The input is laid out in 16 segments, of which 13 actually need to be
loaded. There are no really efficient ways to deal with this:
1) If we load 8 segments wit unit stride, then narrow to 16 segments with
right shifts, we can only get one half-size vector per segment, or just 2
elements per vector (EMUL=1/2) - at least with 128-bit vectors.
This ends up unsurprisingly about as fas as the C code.
2) The current approach is to load with strides. We keep that approach,
but improve it using three 4-segmented loads instead of 12 single-segment
loads. This divides the number of distinct loaded addresses by 4.
3) A potential third approach would be to avoid segmentation altogether
and splat the scalar coefficient into vectors. Then we can use a
unit-stride and maximum EMUL. But the downside then is that we have to
multiply the 3 (of 16) unused segments with zero as part of the
multiply-accumulate operations.
In addition, we also reuse vectors mid-loop so as to increase the EMUL
from 1 to 2, which also improves performance a little bit.
Oeverall the gains are quite small with the device under test, as it does
not deal with segmented loads very well. But at least the code is tidier,
and should enjoy bigger speed-ups on better hardware implementation.
Before:
ps_hybrid_analysis_c: 1819.2
ps_hybrid_analysis_rvv_f32: 1037.0 (before)
ps_hybrid_analysis_rvv_f32: 990.0 (after)
This stores the constant coefficients deinterleaved, so that they can be
loaded directly with NF=0. Unfortunately, we cannot optimise loading the
input, due to insufficient memory alignment (not 32-bit).
Before:
g722_apply_qmf_c: 82.5
g722_apply_qmf_rvv_i32: 78.2
After:
g722_apply_qmf_c: 82.5
g722_apply_qmf_rvv_i32: 65.2
Gathers are (unsurprisingly) a notable exception to the rule that R-V V
gets faster with larger group multipliers. So roll the function to speed
it up.
Before:
vector_fmul_reverse_fixed_c: 2840.7
vector_fmul_reverse_fixed_rvv_i32: 2430.2
After:
vector_fmul_reverse_fixed_c: 2841.0
vector_fmul_reverse_fixed_rvv_i32: 962.2
It might be possible to further optimise the function by moving the
reverse-subtract out of the loop and adding ad-hoc tail handling.
When encoders don't support global header like MediaCodec, FLV
muxer needs to add extract_extradata bsf automatically. The codec
list doesn't include VP9 since it's not supported by
extract_extradata.
Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>
Validate that a hw_frames_ctx is available before using it for
the AVHWAccel.free_frame_priv callback, and don't require it to
be present when the callback is not in use by the HWAccel.
v2: check for free_frame_priv (Hendrik)
v3: return EINVAL (Christoph Reiter)
v4: better commit message (Hendrik)
v5: fix typo with missed frames_ctx (Lynne)
See[1]: https://github.com/msys2/MINGW-packages/pull/19050
Fixes: be07145109 ("avcodec: add AVHWAccel.free_frame_priv callback")
CC: Lynne <dev@lynne.ee>
CC: Christoph Reiter <reiter.christoph@gmail.com>
Signed-off-by: Dmitry Rogozhkin <dmitry.v.rogozhkin@intel.com>
For fate-h264_mp4toannexb_ticket5927 and
fate-h264_mp4toannexb_ticket5927_2, they work by accident
previously. The sample file has two 'avc1' entries, and video
samples use the second one. It means packets should be decoded with
new extradata in side data. Before this patch, only extradata was
kept in the output, new extradata has been dropped. The output can
be decoded because the two extradata are almost the same, except
level indication. This patch fixed the issue, and add another
fate test.
Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>
If there is a single group of SPS/PPS before an IDR frame, but no
SPS/PPS after that, we will miss the chance to reset
idr_sps_seen/idr_pps_seen. No SPS/PPS are inserted afterwards.
This patch saves in-band SPS/PPS and insert them before IDR frames
when necessary.
Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>
start_code_size depends on whether PS comes from out-of-band or
in-band. Make the code more readable.
Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>
This avoids SEI and IDR recovery flags affecting each other
Also eliminate litteral numbers from recovery handling
This should make the code clearer
Improves: tickets/4738/tickets_cut.ts
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This is only supported at compilation time. If Zfhmin is supported, then
conversions are fast, which is what the flag is used for. At this time,
run-tiem detection is not possible, as in not supported by Linux. But even
if it were, the current FFmpeg approach seems unable to deal with it (same
problem as on x86, really).
In this case, the inner loop computing the scalar product can be reduced
to just one multiplication and one sum even with 128-bit vectors. The
result is a lot simpler, but also brings more modest performance gains:
flac_lpc_16_13_c: 15241.0
flac_lpc_16_13_rvv_i32: 11230.0
flac_lpc_16_16_c: 17884.0
flac_lpc_16_16_rvv_i32: 12125.7
flac_lpc_16_29_c: 27847.7
flac_lpc_16_29_rvv_i32: 10494.0
flac_lpc_16_32_c: 30051.5
flac_lpc_16_32_rvv_i32: 10355.0
The entire set of 32 coefficients and corresponding past 32 samples can
fit in a single vector (with LMUL=8) exactly, but... since widening
double the needed vector sizes, we still end up too short with 128-bit
vectors. This adds a very simple version for future 256+-bit hardware,
and for pred_orders values up to 16, and a bit more involved loop for
for 128-bit hardware with pred_orders between 17 and 32.
With 128-bit hardware, the benchmarks look like this:
flac_lpc_32_13_c: 30152.0
flac_lpc_32_13_rvv_i32: 10244.7
flac_lpc_32_16_c: 37314.2
flac_lpc_32_16_rvv_i32: 10126.2
flac_lpc_32_29_c: 61910.0
flac_lpc_32_29_rvv_i32: 14495.2
flac_lpc_32_32_c: 68204.0
flac_lpc_32_32_rvv_i32: 13273.7
Rémi Denis-Courmont
Paul B Mahol
James Almer
Michael Niedermayer
Zhao Zhili
Dmitry Rogozhkin
Jun Zhao
Dawid Kozinski
Anton Khirnov