ffmpeg/libavcodec/arm
Martin Storsjö 1e5d87eec3 arm: Add NEON optimizations for 10 and 12 bit vp9 loop filter
This work is sponsored by, and copyright, Google.

This is pretty much similar to the 8 bpp version, but in some senses
simpler. All input pixels are 16 bits, and all intermediates also fit
in 16 bits, so there's no lengthening/narrowing in the filter at all.

For the full 16 pixel wide filter, we can only process 4 pixels at a time
(using an implementation very much similar to the one for 8 bpp),
but we can do 8 pixels at a time for the 4 and 8 pixel wide filters with
a different implementation of the core filter.

Examples of relative speedup compared to the C version, from checkasm:
                                   Cortex    A7     A8     A9    A53
vp9_loop_filter_h_4_8_10bpp_neon:          1.83   2.16   1.40   2.09
vp9_loop_filter_h_8_8_10bpp_neon:          1.39   1.67   1.24   1.70
vp9_loop_filter_h_16_8_10bpp_neon:         1.56   1.47   1.10   1.81
vp9_loop_filter_h_16_16_10bpp_neon:        1.94   1.69   1.33   2.24
vp9_loop_filter_mix2_h_44_16_10bpp_neon:   2.01   2.27   1.67   2.39
vp9_loop_filter_mix2_h_48_16_10bpp_neon:   1.84   2.06   1.45   2.19
vp9_loop_filter_mix2_h_84_16_10bpp_neon:   1.89   2.20   1.47   2.29
vp9_loop_filter_mix2_h_88_16_10bpp_neon:   1.69   2.12   1.47   2.08
vp9_loop_filter_mix2_v_44_16_10bpp_neon:   3.16   3.98   2.50   4.05
vp9_loop_filter_mix2_v_48_16_10bpp_neon:   2.84   3.64   2.25   3.77
vp9_loop_filter_mix2_v_84_16_10bpp_neon:   2.65   3.45   2.16   3.54
vp9_loop_filter_mix2_v_88_16_10bpp_neon:   2.55   3.30   2.16   3.55
vp9_loop_filter_v_4_8_10bpp_neon:          2.85   3.97   2.24   3.68
vp9_loop_filter_v_8_8_10bpp_neon:          2.27   3.19   1.96   3.08
vp9_loop_filter_v_16_8_10bpp_neon:         3.42   2.74   2.26   4.40
vp9_loop_filter_v_16_16_10bpp_neon:        2.86   2.44   1.93   3.88

The speedup vs C code measured in checkasm is around 1.1-4x.
These numbers are quite inconclusive though, since the checkasm test
runs multiple filterings on top of each other, so later rounds might
end up with different codepaths (different decisions on which filter
to apply, based on input pixel differences).

Based on START_TIMER/STOP_TIMER wrapping around a few individual
functions, the speedup vs C code is around 2-4x.

Signed-off-by: Martin Storsjö <martin@martin.st>
2017-01-24 22:35:59 +02:00
..
Makefile arm: Add NEON optimizations for 10 and 12 bit vp9 loop filter 2017-01-24 22:35:59 +02:00
aac.h
aacpsdsp_init_arm.c
aacpsdsp_neon.S
ac3dsp_arm.S
ac3dsp_armv6.S
ac3dsp_init_arm.c
ac3dsp_neon.S
asm-offsets.h
audiodsp_arm.h
audiodsp_init_arm.c
audiodsp_init_neon.c
audiodsp_neon.S
blockdsp_arm.h
blockdsp_init_arm.c
blockdsp_init_neon.c
blockdsp_neon.S
cabac.h
dca.h
fft_fixed_init_arm.c
fft_fixed_neon.S
fft_init_arm.c
fft_neon.S
fft_vfp.S
flacdsp_arm.S
flacdsp_init_arm.c
fmtconvert_init_arm.c
fmtconvert_neon.S
fmtconvert_vfp.S
g722dsp_init_arm.c
g722dsp_neon.S
h264chroma_init_arm.c
h264cmc_neon.S avcodec: fix vc1dsp dependencies 2016-09-25 13:11:45 +02:00
h264dsp_init_arm.c
h264dsp_neon.S
h264idct_neon.S
h264pred_init_arm.c
h264pred_neon.S
h264qpel_init_arm.c
h264qpel_neon.S
hevcdsp_arm.h
hevcdsp_deblock_neon.S
hevcdsp_idct_neon.S
hevcdsp_init_arm.c
hevcdsp_init_neon.c
hevcdsp_qpel_neon.S
hpeldsp_arm.S
hpeldsp_arm.h
hpeldsp_armv6.S
hpeldsp_init_arm.c
hpeldsp_init_armv6.c
hpeldsp_init_neon.c
hpeldsp_neon.S
idct.h
idctdsp_arm.S
idctdsp_arm.h
idctdsp_armv6.S
idctdsp_init_arm.c
idctdsp_init_armv5te.c
idctdsp_init_armv6.c
idctdsp_init_neon.c
idctdsp_neon.S
int_neon.S
jrevdct_arm.S
lossless_audiodsp_init_arm.c
lossless_audiodsp_neon.S
mathops.h
mdct_fixed_neon.S
mdct_neon.S
mdct_vfp.S
me_cmp_armv6.S
me_cmp_init_arm.c
mlpdsp_armv5te.S
mlpdsp_armv6.S Merge commit '41ed7ab45fc693f7d7fc35664c0233f4c32d69bb' 2016-06-21 21:55:34 +02:00
mlpdsp_init_arm.c
mpegaudiodsp_fixed_armv6.S
mpegaudiodsp_init_arm.c
mpegvideo_arm.c
mpegvideo_arm.h
mpegvideo_armv5te.c Merge commit '41ed7ab45fc693f7d7fc35664c0233f4c32d69bb' 2016-06-21 21:55:34 +02:00
mpegvideo_armv5te_s.S
mpegvideo_neon.S
mpegvideoencdsp_armv6.S
mpegvideoencdsp_init_arm.c
neon.S
neontest.c avcodec: fix arguments on xmm/neon clobber test wrappers 2016-10-02 02:15:47 -03:00
pixblockdsp_armv6.S
pixblockdsp_init_arm.c
rdft_init_arm.c arm/rdft_init: fix license header 2016-04-12 15:01:19 -03:00
rdft_neon.S
rv34dsp_init_arm.c
rv34dsp_neon.S
rv40dsp_init_arm.c
rv40dsp_neon.S
sbrdsp_init_arm.c
sbrdsp_neon.S
simple_idct_arm.S Merge commit '41ed7ab45fc693f7d7fc35664c0233f4c32d69bb' 2016-06-21 21:55:34 +02:00
simple_idct_armv5te.S
simple_idct_armv6.S
simple_idct_neon.S
startcode.h
startcode_armv6.S
synth_filter_init_arm.c
synth_filter_neon.S
synth_filter_vfp.S
vc1dsp.h
vc1dsp_init_arm.c
vc1dsp_init_neon.c
vc1dsp_neon.S
videodsp_arm.h
videodsp_armv5te.S
videodsp_init_arm.c
videodsp_init_armv5te.c
vorbisdsp_init_arm.c
vorbisdsp_neon.S
vp3dsp_init_arm.c
vp3dsp_neon.S
vp6dsp_init_arm.c
vp6dsp_neon.S
vp8.h
vp8_armv6.S
vp8dsp.h
vp8dsp_armv6.S Merge commit '5f74bd31a9bd1ac7655103b11743c12d38e0419f' 2016-11-17 15:05:07 +01:00
vp8dsp_init_arm.c
vp8dsp_init_armv6.c
vp8dsp_init_neon.c
vp8dsp_neon.S Merge commit 'e8b96a77010dd62624c3c65c357d7ae3b397ceaa' 2016-11-14 15:21:49 +01:00
vp9dsp_init.h arm: Add NEON optimizations for 10 and 12 bit vp9 MC 2017-01-24 22:35:50 +02:00
vp9dsp_init_10bpp_arm.c arm: Add NEON optimizations for 10 and 12 bit vp9 MC 2017-01-24 22:35:50 +02:00
vp9dsp_init_12bpp_arm.c arm: Add NEON optimizations for 10 and 12 bit vp9 MC 2017-01-24 22:35:50 +02:00
vp9dsp_init_16bpp_arm_template.c arm: Add NEON optimizations for 10 and 12 bit vp9 loop filter 2017-01-24 22:35:59 +02:00
vp9dsp_init_arm.c arm: Add NEON optimizations for 10 and 12 bit vp9 MC 2017-01-24 22:35:50 +02:00
vp9itxfm_16bpp_neon.S arm: Add NEON optimizations for 10 and 12 bit vp9 itxfm 2017-01-24 22:35:56 +02:00
vp9itxfm_neon.S arm: vp9itxfm: Skip empty slices in the first pass of idct_idct 16x16 and 32x32 2017-01-14 21:13:30 +01:00
vp9lpf_16bpp_neon.S arm: Add NEON optimizations for 10 and 12 bit vp9 loop filter 2017-01-24 22:35:59 +02:00
vp9lpf_neon.S arm: vp9: Add NEON loop filters 2016-11-15 15:10:03 -05:00
vp9mc_16bpp_neon.S arm: Add NEON optimizations for 10 and 12 bit vp9 MC 2017-01-24 22:35:50 +02:00
vp9mc_neon.S arm: vp9mc: Fix vertical alignment of operands 2017-01-14 21:13:37 +01:00
vp56_arith.h