Skip to content
Projects
Groups
Snippets
Help
Loading...
Help
Contribute to GitLab
Sign in / Register
Toggle navigation
F
ffmpeg.wasm-core
Project
Project
Details
Activity
Cycle Analytics
Repository
Repository
Files
Commits
Branches
Tags
Contributors
Graph
Compare
Charts
Issues
0
Issues
0
List
Board
Labels
Milestones
Merge Requests
0
Merge Requests
0
CI / CD
CI / CD
Pipelines
Jobs
Schedules
Charts
Wiki
Wiki
Snippets
Snippets
Members
Members
Collapse sidebar
Close sidebar
Activity
Graph
Charts
Create a new issue
Jobs
Commits
Issue Boards
Open sidebar
Linshizhi
ffmpeg.wasm-core
Commits
c0eb0bb7
Commit
c0eb0bb7
authored
May 27, 2002
by
Zdenek Kabelac
Browse files
Options
Browse Files
Download
Email Patches
Plain Diff
* more PIC friendly and faster code
Originally committed as revision 616 to
svn://svn.ffmpeg.org/ffmpeg/trunk
parent
5f623001
Show whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
63 additions
and
96 deletions
+63
-96
dsputil_mmx_avg.h
libavcodec/i386/dsputil_mmx_avg.h
+63
-96
No files found.
libavcodec/i386/dsputil_mmx_avg.h
View file @
c0eb0bb7
...
...
@@ -27,32 +27,32 @@
static
void
DEF
(
put_pixels_x2
)(
UINT8
*
block
,
const
UINT8
*
pixels
,
int
line_size
,
int
h
)
{
__asm
__volatile
(
"xorl %%eax, %%eax
\n\t
"
".balign 16
\n\t
"
"lea (%3, %3), %%eax
\n\t
"
"1:
\n\t
"
"movq (%1, %%eax), %%mm0
\n\t
"
"movq 1(%1, %%eax), %%mm1
\n\t
"
"movq (%2, %%eax), %%mm2
\n\t
"
"movq 1(%2, %%eax), %%mm3
\n\t
"
"movq (%1), %%mm0
\n\t
"
"movq (%1, %3), %%mm2
\n\t
"
"movq 1(%1), %%mm1
\n\t
"
"movq 1(%1, %3), %%mm3
\n\t
"
"addl %%eax, %1
\n\t
"
PAVGB
" %%mm1, %%mm0
\n\t
"
PAVGB
" %%mm3, %%mm2
\n\t
"
"movq %%mm0, (%3, %%eax)
\n\t
"
"movq %%mm2, (%4, %%eax)
\n\t
"
"addl %5, %%eax
\n\t
"
"movq (%1, %%eax), %%mm0
\n\t
"
"movq 1(%1, %%eax), %%mm1
\n\t
"
"movq (%2, %%eax), %%mm2
\n\t
"
"movq 1(%2, %%eax), %%mm3
\n\t
"
"movq %%mm0, (%2)
\n\t
"
"movq %%mm2, (%2, %3)
\n\t
"
"movq (%1), %%mm0
\n\t
"
"movq 1(%1), %%mm1
\n\t
"
"movq (%1, %3), %%mm2
\n\t
"
"movq 1(%1, %3), %%mm3
\n\t
"
"addl %%eax, %2
\n\t
"
"addl %%eax, %1
\n\t
"
PAVGB
" %%mm1, %%mm0
\n\t
"
PAVGB
" %%mm3, %%mm2
\n\t
"
"movq %%mm0, (%3, %%eax)
\n\t
"
"movq %%mm2, (%
4, %%eax)
\n\t
"
"addl %
5, %%eax
\n\t
"
"movq %%mm0, (%2)
\n\t
"
"movq %%mm2, (%
2, %3)
\n\t
"
"addl %
%eax, %2
\n\t
"
"subl $4, %0
\n\t
"
" jnz 1b
\n\t
"
:
"+g"
(
h
)
:
"D"
(
pixels
),
"S"
(
pixels
+
line_size
),
"r"
(
block
),
"r"
(
block
+
line_size
),
"g"
(
line_size
<<
1
)
:
"+g"
(
h
),
"+S"
(
pixels
),
"+D"
(
block
)
:
"c"
(
line_size
)
:
"%eax"
,
"memory"
);
}
...
...
@@ -60,102 +60,68 @@ static void DEF(put_pixels_x2)(UINT8 *block, const UINT8 *pixels, int line_size,
static
void
DEF
(
put_no_rnd_pixels_x2
)(
UINT8
*
block
,
const
UINT8
*
pixels
,
int
line_size
,
int
h
)
{
__asm
__volatile
(
"
xorl %%eax
, %%eax
\n\t
"
"
lea (%3, %3)
, %%eax
\n\t
"
MOVQ_BONE
(
%%
mm7
)
".balign 16
\n\t
"
"1:
\n\t
"
"movq (%1, %%eax), %%mm0
\n\t
"
"movq 1(%1, %%eax), %%mm1
\n\t
"
"movq (%2, %%eax), %%mm2
\n\t
"
"movq 1(%2, %%eax), %%mm3
\n\t
"
"movq (%1), %%mm0
\n\t
"
"movq (%1, %3), %%mm2
\n\t
"
"movq 1(%1), %%mm1
\n\t
"
"movq 1(%1, %3), %%mm3
\n\t
"
"addl %%eax, %1
\n\t
"
"psubusb %%mm7, %%mm0
\n\t
"
"psubusb %%mm7, %%mm2
\n\t
"
PAVGB
" %%mm1, %%mm0
\n\t
"
PAVGB
" %%mm3, %%mm2
\n\t
"
"movq %%mm0, (%3, %%eax)
\n\t
"
"movq %%mm2, (%4, %%eax)
\n\t
"
"addl %5, %%eax
\n\t
"
"movq (%1, %%eax), %%mm0
\n\t
"
"movq 1(%1, %%eax), %%mm1
\n\t
"
"movq (%2, %%eax), %%mm2
\n\t
"
"movq 1(%2, %%eax), %%mm3
\n\t
"
"movq %%mm0, (%2)
\n\t
"
"movq %%mm2, (%2, %3)
\n\t
"
"movq (%1), %%mm0
\n\t
"
"movq 1(%1), %%mm1
\n\t
"
"movq (%1, %3), %%mm2
\n\t
"
"movq 1(%1, %3), %%mm3
\n\t
"
"addl %%eax, %2
\n\t
"
"addl %%eax, %1
\n\t
"
"psubusb %%mm7, %%mm0
\n\t
"
"psubusb %%mm7, %%mm2
\n\t
"
PAVGB
" %%mm1, %%mm0
\n\t
"
PAVGB
" %%mm3, %%mm2
\n\t
"
"movq %%mm0, (%3, %%eax)
\n\t
"
"movq %%mm2, (%
4, %%eax)
\n\t
"
"addl %
5, %%eax
\n\t
"
"movq %%mm0, (%2)
\n\t
"
"movq %%mm2, (%
2, %3)
\n\t
"
"addl %
%eax, %2
\n\t
"
"subl $4, %0
\n\t
"
" jnz 1b
\n\t
"
:
"+g"
(
h
)
:
"D"
(
pixels
),
"S"
(
pixels
+
line_size
),
"r"
(
block
),
"r"
(
block
+
line_size
),
"r"
(
line_size
<<
1
)
"jnz 1b
\n\t
"
:
"+g"
(
h
),
"+S"
(
pixels
),
"+D"
(
block
)
:
"c"
(
line_size
)
:
"%eax"
,
"memory"
);
}
static
void
DEF
(
put_pixels_y2
)(
UINT8
*
block
,
const
UINT8
*
pixels
,
int
line_size
,
int
h
)
{
#if 1
// Michael - measure me
__asm
__volatile
(
"lea (%3, %3), %%eax
\n\t
"
"movq (%1), %%mm0
\n\t
"
"subl %3, %2
\n\t
"
".balign 16
\n\t
"
"1:
\n\t
"
"movq (%1, %3), %%mm1
\n\t
"
"movq (%1, %%eax), %%mm2
\n\t
"
"addl %%eax, %1
\n\t
"
PAVGB
" %%mm1, %%mm0
\n\t
"
PAVGB
" %%mm2, %%mm1
\n\t
"
"addl %%eax, %1
\n\t
"
"movq %%mm0, (%2, %3)
\n\t
"
"movq %%mm1, (%2, %%eax)
\n\t
"
"movq (%1, %3), %%mm1
\n\t
"
"movq (%1, %%eax), %%mm0
\n\t
"
PAVGB
" %%mm1, %%mm2
\n\t
"
PAVGB
" %%mm0, %%mm1
\n\t
"
"addl %%eax, %2
\n\t
"
"addl %%eax, %1
\n\t
"
PAVGB
" %%mm1, %%mm2
\n\t
"
PAVGB
" %%mm0, %%mm1
\n\t
"
"movq %%mm2, (%2, %3)
\n\t
"
"movq %%mm1, (%2, %%eax)
\n\t
"
"addl %%eax, %2
\n\t
"
"subl $4, %0
\n\t
"
"jnz 1b
\n\t
"
:
"+g"
(
h
),
"+
D"
(
pixels
),
"+S
"
(
block
)
:
"+g"
(
h
),
"+
S"
(
pixels
),
"+D
"
(
block
)
:
"c"
(
line_size
)
:
"%eax"
,
"memory"
);
#else
// kabi measure me
__asm
__volatile
(
"movq (%2), %%mm0
\n\t
"
"addl %1, %2
\n\t
"
"xorl %%eax, %%eax
\n\t
"
"leal (%1, %2), %%edi
\n\t
"
"leal (%1, %3), %%esi
\n\t
"
"addl %1, %1
\n\t
"
".balign 16
\n\t
"
"1:
\n\t
"
"movq (%2 , %%eax), %%mm1
\n\t
"
"movq (%%edi, %%eax), %%mm2
\n\t
"
PAVGB
" %%mm1, %%mm0
\n\t
"
PAVGB
" %%mm2, %%mm1
\n\t
"
"movq %%mm0, (%3 , %%eax)
\n\t
"
"movq %%mm1, (%%esi, %%eax)
\n\t
"
"addl %1, %%eax
\n\t
"
"movq (%2 , %%eax), %%mm1
\n\t
"
"movq (%%edi, %%eax), %%mm0
\n\t
"
PAVGB
" %%mm1, %%mm2
\n\t
"
PAVGB
" %%mm0, %%mm1
\n\t
"
"movq %%mm2, (%3 , %%eax)
\n\t
"
"movq %%mm1, (%%esi, %%eax)
\n\t
"
"addl %1, %%eax
\n\t
"
"subl $4, %0
\n\t
"
" jnz 1b
\n\t
"
:
"+g"
(
h
),
"+r"
(
line_size
),
"+r"
(
pixels
)
:
"r"
(
block
)
:
"%eax"
,
"%esi"
,
"%edi"
,
"memory"
);
#endif
}
/* GL: this function does incorrect rounding if overflow */
...
...
@@ -163,31 +129,32 @@ static void DEF(put_no_rnd_pixels_y2)(UINT8 *block, const UINT8 *pixels, int lin
{
__asm
__volatile
(
MOVQ_BONE
(
%%
mm7
)
"
xorl %%eax, %%eax
\n\t
"
"
lea (%3, %3), %%eax
\n\t
"
"movq (%1), %%mm0
\n\t
"
"
.balign 16
\n\t
"
"
subl %3, %2
\n\t
"
"1:
\n\t
"
"movq (%2, %%eax), %%mm1
\n\t
"
"movq (%3, %%eax), %%mm2
\n\t
"
"movq (%1, %3), %%mm1
\n\t
"
"movq (%1, %%eax), %%mm2
\n\t
"
"addl %%eax, %1
\n\t
"
"psubusb %%mm7, %%mm1
\n\t
"
PAVGB
" %%mm1, %%mm0
\n\t
"
PAVGB
" %%mm2, %%mm1
\n\t
"
"movq %%mm0, (%4, %%eax)
\n\t
"
"movq %%mm1, (%5, %%eax)
\n\t
"
"addl %6, %%eax
\n\t
"
"movq (%2, %%eax), %%mm1
\n\t
"
"movq (%3, %%eax), %%mm0
\n\t
"
"movq %%mm0, (%2, %3)
\n\t
"
"movq %%mm1, (%2, %%eax)
\n\t
"
"movq (%1, %3), %%mm1
\n\t
"
"movq (%1, %%eax), %%mm0
\n\t
"
"addl %%eax, %2
\n\t
"
"addl %%eax, %1
\n\t
"
"psubusb %%mm7, %%mm1
\n\t
"
PAVGB
" %%mm1, %%mm2
\n\t
"
PAVGB
" %%mm0, %%mm1
\n\t
"
"movq %%mm2, (%
4, %%eax)
\n\t
"
"movq %%mm1, (%
5
, %%eax)
\n\t
"
"addl %
6, %%eax
\n\t
"
"movq %%mm2, (%
2, %3)
\n\t
"
"movq %%mm1, (%
2
, %%eax)
\n\t
"
"addl %
%eax, %2
\n\t
"
"subl $4, %0
\n\t
"
" jnz 1b
\n\t
"
:
"+g"
(
h
)
:
"D"
(
pixels
),
"S"
(
pixels
+
line_size
),
"r"
(
pixels
+
line_size
*
2
),
"r"
(
block
),
"r"
(
block
+
line_size
),
"g"
(
line_size
<<
1
)
"jnz 1b
\n\t
"
:
"+g"
(
h
),
"+S"
(
pixels
),
"+D"
(
block
)
:
"c"
(
line_size
)
:
"%eax"
,
"memory"
);
}
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment