r2, 4
     add         r1, STEP
diff --git a/libavutil/avutil.h b/libavutil/avutil.h
index 0f1fd0a..2ec2f2e 100644
--- a/libavutil/avutil.h
+++ b/libavutil/avutil.h
@@ -153,7 +153,7 @@
  */
 
 #define LIBAVUTIL_VERSION_MAJOR 51
-#define LIBAVUTIL_VERSION_MINOR 41
+#define LIBAVUTIL_VERSION_MINOR 42
 #define LIBAVUTIL_VERSION_MICRO 100
 
 #define LIBAVUTIL_VERSION_INT   AV_VERSION_INT(LIBAVUTIL_VERSION_MAJOR, \
diff --git a/libavutil/cpu.c b/libavutil/cpu.c
index fa64a83..fed6093 100644
--- a/libavutil/cpu.c
+++ b/libavutil/cpu.c
@@ -23,7 +23,7 @@ static int flags, checked;
 
 void av_force_cpu_flags(int arg){
     flags   = arg;
-    checked = 1;
+    checked = arg != -1;
 }
 
 int av_get_cpu_flags(void)
@@ -39,6 +39,13 @@ int av_get_cpu_flags(void)
     return flags;
 }
 
+void av_set_cpu_flags_mask(int mask)
+{
+    checked       = 0;
+    flags         = av_get_cpu_flags() & mask;
+    checked       = 1;
+}
+
 #ifdef TEST
 
 #undef printf
diff --git a/libavutil/cpu.h b/libavutil/cpu.h
index 691ee9c..bb80c8d 100644
--- a/libavutil/cpu.h
+++ b/libavutil/cpu.h
@@ -21,6 +21,8 @@
 #ifndef AVUTIL_CPU_H
 #define AVUTIL_CPU_H
 
+#include "attributes.h"
+
 #define AV_CPU_FLAG_FORCE    0x80000000 /* force usage of selected flags (OR) */
 
     /* lower 16 bits - CPU features */
@@ -49,12 +51,19 @@
  */
 int av_get_cpu_flags(void);
 
-
 /**
  * Disables cpu detection and forces the specified flags.
  */
 void av_force_cpu_flags(int flags);
 
+/**
+ * Set a mask on flags returned by av_get_cpu_flags().
+ * This function is mainly useful for testing.
+ * Please use av_force_cpu_flags() and av_get_cpu_flags() instead which are more flexible
+ *
+ * @warning this function is not thread safe.
+ */
+attribute_deprecated void av_set_cpu_flags_mask(int mask);
 
 /* The following CPU-specific functions shall not be called directly. */
 int ff_get_cpu_flags_arm(void);
diff --git a/libswscale/ppc/swscale_altivec.c b/libswscale/ppc/swscale_altivec.c
index b6aed04..df8afb0 100644
--- a/libswscale/ppc/swscale_altivec.c
+++ b/libswscale/ppc/swscale_altivec.c
@@ -144,7 +144,7 @@ static void yuv2planeX_altivec(const int16_t *filter, int filterSize,
 
 static void hScale_altivec_real(SwsContext *c, int16_t *dst, int dstW,
                                 const uint8_t *src, const int16_t *filter,
-                                const int16_t *filterPos, int filterSize)
+                                const int32_t *filterPos, int filterSize)
 {
     register int i;
     DECLARE_ALIGNED(16, int, tempo)[4];
diff --git a/libswscale/swscale.c b/libswscale/swscale.c
index 6dc02b4..38856d9 100644
--- a/libswscale/swscale.c
+++ b/libswscale/swscale.c
@@ -63,7 +63,7 @@ static av_always_inline void fillPlane(uint8_t* plane, int stride,
 
 static void hScale16To19_c(SwsContext *c, int16_t *_dst, int dstW, const uint8_t *_src,
                            const int16_t *filter,
-                           const int16_t *filterPos, int filterSize)
+                           const int32_t *filterPos, int filterSize)
 {
     int i;
     int32_t *dst = (int32_t *) _dst;
@@ -89,7 +89,7 @@ static void hScale16To19_c(SwsContext *c, int16_t *_dst, int dstW, const uint8_t
 
 static void hScale16To15_c(SwsContext *c, int16_t *dst, int dstW, const uint8_t *_src,
                            const int16_t *filter,
-                           const int16_t *filterPos, int filterSize)
+                           const int32_t *filterPos, int filterSize)
 {
     int i;
     const uint16_t *src = (const uint16_t *) _src;
@@ -113,7 +113,7 @@ static void hScale16To15_c(SwsContext *c, int16_t *dst, int dstW, const uint8_t
 
 // bilinear / bicubic scaling
 static void hScale8To15_c(SwsContext *c, int16_t *dst, int dstW, const uint8_t *src,
-                          const int16_t *filter, const int16_t *filterPos,
+                          const int16_t *filter, const int32_t *filterPos,
                           int filterSize)
 {
     int i;
@@ -131,7 +131,7 @@ static void hScale8To15_c(SwsContext *c, int16_t *dst, int dstW, const uint8_t *
 }
 
 static void hScale8To19_c(SwsContext *c, int16_t *_dst, int dstW, const uint8_t *src,
-                          const int16_t *filter, const int16_t *filterPos,
+                          const int16_t *filter, const int32_t *filterPos,
                           int filterSize)
 {
     int i;
@@ -234,7 +234,7 @@ static void hyscale_fast_c(SwsContext *c, int16_t *dst, int dstWidth,
 static av_always_inline void hyscale(SwsContext *c, int16_t *dst, int dstWidth,
                                      const uint8_t *src_in[4], int srcW, int xInc,
                                      const int16_t *hLumFilter,
-                                     const int16_t *hLumFilterPos, int hLumFilterSize,
+                                     const int32_t *hLumFilterPos, int hLumFilterSize,
                                      uint8_t *formatConvBuffer,
                                      uint32_t *pal, int isAlpha)
 {
@@ -282,7 +282,7 @@ static void hcscale_fast_c(SwsContext *c, int16_t *dst1, int16_t *dst2,
 static av_always_inline void hcscale(SwsContext *c, int16_t *dst1, int16_t *dst2, int dstWidth,
                                      const uint8_t *src_in[4],
                                      int srcW, int xInc, const int16_t *hChrFilter,
-                                     const int16_t *hChrFilterPos, int hChrFilterSize,
+                                     const int32_t *hChrFilterPos, int hChrFilterSize,
                                      uint8_t *formatConvBuffer, uint32_t *pal)
 {
     const uint8_t *src1 = src_in[1], *src2 = src_in[2];
@@ -326,10 +326,10 @@ static int swScale(SwsContext *c, const uint8_t* src[],
     const int chrXInc= c->chrXInc;
     const enum PixelFormat dstFormat= c->dstFormat;
     const int flags= c->flags;
-    int16_t *vLumFilterPos= c->vLumFilterPos;
-    int16_t *vChrFilterPos= c->vChrFilterPos;
-    int16_t *hLumFilterPos= c->hLumFilterPos;
-    int16_t *hChrFilterPos= c->hChrFilterPos;
+    int32_t *vLumFilterPos= c->vLumFilterPos;
+    int32_t *vChrFilterPos= c->vChrFilterPos;
+    int32_t *hLumFilterPos= c->hLumFilterPos;
+    int32_t *hChrFilterPos= c->hChrFilterPos;
     int16_t *hLumFilter= c->hLumFilter;
     int16_t *hChrFilter= c->hChrFilter;
     int32_t *lumMmxFilter= c->lumMmxFilter;
diff --git a/libswscale/swscale_internal.h b/libswscale/swscale_internal.h
index 9304356..e73806e 100644
--- a/libswscale/swscale_internal.h
+++ b/libswscale/swscale_internal.h
@@ -299,10 +299,10 @@ typedef struct SwsContext {
     int16_t *hChrFilter;          ///< Array of horizontal filter coefficients for chroma     planes.
     int16_t *vLumFilter;          ///< Array of vertical   filter coefficients for luma/alpha planes.
     int16_t *vChrFilter;          ///< Array of vertical   filter coefficients for chroma     planes.
-    int16_t *hLumFilterPos;       ///< Array of horizontal filter starting positions for each dst[i] for luma/alpha planes.
-    int16_t *hChrFilterPos;       ///< Array of horizontal filter starting positions for each dst[i] for chroma     planes.
-    int16_t *vLumFilterPos;       ///< Array of vertical   filter starting positions for each dst[i] for luma/alpha planes.
-    int16_t *vChrFilterPos;       ///< Array of vertical   filter starting positions for each dst[i] for chroma     planes.
+    int32_t *hLumFilterPos;       ///< Array of horizontal filter starting positions for each dst[i] for luma/alpha planes.
+    int32_t *hChrFilterPos;       ///< Array of horizontal filter starting positions for each dst[i] for chroma     planes.
+    int32_t *vLumFilterPos;       ///< Array of vertical   filter starting positions for each dst[i] for luma/alpha planes.
+    int32_t *vChrFilterPos;       ///< Array of vertical   filter starting positions for each dst[i] for chroma     planes.
     int hLumFilterSize;           ///< Horizontal filter size for luma/alpha pixels.
     int hChrFilterSize;           ///< Horizontal filter size for chroma     pixels.
     int vLumFilterSize;           ///< Vertical   filter size for luma/alpha pixels.
@@ -515,10 +515,10 @@ typedef struct SwsContext {
     /** @{ */
     void (*hyScale)(struct SwsContext *c, int16_t *dst, int dstW,
                     const uint8_t *src, const int16_t *filter,
-                    const int16_t *filterPos, int filterSize);
+                    const int32_t *filterPos, int filterSize);
     void (*hcScale)(struct SwsContext *c, int16_t *dst, int dstW,
                     const uint8_t *src, const int16_t *filter,
-                    const int16_t *filterPos, int filterSize);
+                    const int32_t *filterPos, int filterSize);
     /** @} */
 
     /// Color range conversion function for luma plane if needed.
diff --git a/libswscale/utils.c b/libswscale/utils.c
index 6391d51..7e8285d 100644
--- a/libswscale/utils.c
+++ b/libswscale/utils.c
@@ -191,7 +191,7 @@ static double getSplineCoeff(double a, double b, double c, double d, double dist
                                          dist-1.0);
 }
 
-static int initFilter(int16_t **outFilter, int16_t **filterPos, int *outFilterSize, int xInc,
+static int initFilter(int16_t **outFilter, int32_t **filterPos, int *outFilterSize, int xInc,
                       int srcW, int dstW, int filterAlign, int one, int flags, int cpu_flags,
                       SwsVector *srcFilter, SwsVector *dstFilter, double param[2])
 {
@@ -207,7 +207,7 @@ static int initFilter(int16_t **outFilter, int16_t **filterPos, int *outFilterSi
     emms_c(); //FIXME this should not be required but it IS (even for non-MMX versions)
 
     // NOTE: the +3 is for the MMX(+1)/SSE(+3) scaler which reads over the end
-    FF_ALLOC_OR_GOTO(NULL, *filterPos, (dstW+3)*sizeof(int16_t), fail);
+    FF_ALLOC_OR_GOTO(NULL, *filterPos, (dstW+3)*sizeof(**filterPos), fail);
 
     if (FFABS(xInc - 0x10000) <10) { // unscaled
         int i;
diff --git a/libswscale/x86/scale.asm b/libswscale/x86/scale.asm
index 3c8b6fa..b8cea49 100644
--- a/libswscale/x86/scale.asm
+++ b/libswscale/x86/scale.asm
@@ -38,7 +38,7 @@ SECTION .text
 ;                               (SwsContext *c, int{16,32}_t *dst,
 ;                                int dstW, const uint{8,16}_t *src,
 ;                                const int16_t *filter,
-;                                const int16_t *filterPos, int filterSize);
+;                                const int32_t *filterPos, int filterSize);
 ;
 ; Scale one horizontal line. Input is either 8-bits width or 16-bits width
 ; ($source_width can be either 8, 9, 10 or 16, difference is whether we have to
@@ -53,6 +53,9 @@ SECTION .text
 cglobal hscale%1to%2_%4_%5, %6, 7, %7
 %if ARCH_X86_64
     movsxd        r2, r2d
+%define mov32 movsxd
+%else ; x86-32
+%define mov32 mov
 %endif ; x86-64
 %if %2 == 19
 %if mmsize == 8 ; mmx
@@ -95,14 +98,14 @@ cglobal hscale%1to%2_%4_%5, %6, 7, %7
 %else ; %2 == 19
     lea           r1, [r1+r2*(4>>r2shr)]
 %endif ; %2 == 15/19
-    lea           r5, [r5+r2*(2>>r2shr)]
+    lea           r5, [r5+r2*(4>>r2shr)]
     neg           r2
 
 .loop:
 %if %3 == 4 ; filterSize == 4 scaling
     ; load 2x4 or 4x4 source pixels into m0/m1
-    movsx         r0, word [r5+r2*2+0]   ; filterPos[0]
-    movsx         r6, word [r5+r2*2+2]   ; filterPos[1]
+    mov32         r0, dword [r5+r2*4+0]  ; filterPos[0]
+    mov32         r6, dword [r5+r2*4+4]  ; filterPos[1]
     movlh         m0, [r3+r0*srcmul]     ; src[filterPos[0] + {0,1,2,3}]
 %if mmsize == 8
     movlh         m1, [r3+r6*srcmul]     ; src[filterPos[1] + {0,1,2,3}]
@@ -112,8 +115,8 @@ cglobal hscale%1to%2_%4_%5, %6, 7, %7
 %else ; %1 == 8
     movd          m4, [r3+r6*srcmul]     ; src[filterPos[1] + {0,1,2,3}]
 %endif
-    movsx         r0, word [r5+r2*2+4]   ; filterPos[2]
-    movsx         r6, word [r5+r2*2+6]   ; filterPos[3]
+    mov32         r0, dword [r5+r2*4+8]  ; filterPos[2]
+    mov32         r6, dword [r5+r2*4+12] ; filterPos[3]
     movlh         m1, [r3+r0*srcmul]     ; src[filterPos[2] + {0,1,2,3}]
 %if %1 > 8
     movhps        m1, [r3+r6*srcmul]     ; src[filterPos[3] + {0,1,2,3}]
@@ -156,8 +159,8 @@ cglobal hscale%1to%2_%4_%5, %6, 7, %7
 %endif ; mmx/sse2/ssse3/sse4
 %else ; %3 == 8, i.e. filterSize == 8 scaling
     ; load 2x8 or 4x8 source pixels into m0, m1, m4 and m5
-    movsx         r0, word [r5+r2*1+0]   ; filterPos[0]
-    movsx         r6, word [r5+r2*1+2]   ; filterPos[1]
+    mov32         r0, dword [r5+r2*2+0]  ; filterPos[0]
+    mov32         r6, dword [r5+r2*2+4]  ; filterPos[1]
     movbh         m0, [r3+ r0   *srcmul] ; src[filterPos[0] + {0,1,2,3,4,5,6,7}]
 %if mmsize == 8
     movbh         m1, [r3+(r0+4)*srcmul] ; src[filterPos[0] + {4,5,6,7}]
@@ -165,8 +168,8 @@ cglobal hscale%1to%2_%4_%5, %6, 7, %7
     movbh         m5, [r3+(r6+4)*srcmul] ; src[filterPos[1] + {4,5,6,7}]
 %else ; mmsize == 16
     movbh         m1, [r3+ r6   *srcmul] ; src[filterPos[1] + {0,1,2,3,4,5,6,7}]
-    movsx         r0, word [r5+r2*1+4]   ; filterPos[2]
-    movsx         r6, word [r5+r2*1+6]   ; filterPos[3]
+    mov32         r0, dword [r5+r2*2+8]  ; filterPos[2]
+    mov32         r6, dword [r5+r2*2+12] ; filterPos[3]
     movbh         m4, [r3+ r0   *srcmul] ; src[filterPos[2] + {0,1,2,3,4,5,6,7}]
     movbh         m5, [r3+ r6   *srcmul] ; src[filterPos[3] + {0,1,2,3,4,5,6,7}]
 %endif ; mmsize == 8/16
@@ -251,7 +254,7 @@ cglobal hscale%1to%2_%4_%5, %6, 7, %7
 %define r1x     r1
 %define filter2 r6m
 %endif ; x86-32/64
-    lea           r5, [r5+r2*2]
+    lea           r5, [r5+r2*4]
 %if %2 == 15
     lea           r1, [r1+r2*2]
 %else ; %2 == 19
@@ -261,8 +264,8 @@ cglobal hscale%1to%2_%4_%5, %6, 7, %7
     neg           r2
 
 .loop:
-    movsx         r0, word [r5+r2*2+0]   ; filterPos[0]
-    movsx        r1x, word [r5+r2*2+2]   ; filterPos[1]
+    mov32         r0, dword [r5+r2*4+0]  ; filterPos[0]
+    mov32        r1x, dword [r5+r2*4+4]  ; filterPos[1]
     ; FIXME maybe do 4px/iteration on x86-64 (x86-32 wouldn't have enough regs)?
     pxor          m4, m4
     pxor          m5, m5
@@ -293,7 +296,7 @@ cglobal hscale%1to%2_%4_%5, %6, 7, %7
     jl .innerloop
 
 %ifidn %4, X4
-    movsx        r1x, word [r5+r2*2+2]   ; filterPos[1]
+    mov32        r1x, dword [r5+r2*4+4]  ; filterPos[1]
     movlh         m0, [src_reg+r0 *srcmul] ; split last 4 srcpx of dstpx[0]
     sub          r1x, r6                   ; and first 4 srcpx of dstpx[1]
 %if %1 > 8
diff --git a/libswscale/x86/swscale_mmx.c b/libswscale/x86/swscale_mmx.c
index 28fc004..0c8732c 100644
--- a/libswscale/x86/swscale_mmx.c
+++ b/libswscale/x86/swscale_mmx.c
@@ -94,8 +94,8 @@ void updateMMXDitherTables(SwsContext *c, int dstY, int lumBufIndex, int chrBufI
     int16_t **alpPixBuf= c->alpPixBuf;
     const int vLumBufSize= c->vLumBufSize;
     const int vChrBufSize= c->vChrBufSize;
-    int16_t *vLumFilterPos= c->vLumFilterPos;
-    int16_t *vChrFilterPos= c->vChrFilterPos;
+    int32_t *vLumFilterPos= c->vLumFilterPos;
+    int32_t *vChrFilterPos= c->vChrFilterPos;
     int16_t *vLumFilter= c->vLumFilter;
     int16_t *vChrFilter= c->vChrFilter;
     int32_t *lumMmxFilter= c->lumMmxFilter;
@@ -266,7 +266,7 @@ extern void ff_hscale ## from_bpc ## to ## to_bpc ## _ ## filter_n ## _ ## opt(
                                                 SwsContext *c, int16_t *data, \
                                                 int dstW, const uint8_t *src, \
                                                 const int16_t *filter, \
-                                                const int16_t *filterPos, int filterSize)
+                                                const int32_t *filterPos, int filterSize)
 
 #define SCALE_FUNCS(filter_n, opt) \
     SCALE_FUNC(filter_n,  8, 15, opt); \
diff --git a/libswscale/x86/swscale_template.c b/libswscale/x86/swscale_template.c
index 4eee894..cf8e802 100644
--- a/libswscale/x86/swscale_template.c
+++ b/libswscale/x86/swscale_template.c
@@ -1450,7 +1450,7 @@ static void RENAME(hyscale_fast)(SwsContext *c, int16_t *dst,
                                  int dstWidth, const uint8_t *src,
                                  int srcW, int xInc)
 {
-    int16_t *filterPos = c->hLumFilterPos;
+    int32_t *filterPos = c->hLumFilterPos;
     int16_t *filter    = c->hLumFilter;
     void    *mmx2FilterCode= c->lumMmx2FilterCode;
     int i;
@@ -1546,7 +1546,7 @@ static void RENAME(hcscale_fast)(SwsContext *c, int16_t *dst1, int16_t *dst2,
                                  int dstWidth, const uint8_t *src1,
                                  const uint8_t *src2, int srcW, int xInc)
 {
-    int16_t *filterPos = c->hChrFilterPos;
+    int32_t *filterPos = c->hChrFilterPos;
     int16_t *filter    = c->hChrFilter;
     void    *mmx2FilterCode= c->chrMmx2FilterCode;
     int i;
diff --git a/tests/Makefile b/tests/Makefile
index 45ce130..d9856dc 100644
--- a/tests/Makefile
+++ b/tests/Makefile
@@ -142,7 +142,7 @@ fate:: $(FATE)
 
 $(FATE): $(TOOL)$(EXESUF) $(FATE_UTILS:%=tests/%$(HOSTEXESUF))
 	@echo "TEST    $(@:fate-%=%)"
-	$(Q)$(SRC_PATH)/tests/fate-run.sh $@ "$(SAMPLES)" "$(TARGET_EXEC)" "$(TARGET_PATH)" '$(CMD)' '$(CMP)' '$(REF)' '$(FUZZ)' '$(THREADS)' '$(THREAD_TYPE)' '$(TOOL)'
+	$(Q)$(SRC_PATH)/tests/fate-run.sh $@ "$(SAMPLES)" "$(TARGET_EXEC)" "$(TARGET_PATH)" '$(CMD)' '$(CMP)' '$(REF)' '$(FUZZ)' '$(THREADS)' '$(THREAD_TYPE)' '$(CPUFLAGS)'
 
 fate-list:
 	@printf '%s\n' $(sort $(FATE))
diff --git a/tests/fate-run.sh b/tests/fate-run.sh
index d7b5e7e..f7a51ad 100755
--- a/tests/fate-run.sh
+++ b/tests/fate-run.sh
@@ -17,7 +17,7 @@ ref=${7:-"${base}/ref/fate/${test}"}
 fuzz=$8
 threads=${9:-1}
 thread_type=${10:-frame+slice}
-tool=${11}
+cpuflags=${11:-all}
 
 outdir="tests/data/fate"
 outfile="${outdir}/${test}"
@@ -51,7 +51,7 @@ run(){
 }
 
 avconv(){
-    run $tool -nostats -threads $threads -thread_type $thread_type "$@"
+    run ffmpeg -nostats -threads $threads -thread_type $thread_type -cpuflags $cpuflags "$@"
 }
 
 framecrc(){
@@ -77,7 +77,7 @@ pcm(){
 regtest(){
     t="${test#$2-}"
     ref=${base}/ref/$2/$t
-    ${base}/${1}-regression.sh $t $2 $3 "$target_exec" "$target_path" "$threads" "$thread_type" "$tool" "$samples"
+    ${base}/${1}-regression.sh $t $2 $3 "$target_exec" "$target_path" "$threads" "$thread_type" "$cpuflags" "$samples"
 }
 
 codectest(){
diff --git a/tests/regression-funcs.sh b/tests/regression-funcs.sh
index 1cafb3f..7244b66 100755
--- a/tests/regression-funcs.sh
+++ b/tests/regression-funcs.sh
@@ -10,7 +10,7 @@ raw_src_dir=$3
 target_exec=$4
 target_path=$5
 threads=${6:-1}
-tool=$8
+cpuflags=${8:-all}
 samples=$9
 
 datadir="./tests/data"
@@ -20,7 +20,7 @@ this="$test.$test_ref"
 outfile="$datadir/$test_ref/"
 
 # various files
-avconv="$target_exec ${target_path}/${tool}"
+avconv="$target_exec ${target_path}/ffmpeg"
 tiny_psnr="tests/tiny_psnr"
 raw_src="${target_path}/$raw_src_dir/%02d.pgm"
 raw_dst="$datadir/$this.out.yuv"
@@ -45,7 +45,7 @@ echov(){
 
 . $(dirname $0)/md5.sh
 
-AVCONV_OPTS="-nostats -y"
+AVCONV_OPTS="-nostats -y -cpuflags $cpuflags"
 COMMON_OPTS="-flags +bitexact -idct simple -sws_flags +accurate_rnd+bitexact"
 DEC_OPTS="$COMMON_OPTS -threads $threads"
 ENC_OPTS="$COMMON_OPTS -threads 1 -dct fastint"