Skip to content

L
libswscale
Commit e76f1aa2 authored by michael's avatar michael
Browse files
Options

minor speedup 

cleanup


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@2571 b3059339-0415-0410-9bf9-f77b7e298cf2
parent e9a032ef
Show whitespace changes
Inline Side-by-side
Showing with 54 additions and 24 deletions
postprocess.c
View file @ e76f1aa2
......@@ -62,6 +62,7 @@ split this huge file
fix warnings (unused vars, ...)
noise reduction filters
border remover
optimize c versions
...
Notes:
......@@ -417,7 +418,6 @@ static inline void doVertLowPass(uint8_t *src, int stride, int QP)
#if defined (HAVE_MMX2) || defined (HAVE_3DNOW)
src+= stride*3;
asm volatile( //"movv %0 %1 %2\n\t"
"pushl %0 \n\t"
"movq pQPb, %%mm0 \n\t" // QP,..., QP
"movq (%0), %%mm6 \n\t"
......@@ -535,7 +535,7 @@ static inline void doVertLowPass(uint8_t *src, int stride, int QP)
PAVGB(%%mm3, %%mm0) // 112 /4
PAVGB(%%mm0, %%mm5) // 112246 /16
"movq %%mm5, (%%eax, %1, 4) \n\t" // X
"popl %0\n\t"
"subl %1, %0 \n\t"
:
: "r" (src), "r" (stride)
......@@ -1167,7 +1167,21 @@ static inline void doVertDefFilter(uint8_t src[], int stride, int QP)
"movq temp0, %%mm2 \n\t" // 2L0 - 5L1 + 5L2 - 2L3
"movq temp1, %%mm3 \n\t" // 2H0 - 5H1 + 5H2 - 2H3
//FIXME pxor, psubw, pmax for abs
#ifdef HAVE_MMX2
"movq %%mm7, %%mm6 \n\t" // 0
"psubw %%mm0, %%mm6 \n\t"
"pmaxsw %%mm6, %%mm0 \n\t" // |2L4 - 5L5 + 5L6 - 2L7|
"movq %%mm7, %%mm6 \n\t" // 0
"psubw %%mm1, %%mm6 \n\t"
"pmaxsw %%mm6, %%mm1 \n\t" // |2H4 - 5H5 + 5H6 - 2H7|
"movq %%mm7, %%mm6 \n\t" // 0
"psubw %%mm2, %%mm6 \n\t"
"pmaxsw %%mm6, %%mm2 \n\t" // |2L0 - 5L1 + 5L2 - 2L3|
"movq %%mm7, %%mm6 \n\t" // 0
"psubw %%mm3, %%mm6 \n\t"
"pmaxsw %%mm6, %%mm3 \n\t" // |2H0 - 5H1 + 5H2 - 2H3|
#else
"movq %%mm7, %%mm6 \n\t" // 0
"pcmpgtw %%mm0, %%mm6 \n\t"
"pxor %%mm6, %%mm0 \n\t"
......@@ -1176,7 +1190,6 @@ static inline void doVertDefFilter(uint8_t src[], int stride, int QP)
"pcmpgtw %%mm1, %%mm6 \n\t"
"pxor %%mm6, %%mm1 \n\t"
"psubw %%mm6, %%mm1 \n\t" // |2H4 - 5H5 + 5H6 - 2H7|
"movq %%mm7, %%mm6 \n\t" // 0
"pcmpgtw %%mm2, %%mm6 \n\t"
"pxor %%mm6, %%mm2 \n\t"
......@@ -1185,6 +1198,7 @@ static inline void doVertDefFilter(uint8_t src[], int stride, int QP)
"pcmpgtw %%mm3, %%mm6 \n\t"
"pxor %%mm6, %%mm3 \n\t"
"psubw %%mm6, %%mm3 \n\t" // |2H0 - 5H1 + 5H2 - 2H3|
#endif
#ifdef HAVE_MMX2
"pminsw %%mm2, %%mm0 \n\t"
......@@ -2996,6 +3010,7 @@ static void postProcess(uint8_t src[], int srcStride, uint8_t dst[], int dstStri
long long memcpyTime=0, vertTime=0, horizTime=0, sumTime;
sumTime= rdtsc();
#endif
//mode= 0x7F;
if(tempDst==NULL)
{
......
postprocess_template.c
View file @ e76f1aa2
......@@ -62,6 +62,7 @@ split this huge file
fix warnings (unused vars, ...)
noise reduction filters
border remover
optimize c versions
...
Notes:
......@@ -417,7 +418,6 @@ static inline void doVertLowPass(uint8_t *src, int stride, int QP)
#if defined (HAVE_MMX2) || defined (HAVE_3DNOW)
src+= stride*3;
asm volatile( //"movv %0 %1 %2\n\t"
"pushl %0 \n\t"
"movq pQPb, %%mm0 \n\t" // QP,..., QP
"movq (%0), %%mm6 \n\t"
......@@ -535,7 +535,7 @@ static inline void doVertLowPass(uint8_t *src, int stride, int QP)
PAVGB(%%mm3, %%mm0) // 112 /4
PAVGB(%%mm0, %%mm5) // 112246 /16
"movq %%mm5, (%%eax, %1, 4) \n\t" // X
"popl %0\n\t"
"subl %1, %0 \n\t"
:
: "r" (src), "r" (stride)
......@@ -1167,7 +1167,21 @@ static inline void doVertDefFilter(uint8_t src[], int stride, int QP)
"movq temp0, %%mm2 \n\t" // 2L0 - 5L1 + 5L2 - 2L3
"movq temp1, %%mm3 \n\t" // 2H0 - 5H1 + 5H2 - 2H3
//FIXME pxor, psubw, pmax for abs
#ifdef HAVE_MMX2
"movq %%mm7, %%mm6 \n\t" // 0
"psubw %%mm0, %%mm6 \n\t"
"pmaxsw %%mm6, %%mm0 \n\t" // |2L4 - 5L5 + 5L6 - 2L7|
"movq %%mm7, %%mm6 \n\t" // 0
"psubw %%mm1, %%mm6 \n\t"
"pmaxsw %%mm6, %%mm1 \n\t" // |2H4 - 5H5 + 5H6 - 2H7|
"movq %%mm7, %%mm6 \n\t" // 0
"psubw %%mm2, %%mm6 \n\t"
"pmaxsw %%mm6, %%mm2 \n\t" // |2L0 - 5L1 + 5L2 - 2L3|
"movq %%mm7, %%mm6 \n\t" // 0
"psubw %%mm3, %%mm6 \n\t"
"pmaxsw %%mm6, %%mm3 \n\t" // |2H0 - 5H1 + 5H2 - 2H3|
#else
"movq %%mm7, %%mm6 \n\t" // 0
"pcmpgtw %%mm0, %%mm6 \n\t"
"pxor %%mm6, %%mm0 \n\t"
......@@ -1176,7 +1190,6 @@ static inline void doVertDefFilter(uint8_t src[], int stride, int QP)
"pcmpgtw %%mm1, %%mm6 \n\t"
"pxor %%mm6, %%mm1 \n\t"
"psubw %%mm6, %%mm1 \n\t" // |2H4 - 5H5 + 5H6 - 2H7|
"movq %%mm7, %%mm6 \n\t" // 0
"pcmpgtw %%mm2, %%mm6 \n\t"
"pxor %%mm6, %%mm2 \n\t"
......@@ -1185,6 +1198,7 @@ static inline void doVertDefFilter(uint8_t src[], int stride, int QP)
"pcmpgtw %%mm3, %%mm6 \n\t"
"pxor %%mm6, %%mm3 \n\t"
"psubw %%mm6, %%mm3 \n\t" // |2H0 - 5H1 + 5H2 - 2H3|
#endif
#ifdef HAVE_MMX2
"pminsw %%mm2, %%mm0 \n\t"
......@@ -2996,6 +3010,7 @@ static void postProcess(uint8_t src[], int srcStride, uint8_t dst[], int dstStri
long long memcpyTime=0, vertTime=0, horizTime=0, sumTime;
sumTime= rdtsc();
#endif
//mode= 0x7F;
if(tempDst==NULL)
{
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment