* commit '1bd890ad173d79e7906c5e1d06bf0a06cca4519d':
hevc: Separate adding residual to prediction from IDCT
This commit should be a noop but isn't because of the following renames:
- transform_add → add_residual
- transform_skip → dequant
- idct_4x4_luma → transform_4x4_luma
Merged-by: Clément Bœsch <cboesch@gopro.com>
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>
This work is sponsored by, and copyright, Google.
This is structured similarly to the 8 bit version. In the 8 bit
version, the coefficients are 16 bits, and intermediates are 32 bits.
Here, the coefficients are 32 bit. For the 4x4 transforms for 10 bit
content, the intermediates also fit in 32 bits, but for all other
transforms (4x4 for 12 bit content, and 8x8 and larger for both 10
and 12 bit) the intermediates are 64 bit.
For the existing 8 bit case, the 8x8 transform fit all coefficients in
registers; for 10/12 bit, when the coefficients are 32 bit, the 8x8
transform also has to be done in slices of 4 pixels (just as 16x16 and
32x32 for 8 bit).
The slice width also shrinks from 4 elements to 2 elements in parallel
for the 16x16 and 32x32 cases.
The 16 bit coefficients from idct_coeffs and similar tables also need
to be lenghtened to 32 bit in order to be used in multiplication with
vectors with 32 bit elements. This leads to the fixed coefficient
vectors needing more space, leading to more cases where they have to
be reloaded within the transform (in iadst16).
This technically would need testing in checkasm for subpartitions
in increments of 2, but that slows down normal checkasm runs
excessively.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_sub4_add_10_neon: 4.83 11.36 5.22 6.77
vp9_inv_adst_adst_8x8_sub8_add_10_neon: 4.12 7.60 4.06 4.84
vp9_inv_adst_adst_16x16_sub16_add_10_neon: 3.93 8.16 4.52 5.35
vp9_inv_dct_dct_4x4_sub1_add_10_neon: 1.36 2.57 1.41 1.61
vp9_inv_dct_dct_4x4_sub4_add_10_neon: 4.24 8.66 5.06 5.81
vp9_inv_dct_dct_8x8_sub1_add_10_neon: 2.63 4.18 1.68 2.87
vp9_inv_dct_dct_8x8_sub4_add_10_neon: 4.52 9.47 4.24 5.39
vp9_inv_dct_dct_8x8_sub8_add_10_neon: 3.45 7.34 3.45 4.30
vp9_inv_dct_dct_16x16_sub1_add_10_neon: 3.56 6.21 2.47 4.32
vp9_inv_dct_dct_16x16_sub2_add_10_neon: 5.68 12.73 5.28 7.07
vp9_inv_dct_dct_16x16_sub8_add_10_neon: 4.42 9.28 4.24 5.45
vp9_inv_dct_dct_16x16_sub16_add_10_neon: 3.41 7.29 3.35 4.19
vp9_inv_dct_dct_32x32_sub1_add_10_neon: 4.52 8.35 3.83 6.40
vp9_inv_dct_dct_32x32_sub2_add_10_neon: 5.86 13.19 6.14 7.04
vp9_inv_dct_dct_32x32_sub16_add_10_neon: 4.29 8.11 4.59 5.06
vp9_inv_dct_dct_32x32_sub32_add_10_neon: 3.31 5.70 3.56 3.84
vp9_inv_wht_wht_4x4_sub4_add_10_neon: 1.89 2.80 1.82 1.97
The speedup compared to the C functions is around 1.3 to 7x for the
full transforms, even higher for the smaller subpartitions.
Signed-off-by: Martin Storsjö <martin@martin.st>
This work is sponsored by, and copyright, Google.
The plain pixel put/copy functions are used from the 8 bit version,
for the double size (e.g. put16 uses ff_vp9_copy32_neon), and a new
copy128 is added.
Compared with the 8 bit version, the filters can no longer use the
trick to accumulate in 16 bit with only saturation at the end, but now
the accumulators need to be 32 bit. This avoids the need to keep track
of which filter index is the largest though, reducing the size of the
executable code for these filters.
For the horizontal filters, we only do 4 or 8 pixels wide in parallel
(while doing two rows at a time), since we don't have enough register
space to filter 16 pixels wide.
For the vertical filters, we still do 4 and 8 pixels in parallel just
as in the 8 bit case, but we need to store the output after every 2
rows instead of after every 4 rows.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_avg4_10bpp_neon: 2.25 2.44 3.05 2.16
vp9_avg8_10bpp_neon: 3.66 8.48 3.86 3.50
vp9_avg16_10bpp_neon: 3.39 8.26 3.37 2.72
vp9_avg32_10bpp_neon: 4.03 10.20 4.07 3.42
vp9_avg64_10bpp_neon: 4.15 10.01 4.13 3.70
vp9_avg_8tap_smooth_4h_10bpp_neon: 3.38 6.22 3.41 4.75
vp9_avg_8tap_smooth_4hv_10bpp_neon: 3.89 6.39 4.30 5.32
vp9_avg_8tap_smooth_4v_10bpp_neon: 5.32 9.73 6.34 7.31
vp9_avg_8tap_smooth_8h_10bpp_neon: 4.45 9.40 4.68 6.87
vp9_avg_8tap_smooth_8hv_10bpp_neon: 4.64 8.91 5.44 6.47
vp9_avg_8tap_smooth_8v_10bpp_neon: 6.44 13.42 8.68 8.79
vp9_avg_8tap_smooth_64h_10bpp_neon: 4.66 9.02 4.84 7.71
vp9_avg_8tap_smooth_64hv_10bpp_neon: 4.61 9.14 4.92 7.10
vp9_avg_8tap_smooth_64v_10bpp_neon: 6.90 14.13 9.57 10.41
vp9_put4_10bpp_neon: 1.33 1.46 2.09 1.33
vp9_put8_10bpp_neon: 1.57 3.42 1.83 1.84
vp9_put16_10bpp_neon: 1.55 4.78 2.17 1.89
vp9_put32_10bpp_neon: 2.06 5.35 2.14 2.30
vp9_put64_10bpp_neon: 3.00 2.41 1.95 1.66
vp9_put_8tap_smooth_4h_10bpp_neon: 3.19 5.81 3.31 4.63
vp9_put_8tap_smooth_4hv_10bpp_neon: 3.86 6.22 4.32 5.21
vp9_put_8tap_smooth_4v_10bpp_neon: 5.40 9.77 6.08 7.21
vp9_put_8tap_smooth_8h_10bpp_neon: 4.22 8.41 4.46 6.63
vp9_put_8tap_smooth_8hv_10bpp_neon: 4.56 8.51 5.39 6.25
vp9_put_8tap_smooth_8v_10bpp_neon: 6.60 12.43 8.17 8.89
vp9_put_8tap_smooth_64h_10bpp_neon: 4.41 8.59 4.54 7.49
vp9_put_8tap_smooth_64hv_10bpp_neon: 4.43 8.58 5.34 6.63
vp9_put_8tap_smooth_64v_10bpp_neon: 7.26 13.92 9.27 10.92
For the larger 8tap filters, the speedup vs C code is around 4-14x.
Signed-off-by: Martin Storsjö <martin@martin.st>
This work is sponsored by, and copyright, Google.
This is more in line with how it will be extended for more bitdepths.
Signed-off-by: Martin Storsjö <martin@martin.st>
This work is sponsored by, and copyright, Google.
Previously all subpartitions except the eob=1 (DC) case ran with
the same runtime:
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_16x16_sub16_add_neon: 3188.1 2435.4 2499.0 1969.0
vp9_inv_dct_dct_32x32_sub32_add_neon: 18531.7 16582.3 14207.6 12000.3
By skipping individual 4x16 or 4x32 pixel slices in the first pass,
we reduce the runtime of these functions like this:
vp9_inv_dct_dct_16x16_sub1_add_neon: 274.6 189.5 211.7 235.8
vp9_inv_dct_dct_16x16_sub2_add_neon: 2064.0 1534.8 1719.4 1248.7
vp9_inv_dct_dct_16x16_sub4_add_neon: 2135.0 1477.2 1736.3 1249.5
vp9_inv_dct_dct_16x16_sub8_add_neon: 2446.7 1828.7 1993.6 1494.7
vp9_inv_dct_dct_16x16_sub12_add_neon: 2832.4 2118.3 2266.5 1735.1
vp9_inv_dct_dct_16x16_sub16_add_neon: 3211.7 2475.3 2523.5 1983.1
vp9_inv_dct_dct_32x32_sub1_add_neon: 756.2 456.7 862.0 553.9
vp9_inv_dct_dct_32x32_sub2_add_neon: 10682.2 8190.4 8539.2 6762.5
vp9_inv_dct_dct_32x32_sub4_add_neon: 10813.5 8014.9 8518.3 6762.8
vp9_inv_dct_dct_32x32_sub8_add_neon: 11859.6 9313.0 9347.4 7514.5
vp9_inv_dct_dct_32x32_sub12_add_neon: 12946.6 10752.4 10192.2 8280.2
vp9_inv_dct_dct_32x32_sub16_add_neon: 14074.6 11946.5 11001.4 9008.6
vp9_inv_dct_dct_32x32_sub20_add_neon: 15269.9 13662.7 11816.1 9762.6
vp9_inv_dct_dct_32x32_sub24_add_neon: 16327.9 14940.1 12626.7 10516.0
vp9_inv_dct_dct_32x32_sub28_add_neon: 17462.7 15776.1 13446.2 11264.7
vp9_inv_dct_dct_32x32_sub32_add_neon: 18575.5 17157.0 14249.3 12015.1
I.e. in general a very minor overhead for the full subpartition case due
to the additional loads and cmps, but a significant speedup for the cases
when we only need to process a small part of the actual input data.
In common VP9 content in a few inspected clips, 70-90% of the non-dc-only
16x16 and 32x32 IDCTs only have nonzero coefficients in the upper left
8x8 or 16x16 subpartitions respectively.
This is cherrypicked from libav commit
9c8bc74c2b.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This avoids reloading them if they haven't been clobbered, if the
first pass also was idct.
This is similar to what was done in the aarch64 version.
This is cherrypicked from libav commit
3c87039a40.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Since the same parameter is used for both input and output,
the name inout is more fitting.
This matches the naming used below in the dmbutterfly macro.
This is cherrypicked from libav commit
79566ec8c7.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This is one instruction less for thumb, and only have got
1/2 arm/thumb specific instructions.
This is cherrypicked from libav commit
e5b0fc170f.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This work is sponsored by, and copyright, Google.
Previously all subpartitions except the eob=1 (DC) case ran with
the same runtime:
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_16x16_sub16_add_neon: 3188.1 2435.4 2499.0 1969.0
vp9_inv_dct_dct_32x32_sub32_add_neon: 18531.7 16582.3 14207.6 12000.3
By skipping individual 4x16 or 4x32 pixel slices in the first pass,
we reduce the runtime of these functions like this:
vp9_inv_dct_dct_16x16_sub1_add_neon: 274.6 189.5 211.7 235.8
vp9_inv_dct_dct_16x16_sub2_add_neon: 2064.0 1534.8 1719.4 1248.7
vp9_inv_dct_dct_16x16_sub4_add_neon: 2135.0 1477.2 1736.3 1249.5
vp9_inv_dct_dct_16x16_sub8_add_neon: 2446.7 1828.7 1993.6 1494.7
vp9_inv_dct_dct_16x16_sub12_add_neon: 2832.4 2118.3 2266.5 1735.1
vp9_inv_dct_dct_16x16_sub16_add_neon: 3211.7 2475.3 2523.5 1983.1
vp9_inv_dct_dct_32x32_sub1_add_neon: 756.2 456.7 862.0 553.9
vp9_inv_dct_dct_32x32_sub2_add_neon: 10682.2 8190.4 8539.2 6762.5
vp9_inv_dct_dct_32x32_sub4_add_neon: 10813.5 8014.9 8518.3 6762.8
vp9_inv_dct_dct_32x32_sub8_add_neon: 11859.6 9313.0 9347.4 7514.5
vp9_inv_dct_dct_32x32_sub12_add_neon: 12946.6 10752.4 10192.2 8280.2
vp9_inv_dct_dct_32x32_sub16_add_neon: 14074.6 11946.5 11001.4 9008.6
vp9_inv_dct_dct_32x32_sub20_add_neon: 15269.9 13662.7 11816.1 9762.6
vp9_inv_dct_dct_32x32_sub24_add_neon: 16327.9 14940.1 12626.7 10516.0
vp9_inv_dct_dct_32x32_sub28_add_neon: 17462.7 15776.1 13446.2 11264.7
vp9_inv_dct_dct_32x32_sub32_add_neon: 18575.5 17157.0 14249.3 12015.1
I.e. in general a very minor overhead for the full subpartition case due
to the additional loads and cmps, but a significant speedup for the cases
when we only need to process a small part of the actual input data.
In common VP9 content in a few inspected clips, 70-90% of the non-dc-only
16x16 and 32x32 IDCTs only have nonzero coefficients in the upper left
8x8 or 16x16 subpartitions respectively.
Signed-off-by: Martin Storsjö <martin@martin.st>
This avoids reloading them if they haven't been clobbered, if the
first pass also was idct.
This is similar to what was done in the aarch64 version.
Signed-off-by: Martin Storsjö <martin@martin.st>
Since the same parameter is used for both input and output,
the name inout is more fitting.
This matches the naming used below in the dmbutterfly macro.
Signed-off-by: Martin Storsjö <martin@martin.st>
This work is sponsored by, and copyright, Google.
The implementation tries to have smart handling of cases
where no pixels need the full filtering for the 8/16 width
filters, skipping both calculation and writeback of the
unmodified pixels in those cases. The actual effect of this
is hard to test with checkasm though, since it tests the
full filtering, and the benefit depends on how many filtered
blocks use the shortcut.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_loop_filter_h_4_8_neon: 2.72 2.68 1.78 3.15
vp9_loop_filter_h_8_8_neon: 2.36 2.38 1.70 2.91
vp9_loop_filter_h_16_8_neon: 1.80 1.89 1.45 2.01
vp9_loop_filter_h_16_16_neon: 2.81 2.78 2.18 3.16
vp9_loop_filter_mix2_h_44_16_neon: 2.65 2.67 1.93 3.05
vp9_loop_filter_mix2_h_48_16_neon: 2.46 2.38 1.81 2.85
vp9_loop_filter_mix2_h_84_16_neon: 2.50 2.41 1.73 2.85
vp9_loop_filter_mix2_h_88_16_neon: 2.77 2.66 1.96 3.23
vp9_loop_filter_mix2_v_44_16_neon: 4.28 4.46 3.22 5.70
vp9_loop_filter_mix2_v_48_16_neon: 3.92 4.00 3.03 5.19
vp9_loop_filter_mix2_v_84_16_neon: 3.97 4.31 2.98 5.33
vp9_loop_filter_mix2_v_88_16_neon: 3.91 4.19 3.06 5.18
vp9_loop_filter_v_4_8_neon: 4.53 4.47 3.31 6.05
vp9_loop_filter_v_8_8_neon: 3.58 3.99 2.92 5.17
vp9_loop_filter_v_16_8_neon: 3.40 3.50 2.81 4.68
vp9_loop_filter_v_16_16_neon: 4.66 4.41 3.74 6.02
The speedup vs C code is around 2-6x. The 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). Disabling the early-exit in the asm
doesn't give a fair comparison either though, since the C code
only does the necessary calcuations for each row.
Based on START_TIMER/STOP_TIMER wrapping around a few individual
functions, the speedup vs C code is around 4-9x.
This is pretty similar in runtime to the corresponding routines
in libvpx. (This is comparing vpx_lpf_vertical_16_neon,
vpx_lpf_horizontal_edge_8_neon and vpx_lpf_horizontal_edge_16_neon
to vp9_loop_filter_h_16_8_neon, vp9_loop_filter_v_16_8_neon
and vp9_loop_filter_v_16_16_neon - note that the naming of horizonal
and vertical is flipped between the libraries.)
In order to have stable, comparable numbers, the early exits in both
asm versions were disabled, forcing the full filtering codepath.
Cortex A7 A8 A9 A53
vp9_loop_filter_h_16_8_neon: 597.2 472.0 482.4 415.0
libvpx vpx_lpf_vertical_16_neon: 626.0 464.5 470.7 445.0
vp9_loop_filter_v_16_8_neon: 500.2 422.5 429.7 295.0
libvpx vpx_lpf_horizontal_edge_8_neon: 586.5 414.5 415.6 383.2
vp9_loop_filter_v_16_16_neon: 905.0 784.7 791.5 546.0
libvpx vpx_lpf_horizontal_edge_16_neon: 1060.2 751.7 743.5 685.2
Our version is consistently faster on on A7 and A53, marginally slower on
A8, and sometimes faster, sometimes slower on A9 (marginally slower in all
three tests in this particular test run).
This is an adapted cherry-pick from libav commit
dd299a2d6d.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083 and
52d196fb30.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
This work is sponsored by, and copyright, Google.
The filter coefficients are signed values, where the product of the
multiplication with one individual filter coefficient doesn't
overflow a 16 bit signed value (the largest filter coefficient is
127). But when the products are accumulated, the resulting sum can
overflow the 16 bit signed range. Instead of accumulating in 32 bit,
we accumulate the largest product (either index 3 or 4) last with a
saturated addition.
(The VP8 MC asm does something similar, but slightly simpler, by
accumulating each half of the filter separately. In the VP9 MC
filters, each half of the filter can also overflow though, so the
largest component has to be handled individually.)
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_avg4_neon: 1.71 1.15 1.42 1.49
vp9_avg8_neon: 2.51 3.63 3.14 2.58
vp9_avg16_neon: 2.95 6.76 3.01 2.84
vp9_avg32_neon: 3.29 6.64 2.85 3.00
vp9_avg64_neon: 3.47 6.67 3.14 2.80
vp9_avg_8tap_smooth_4h_neon: 3.22 4.73 2.76 4.67
vp9_avg_8tap_smooth_4hv_neon: 3.67 4.76 3.28 4.71
vp9_avg_8tap_smooth_4v_neon: 5.52 7.60 4.60 6.31
vp9_avg_8tap_smooth_8h_neon: 6.22 9.04 5.12 9.32
vp9_avg_8tap_smooth_8hv_neon: 6.38 8.21 5.72 8.17
vp9_avg_8tap_smooth_8v_neon: 9.22 12.66 8.15 11.10
vp9_avg_8tap_smooth_64h_neon: 7.02 10.23 5.54 11.58
vp9_avg_8tap_smooth_64hv_neon: 6.76 9.46 5.93 9.40
vp9_avg_8tap_smooth_64v_neon: 10.76 14.13 9.46 13.37
vp9_put4_neon: 1.11 1.47 1.00 1.21
vp9_put8_neon: 1.23 2.17 1.94 1.48
vp9_put16_neon: 1.63 4.02 1.73 1.97
vp9_put32_neon: 1.56 4.92 2.00 1.96
vp9_put64_neon: 2.10 5.28 2.03 2.35
vp9_put_8tap_smooth_4h_neon: 3.11 4.35 2.63 4.35
vp9_put_8tap_smooth_4hv_neon: 3.67 4.69 3.25 4.71
vp9_put_8tap_smooth_4v_neon: 5.45 7.27 4.49 6.52
vp9_put_8tap_smooth_8h_neon: 5.97 8.18 4.81 8.56
vp9_put_8tap_smooth_8hv_neon: 6.39 7.90 5.64 8.15
vp9_put_8tap_smooth_8v_neon: 9.03 11.84 8.07 11.51
vp9_put_8tap_smooth_64h_neon: 6.78 9.48 4.88 10.89
vp9_put_8tap_smooth_64hv_neon: 6.99 8.87 5.94 9.56
vp9_put_8tap_smooth_64v_neon: 10.69 13.30 9.43 14.34
For the larger 8tap filters, the speedup vs C code is around 5-14x.
This is significantly faster than libvpx's implementation of the same
functions, at least when comparing the put_8tap_smooth_64 functions
(compared to vpx_convolve8_horiz_neon and vpx_convolve8_vert_neon from
libvpx).
Absolute runtimes from checkasm:
Cortex A7 A8 A9 A53
vp9_put_8tap_smooth_64h_neon: 20150.3 14489.4 19733.6 10863.7
libvpx vpx_convolve8_horiz_neon: 52623.3 19736.4 21907.7 25027.7
vp9_put_8tap_smooth_64v_neon: 14455.0 12303.9 13746.4 9628.9
libvpx vpx_convolve8_vert_neon: 42090.0 17706.2 17659.9 16941.2
Thus, on the A9, the horizontal filter is only marginally faster than
libvpx, while our version is significantly faster on the other cores,
and the vertical filter is significantly faster on all cores. The
difference is especially large on the A7.
The libvpx implementation does the accumulation in 32 bit, which
probably explains most of the differences.
This is an adapted cherry-pick from libav commits
ffbd1d2b00,
392caa65df,
557c1675cf and
11623217e3.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
This work is sponsored by, and copyright, Google.
The implementation tries to have smart handling of cases
where no pixels need the full filtering for the 8/16 width
filters, skipping both calculation and writeback of the
unmodified pixels in those cases. The actual effect of this
is hard to test with checkasm though, since it tests the
full filtering, and the benefit depends on how many filtered
blocks use the shortcut.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_loop_filter_h_4_8_neon: 2.72 2.68 1.78 3.15
vp9_loop_filter_h_8_8_neon: 2.36 2.38 1.70 2.91
vp9_loop_filter_h_16_8_neon: 1.80 1.89 1.45 2.01
vp9_loop_filter_h_16_16_neon: 2.81 2.78 2.18 3.16
vp9_loop_filter_mix2_h_44_16_neon: 2.65 2.67 1.93 3.05
vp9_loop_filter_mix2_h_48_16_neon: 2.46 2.38 1.81 2.85
vp9_loop_filter_mix2_h_84_16_neon: 2.50 2.41 1.73 2.85
vp9_loop_filter_mix2_h_88_16_neon: 2.77 2.66 1.96 3.23
vp9_loop_filter_mix2_v_44_16_neon: 4.28 4.46 3.22 5.70
vp9_loop_filter_mix2_v_48_16_neon: 3.92 4.00 3.03 5.19
vp9_loop_filter_mix2_v_84_16_neon: 3.97 4.31 2.98 5.33
vp9_loop_filter_mix2_v_88_16_neon: 3.91 4.19 3.06 5.18
vp9_loop_filter_v_4_8_neon: 4.53 4.47 3.31 6.05
vp9_loop_filter_v_8_8_neon: 3.58 3.99 2.92 5.17
vp9_loop_filter_v_16_8_neon: 3.40 3.50 2.81 4.68
vp9_loop_filter_v_16_16_neon: 4.66 4.41 3.74 6.02
The speedup vs C code is around 2-6x. The 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). Disabling the early-exit in the asm
doesn't give a fair comparison either though, since the C code
only does the necessary calcuations for each row.
Based on START_TIMER/STOP_TIMER wrapping around a few individual
functions, the speedup vs C code is around 4-9x.
This is pretty similar in runtime to the corresponding routines
in libvpx. (This is comparing vpx_lpf_vertical_16_neon,
vpx_lpf_horizontal_edge_8_neon and vpx_lpf_horizontal_edge_16_neon
to vp9_loop_filter_h_16_8_neon, vp9_loop_filter_v_16_8_neon
and vp9_loop_filter_v_16_16_neon - note that the naming of horizonal
and vertical is flipped between the libraries.)
In order to have stable, comparable numbers, the early exits in both
asm versions were disabled, forcing the full filtering codepath.
Cortex A7 A8 A9 A53
vp9_loop_filter_h_16_8_neon: 597.2 472.0 482.4 415.0
libvpx vpx_lpf_vertical_16_neon: 626.0 464.5 470.7 445.0
vp9_loop_filter_v_16_8_neon: 500.2 422.5 429.7 295.0
libvpx vpx_lpf_horizontal_edge_8_neon: 586.5 414.5 415.6 383.2
vp9_loop_filter_v_16_16_neon: 905.0 784.7 791.5 546.0
libvpx vpx_lpf_horizontal_edge_16_neon: 1060.2 751.7 743.5 685.2
Our version is consistently faster on on A7 and A53, marginally slower on
A8, and sometimes faster, sometimes slower on A9 (marginally slower in all
three tests in this particular test run).
Signed-off-by: Martin Storsjö <martin@martin.st>
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
Signed-off-by: Martin Storsjö <martin@martin.st>
This fixes crashes since 557c1675cf in linux PIC builds.
Previously, movrelx silently used r12 as helper register, which
doesn't work when r12 is the destination register.
Signed-off-by: Martin Storsjö <martin@martin.st>
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>
This fixes errors like this when building non-pic binaries with armv6
as baseline:
Error: invalid literal constant: pool needs to be closer
Signed-off-by: Martin Storsjö <martin@martin.st>
This work is sponsored by, and copyright, Google.
The filter coefficients are signed values, where the product of the
multiplication with one individual filter coefficient doesn't
overflow a 16 bit signed value (the largest filter coefficient is
127). But when the products are accumulated, the resulting sum can
overflow the 16 bit signed range. Instead of accumulating in 32 bit,
we accumulate the largest product (either index 3 or 4) last with a
saturated addition.
(The VP8 MC asm does something similar, but slightly simpler, by
accumulating each half of the filter separately. In the VP9 MC
filters, each half of the filter can also overflow though, so the
largest component has to be handled individually.)
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_avg4_neon: 1.71 1.15 1.42 1.49
vp9_avg8_neon: 2.51 3.63 3.14 2.58
vp9_avg16_neon: 2.95 6.76 3.01 2.84
vp9_avg32_neon: 3.29 6.64 2.85 3.00
vp9_avg64_neon: 3.47 6.67 3.14 2.80
vp9_avg_8tap_smooth_4h_neon: 3.22 4.73 2.76 4.67
vp9_avg_8tap_smooth_4hv_neon: 3.67 4.76 3.28 4.71
vp9_avg_8tap_smooth_4v_neon: 5.52 7.60 4.60 6.31
vp9_avg_8tap_smooth_8h_neon: 6.22 9.04 5.12 9.32
vp9_avg_8tap_smooth_8hv_neon: 6.38 8.21 5.72 8.17
vp9_avg_8tap_smooth_8v_neon: 9.22 12.66 8.15 11.10
vp9_avg_8tap_smooth_64h_neon: 7.02 10.23 5.54 11.58
vp9_avg_8tap_smooth_64hv_neon: 6.76 9.46 5.93 9.40
vp9_avg_8tap_smooth_64v_neon: 10.76 14.13 9.46 13.37
vp9_put4_neon: 1.11 1.47 1.00 1.21
vp9_put8_neon: 1.23 2.17 1.94 1.48
vp9_put16_neon: 1.63 4.02 1.73 1.97
vp9_put32_neon: 1.56 4.92 2.00 1.96
vp9_put64_neon: 2.10 5.28 2.03 2.35
vp9_put_8tap_smooth_4h_neon: 3.11 4.35 2.63 4.35
vp9_put_8tap_smooth_4hv_neon: 3.67 4.69 3.25 4.71
vp9_put_8tap_smooth_4v_neon: 5.45 7.27 4.49 6.52
vp9_put_8tap_smooth_8h_neon: 5.97 8.18 4.81 8.56
vp9_put_8tap_smooth_8hv_neon: 6.39 7.90 5.64 8.15
vp9_put_8tap_smooth_8v_neon: 9.03 11.84 8.07 11.51
vp9_put_8tap_smooth_64h_neon: 6.78 9.48 4.88 10.89
vp9_put_8tap_smooth_64hv_neon: 6.99 8.87 5.94 9.56
vp9_put_8tap_smooth_64v_neon: 10.69 13.30 9.43 14.34
For the larger 8tap filters, the speedup vs C code is around 5-14x.
This is significantly faster than libvpx's implementation of the same
functions, at least when comparing the put_8tap_smooth_64 functions
(compared to vpx_convolve8_horiz_neon and vpx_convolve8_vert_neon from
libvpx).
Absolute runtimes from checkasm:
Cortex A7 A8 A9 A53
vp9_put_8tap_smooth_64h_neon: 20150.3 14489.4 19733.6 10863.7
libvpx vpx_convolve8_horiz_neon: 52623.3 19736.4 21907.7 25027.7
vp9_put_8tap_smooth_64v_neon: 14455.0 12303.9 13746.4 9628.9
libvpx vpx_convolve8_vert_neon: 42090.0 17706.2 17659.9 16941.2
Thus, on the A9, the horizontal filter is only marginally faster than
libvpx, while our version is significantly faster on the other cores,
and the vertical filter is significantly faster on all cores. The
difference is especially large on the A7.
The libvpx implementation does the accumulation in 32 bit, which
probably explains most of the differences.
Signed-off-by: Martin Storsjö <martin@martin.st>
This avoids SIMD-optimized functions having to sign-extend their
line size argument manually to be able to do pointer arithmetic.
Also adjust parameter names to be "stride" everywhere.
This avoids SIMD-optimized functions having to sign-extend their
stride argument manually to be able to do pointer arithmetic.
Also adjust parameter names to be "stride" everywhere.
GNU as evaluates true as '-1' while Apple's variant and llvm's internal
assembler evaluate it as '1'. The best way to avoid this madness is to
eliminate boolean expressions instead of trying to fix it with
preprocessor directives. Use a direct formula to calculate the
required temporary space on the stack in
ff_put_vp8_{epel,bilin}{4,8,16}_h[246]v[246]_armv6().
Fixes a checkasm segfault in vp8dsp.mc when using llvm's internal
assembler for a non-Apple target.
Restore alphabetical order in lists, break overly long lines, do some
prettyprinting, add some explanatory section comments, group parts
together that belong together logically.
* commit '2008f76054906e9ff6bf744800af0e5a5bfe61be':
dca: remove unused decode_hf function and quant_d tables
Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>
Quite a bit faster than int32_to_float_fmul_array8_c calling
ff_int32_to_float_fmul_scalar_neon through FmtConvertContext.
Number of cycles per int32_to_float_fmul_array8 call while decoding
padded.dts on exynos5422:
before after change
cortex-a7: 1270 951 -25%
cortex-a15: 434 285 -34%
checkasm --bench cycle counts: cortex-a15 cortex-a7
int32_to_float_fmul_array8_c: 1730.4 4384.5
int32_to_float_fmul_array8_neon_c: 571.5 1694.3
int32_to_float_fmul_array8_neon: 374.0 1448.8
Interesting are the differences between
int32_to_float_fmul_array8_neon_c and int32_to_float_fmul_array8_neon.
The former is current behaviour of calling
ff_int32_to_float_fmul_scalar_neon repeatedly from the c function,
The raw numbers differ since checkasm uses different lengths than the
dca decoder.
The vector mode was deprecated in ARMv7-A/VFPv3 and various cpu
implementations do not support it in hardware. Vector mode code will
depending the OS either be emulated in software or result in an illegal
instruction on cpus which does not support it. This was not really
problem in practice since NEON implementations of the same functions are
preferred. It will however become a problem for checkasm which tests
every cpu flag separately.
Since this is a cpu feature newer cpu do not support anymore the
behaviour of this flag differs from the other flags. It can be only
activated by runtime cpu feature selection.
It is only (mis-)used to set the dsp fucntions clear_block(s). But
these functions always work on 16bits-wide elements, which make
the parameter useless and actually harmful, as it causes all content
on more than 8-bits to not use accelerated functions.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
* commit '256ef19844892c6cf8e0386e3287bae970ec6320':
h264: arm: use intra pred8x8 functions only for chroma_format_idc <= 1
Conflicts:
libavcodec/arm/h264pred_init_arm.c
See: 565cabf5c8
Merged-by: Michael Niedermayer <michael@niedermayer.cc>
* commit '581c7f0e12b1fa39f73d683e54d6ecda0772c5a9':
arm: make ff_mlp_filter_channel_arm and ff_mlp_rematrix_channel_arm position independent
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* commit 'f963f80399deb1a2b44c1bac3af7123e8a0c9e46':
arm: Use .data.rel.ro for const data with relocations
Conflicts:
configure
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* commit 'b280c6202b28b371a8d96850194fd69d7ad5dcc0':
arm: fft_vfp: Unify the behaviour in ff_fft_calc_vfp between arm/thumb
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* commit 'ae81576414f2d2083d3118fb4abe1ebc5a7a4c54':
arm: fft_vfp: Add a missing "endconst" when building in thumb mode
Merged-by: Michael Niedermayer <michaelni@gmx.at>
Don't include the function pointer table in the code segment
in arm mode.
This shouldn't have any significant performance effect. It does
end up as a few more instructions than before, for ARM, but
only at the entry to this function, not within the fft functions
themselves.
Signed-off-by: Martin Storsjö <martin@martin.st>
* commit '95c0cec03acec0a80cc1c7db48f3b2355d9e767b':
idctdsp: Add global function pointers for {add|put}_pixels_clamped functions
Conflicts:
libavcodec/arm/idctdsp_init_arm.c
libavcodec/dct.h
libavcodec/idctdsp.c
libavcodec/jrevdct.c
Merged-by: Michael Niedermayer <michaelni@gmx.at>
These function pointers already existed in the ARM code. Adding them globally
allows calls to the function pointers to access arch-optimized versions of the
functions transparently.
* commit 'efd26bedec9a345a5960dbfcbaec888418f2d4e6':
build: Add explanatory comments to (optimization) blocks in the Makefiles
Conflicts:
libavcodec/ppc/Makefile
libavcodec/x86/Makefile
Merged-by: Michael Niedermayer <michaelni@gmx.at>
This reduces code duplication and differences with the fork.
Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Initialise VC1DSPContext for parser as well as for decoder.
Note, the VC-1 code doesn't actually use the function pointer yet.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
* commit '7fb993d338d88f2f62e0a358b6c9f3eb9a3a08ac':
qpeldsp: Mark source pointer in qpel_mc_func function pointer const
Conflicts:
libavcodec/h264qpel_template.c
libavcodec/x86/cavsdsp.c
libavcodec/x86/rv40dsp_init.c
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* commit '6869612f5c7d4d2f20f69a5658328a761deadb1c':
arm: Macroize the test for 'setend' CPU instruction support
Conflicts:
libavcodec/arm/h264dsp_init_arm.c
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* commit '4de8b60684ce13dff3e3d372dae4f49b9e53f755':
idct: Move arm-specific declarations to a header in the arm directory
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* commit '7e18a727d2c2a19f22fcf68875d1b05fd2eafcef':
arm: cosmetics: Consistently use lowercase for shift operators
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* commit '87552d54d3337c3241e8a9e1a05df16eaa821496':
armv6: Accelerate ff_fft_calc for general case (nbits != 4)
Merged-by: Michael Niedermayer <michaelni@gmx.at>
The previous implementation targeted DTS Coherent Acoustics, which only
requires nbits == 4 (fft16()). This case was (and still is) linked directly
rather than being indirected through ff_fft_calc_vfp(), but now the full
range from radix-4 up to radix-65536 is available. This benefits other codecs
such as AAC and AC3.
The implementaion is based upon the C version, with each routine larger than
radix-16 calling a hierarchy of smaller FFT functions, then performing a
post-processing pass. This pass benefits a lot from loop unrolling to
counter the long pipelines in the VFP. A relaxed calling standard also
reduces the overhead of the call hierarchy, and avoiding the excessive
inlining performed by GCC probably helps with I-cache utilisation too.
I benchmarked the result by measuring the number of gperftools samples that
hit anywhere in the AAC decoder (starting from aac_decode_frame()) or
specifically in the FFT routines (fft4() to fft512() and pass()) for the
same sample AAC stream:
Before After
Mean StdDev Mean StdDev Confidence Change
Audio decode 2245.5 53.1 1599.6 43.8 100.0% +40.4%
FFT routines 940.6 22.0 348.1 20.8 100.0% +170.2%
Signed-off-by: Martin Storsjö <martin@martin.st>
The previous implementation targeted DTS Coherent Acoustics, which only
requires mdct_bits == 6. This relatively small size lent itself to
unrolling the loops a small number of times, and encoding offsets
calculated at assembly time within the load/store instructions of each
iteration.
In the more general case (codecs such as AAC and AC3) much larger arrays
are used - mdct_bits == [8, 9, 11]. The old method does not scale for
these cases, so more integer registers are used with non-unrolled versions
of the loops (and with some stack spillage). The postrotation filter loop
is still unrolled by a factor of 2 to permit the double-buffering of some
VFP registers to facilitate overlap of neighbouring iterations.
I benchmarked the result by measuring the number of gperftools samples
that hit anywhere in the AAC decoder (starting from aac_decode_frame())
or specifically in ff_imdct_half_c / ff_imdct_half_vfp, for the same
example AAC stream:
Before After
Mean StdDev Mean StdDev Confidence Change
aac_decode_frame 2368.1 35.8 2117.2 35.3 100.0% +11.8%
ff_imdct_half_* 457.5 22.4 251.2 16.2 100.0% +82.1%
Signed-off-by: Martin Storsjö <martin@martin.st>
