diff options
| author | Christophe Gisquet <christophe.gisquet@gmail.com> | 2015-10-12 19:37:49 +0200 |
|---|---|---|
| committer | Michael Niedermayer <michael@niedermayer.cc> | 2015-10-13 15:34:32 +0200 |
| commit | 7ece8b50b19e140ace13eda6f1a9f45f868c2528 (patch) | |
| tree | ee4e986a0913996db69ad2054cc03f464cd717d8 /libavcodec/x86/simple_idct.h | |
| parent | 4369b9dc7b2b0da594223ce46615ba8e2b4cead6 (diff) | |
x86: simple_idct: 12bits versions
On 12 frames of a 444p 12 bits DNxHR sequence, _put function:
C: 78902 decicycles in idct, 262071 runs, 73 skips
avx: 32478 decicycles in idct, 262045 runs, 99 skips
Difference between the 2:
stddev: 0.39 PSNR:104.47 MAXDIFF: 2
This is unavoidable and due to the scale factors used in the x86
version, which cannot match the C ones.
In addition, the trick of adding an initial bias to the input of a
pass can overflow, as the input coefficients are already 15bits,
which is the maximum this function can handle.
Overall, however, the omse on 12 bits samples goes from 0.16916 to
0.16883. Reducing rowshift by 1 improves to 0.0908, but causes
overflows.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Diffstat (limited to 'libavcodec/x86/simple_idct.h')
| -rw-r--r-- | libavcodec/x86/simple_idct.h | 6 |
1 files changed, 6 insertions, 0 deletions
diff --git a/libavcodec/x86/simple_idct.h b/libavcodec/x86/simple_idct.h index e8f59c1865..8eeb31e299 100644 --- a/libavcodec/x86/simple_idct.h +++ b/libavcodec/x86/simple_idct.h @@ -31,4 +31,10 @@ void ff_simple_idct10_avx(int16_t *block); void ff_simple_idct10_put_sse2(uint8_t *dest, int line_size, int16_t *block); void ff_simple_idct10_put_avx(uint8_t *dest, int line_size, int16_t *block); +void ff_simple_idct12_sse2(int16_t *block); +void ff_simple_idct12_avx(int16_t *block); + +void ff_simple_idct12_put_sse2(uint8_t *dest, int line_size, int16_t *block); +void ff_simple_idct12_put_avx(uint8_t *dest, int line_size, int16_t *block); + #endif /* AVCODEC_X86_SIMPLE_IDCT_H */ |