Run the AV1 and VP9 codecs on Arm Linux
Introduction
Build and Run the AV1 codec
Build and Run the VP9 codec
Review
Next Steps

  Log an issue

  Fork and edit

  Discuss on Discord

What is the VP9 codec?

VP9 is a next-generation video compression format developed by the WebM Project .

The reference software implementation for VP9 encoding and decoding can be found in the libvpx library. Similar to AV1, significant effort has been made to optimize the libvpx implementation on Arm platforms. This includes making use of the Neon and SVE2 vector extensions available on Arm Neoverse platforms. You can learn more on Google for Developers .

Download and build libvpx from source

Download the libvpx source code:

    

        
        
            git clone https://chromium.googlesource.com/webm/libvpx
cd libvpx

The configuration allows for different options to be enabled and disabled.

To see the options run:

    

        
        
            ./configure --help

Configure and build with VP8 disabled, VP9 enabled, and unit tests enabled:

    

        
        
            ./configure --disable-vp8 --enable-vp9 --enable-vp9-highbitdepth --enable-unit-tests
make -j$(nproc)

For more information, refer to the README .

Run libvpx unit tests

You can run libvpx unit tests using make. For example, to run the decoder test run:

    

        
        
            make testdata
make test TEST_NAME=vp9_decoder_test

The output is:

    

        
        Note: Google Test filter = *-:NEON_I8MM.*:NEON_I8MM/*:SVE.*:SVE/*:SVE2.*:SVE2/*
Note: This is test shard 1 of 10.
[==========] Running 1592 tests from 132 test suites.
[----------] Global test environment set-up.
[----------] 1 test from ByteAlignmentTest
[ RUN      ] ByteAlignmentTest.SwitchByteAlignment
[       OK ] ByteAlignmentTest.SwitchByteAlignment (5248 ms)
[----------] 1 test from ByteAlignmentTest (5248 ms total)

[----------] 4 tests from EncodeAPI
[ RUN      ] EncodeAPI.SetRoi
[       OK ] EncodeAPI.SetRoi (4 ms)
[ RUN      ] EncodeAPI.DynamicDeadlineChange
[       OK ] EncodeAPI.DynamicDeadlineChange (4 ms)
[ RUN      ] EncodeAPI.Buganizer311294795
[       OK ] EncodeAPI.Buganizer311294795 (212 ms)
[ RUN      ] EncodeAPI.Buganizer329179808RowMT1BitDepth10
[       OK ] EncodeAPI.Buganizer329179808RowMT1BitDepth10 (26 ms)
[----------] 4 tests from EncodeAPI (248 ms total)

< output omitted >

Performance Benchmarking

You can benchmark either the encoding or decoding processes. In this section, you will focus on encoding.

For Performance benchmarking, you can select video encoding either on-demand or live stream.

Note

The video files are large so adjust the paths to the y4m files in the commands below. You can also copy them to the current directory.

On-demand video encoding

For on-demand encoding, you can experiment different number of processors and monitor performance.

For example, run with --good and use the --cpu-used argument to vary the number from 2 to 6.

Note

The naming of --cpu-used flag is to trade-off encoding speed for resulting video quality/compression, not to determine how many CPUs to use for parallel encoding. Lower numbers indicate better quality but longer encoding time. For VoD a reasonable range of quality settings is 0-4 inclusive.

Run standard bit depth and change the CPU count, then see the results using:

    

        
        
            ./vpxenc --good --cpu-used=2 --profile=0 --bit-depth=8 -o output.mkv Bosphorus_1920x1080_120fps_420_8bit_YUV.y4m

Try the above command with different --cpu-used values:

    

        
        
            ./vpxenc --good --cpu-used=6 --profile=0 --bit-depth=8 -o output.mkv Bosphorus_1920x1080_120fps_420_8bit_YUV.y4m

You can do the same for high bit depth:

    

        
        
            ./vpxenc --good --cpu-used=2 --profile=2 --bit-depth=10 -o output.mkv Bosphorus_3840x2160_120fps_420_10bit_YUV.y4m

Live stream video encoding

For live stream encoding, you can experiment different number of processors and monitor performance using the --cpus-used in the range from 5 to 9.

Note

The naming of --cpu-used flag is to trade-off encoding speed for resulting video quality/compression, not to determine how many CPUs to use for parallel encoding. Lower numbers indicate better quality but longer encoding time. For live stream a reasonable range of quality settings is 5-9.

For standard bit depth with 8 CPUs, run:

    

        
        
            ./vpxenc --rt --cpu-used=8 --profile=0 --bit-depth=8 -o output.mkv Bosphorus_1920x1080_120fps_420_8bit_YUV.y4m

For high bit depth, run:

    

        
        
            ./vpxenc --rt --cpu-used=8 --profile=2 --bit-depth=10 -o output.mkv Bosphorus_3840x2160_120fps_420_10bit_YUV.y4m

View Results

The encoding frame rate (frames per second) for the video files is output at the end of each run.

Shown below is the example output from running the libvpx codec on the 8-bit FHD sample video file with different –cpu-used settings to compare the encoding time.

The output from running the codec with --cpu-used=2 is similar to:

–cpu-used=2

    

        
        Pass 1/2 frame  600/601   129816B    1730b/f   51926b/s   29486 ms (20.35 fps)
Pass 2/2 frame  600/576   625752B  135925 ms 4.41 fps [ETA  0:00:05]     196F    260F    220F    227F    242F    166F     66F   4066F    316F    386F    337F    24Pass 2/2 frame  600/600   639729B    8529b/f  255891b/s  109956 ms (5.46 fps)

The output from running the codec with --cpu-used=6 is similar to:

    

        
        
            Pass 1/2 frame  600/601   129816B    1730b/f   51926b/s   29446 ms (20.38 fps)
Pass 2/2 frame  600/576   629484B  101687 ms 5.90 fps [ETA  0:00:04]     344F    249F    215F    206F    195F    168F     70F   5095F    186F    230F    206F    21Pass 2/2 frame  600/600   643500B    8580b/f  257400b/s   74431 ms (8.06 fps)

As demonstrated in these examples, --cpu-used=6 is 30% faster than --cpu-used=2.

Back
Next