diff options
| author | Martin Storsjö <martin@martin.st> | 2016-11-10 11:07:39 +0200 |
|---|---|---|
| committer | Martin Storsjö <martin@martin.st> | 2016-11-10 11:18:22 +0200 |
| commit | 557c1675cf0e803b2fee43b4c8b58433842c84d0 (patch) | |
| tree | 4cb2063cbbc569fc45faf05ae35d3c66fe29e69b /libavcodec/arm/vp9dsp_init_arm.c | |
| parent | 383d96aa2229f644d9bd77b821ed3a309da5e9fc (diff) | |
| download | ffmpeg-streaming-557c1675cf0e803b2fee43b4c8b58433842c84d0.zip ffmpeg-streaming-557c1675cf0e803b2fee43b4c8b58433842c84d0.tar.gz | |
arm: vp9mc: Minor adjustments from review of the aarch64 version
This work is sponsored by, and copyright, Google.
The speedup for the large horizontal filters is surprisingly
big on A7 and A53, while there's a minor slowdown (almost within
measurement noise) on A8 and A9.
Cortex A7 A8 A9 A53
orig:
vp9_put_8tap_smooth_64h_neon: 20270.0 14447.3 19723.9 10910.9
new:
vp9_put_8tap_smooth_64h_neon: 20165.8 14466.5 19730.2 10668.8
Signed-off-by: Martin Storsjö <martin@martin.st>
Diffstat (limited to 'libavcodec/arm/vp9dsp_init_arm.c')
| -rw-r--r-- | libavcodec/arm/vp9dsp_init_arm.c | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/libavcodec/arm/vp9dsp_init_arm.c b/libavcodec/arm/vp9dsp_init_arm.c index 1b00177..839037a 100644 --- a/libavcodec/arm/vp9dsp_init_arm.c +++ b/libavcodec/arm/vp9dsp_init_arm.c @@ -43,7 +43,7 @@ static void op##_##filter##sz##_hv_neon(uint8_t *dst, ptrdiff_t dst_stride, const uint8_t *src, ptrdiff_t src_stride, \ int h, int mx, int my) \ { \ - LOCAL_ALIGNED_16(uint8_t, temp, [((sz < 64 ? 2 * sz : 64) + 8) * sz]); \ + LOCAL_ALIGNED_16(uint8_t, temp, [((1 + (sz < 64)) * sz + 8) * sz]); \ /* We only need h + 7 lines, but the horizontal filter assumes an \ * even number of rows, so filter h + 8 lines here. */ \ ff_vp9_put_##filter##sz##_h_neon(temp, sz, \ |
