VideoLAN Publishes Dav2d For Open-Source AV2 Decoder - Phoronix

VideoLAN has made a significant move in the video codec space by publishing dav2d, their open-source implementation of an AV2 decoder. This comes at an interesting time, as the Alliance for Open Media (AOMedia) has yet to finalize the AV2 specification, which was originally targeted for release by the end of 2025. Despite the specification remaining in draft status, VideoLAN's developers have been working on dav2d for months and have now made the code public via the VideoLAN GitLab repository.

The dav2d decoder follows in the footsteps of VideoLAN's successful dav1d implementation for AV1. Like its predecessor, dav2d is designed as a CPU-based decoder that prioritizes both performance and correctness. The cross-platform nature of the decoder ensures it can be deployed across various operating systems and architectures, making it a versatile solution for developers and content consumers alike.

Technical Implementation

From a technical standpoint, dav2d inherits much of its architecture from dav1d, which has proven to be a highly efficient decoder. The VideoLAN team has emphasized that while the code is now public, significant work remains to complete the C implementation, finalize the API, and port the decoder to additional platforms. The developers have also noted that substantial performance optimizations are still possible, particularly around AVX2 and other instruction set architectures for ARM, RISC-V, and other emerging processor architectures.

The availability of an open-source AV2 decoder is particularly significant given the industry's trajectory. Video codecs have become increasingly important as content resolution and frame rates continue to climb. The transition from H.264 to H.265/HEVC and then to AV1 demonstrated the industry's need for more efficient compression technologies that can deliver high-quality video with lower bandwidth requirements.

Performance Considerations

While specific benchmark data for dav2d is not yet available, we can draw some informed predictions based on the evolution from dav1d. The original dav1d decoder demonstrated impressive performance characteristics, often outperforming other open-source AV1 decoders while maintaining high accuracy. The following table compares expected performance characteristics based on the progression from previous codecs:

Decoder	Resolution	Frame Rate	CPU Usage	Memory Footprint	Quality
H.264 (x264)	4K	60fps	15-25%	150-200MB	Good
H.265 (x265)	4K	60fps	25-40%	200-300MB	Very Good
AV1 (dav1d)	4K	60fps	20-35%	180-250MB	Excellent
AV2 (dav2d)	4K	60fps	15-30%	160-220MB	Excellent+

Note: These are estimated values based on codec efficiency improvements and actual dav1d performance data. Actual dav2d performance may vary.

The reduced CPU usage and memory footprint in the projected AV2 decoder are particularly noteworthy for homelab enthusiasts and server administrators. Lower power consumption means reduced heat output and potentially lower operating costs, especially when running multiple simultaneous decoding streams.

Platform Compatibility

One of the key advantages of dav2d is its cross-platform nature. The decoder is designed to work across:

• x86-64 (Intel/AMD) with AVX2 optimization potential
• ARM (AArch64) with NEON optimization potential
• RISC-V (emerging architecture)
• Other platforms as porting progresses

This broad compatibility ensures that dav2d can be integrated into a wide range of applications, from media players like VLC to streaming services and content delivery networks. The open-source nature of the decoder also means that it can be customized and optimized for specific use cases, which is particularly valuable for specialized hardware or software environments.

Build Recommendations

For homelab enthusiasts interested in experimenting with dav2d, here are some build recommendations:

Software Requirements

• Compiler with C11 support (GCC 7+, Clang 5+, MSVC 2019+)
• CMake 3.10+ for build system
• NASM or YASM for assembly optimizations
• Git for source code retrieval

Hardware Recommendations

Use Case	Recommended CPU	Memory	Storage
Development/Testing	4+ core modern CPU (Ryzen 5/7, Intel i5/i7)	16GB+	10GB+ SSD space
Production Streaming Server	8+ core server CPU (EPYC, Xeon)	32GB+	50GB+ NVMe space
Low-Power Homelab	ARM-based SBC (Raspberry Pi 4/5, Rockchip)	8GB+	32GB+ SD card

Compilation Instructions

For those interested in building dav2d from source, the general process involves:

1. Clone the repository: git clone https://code.videolan.org/videolan/dav2d.git
2. Create build directory: mkdir build && cd build
3. Configure with CMake: cmake ..
4. Build: make -j$(nproc)
5. Install: sudo make install

The VideoLAN team has indicated that they're working on providing pre-built binaries for various platforms to simplify the adoption process.

Future Implications

The publication of dav2d serves multiple important purposes in the video codec ecosystem:

1. Early Adoption: Provides a reference implementation that developers can begin experimenting with before the final AV2 specification is released.
2. Specification Feedback: Early implementations can help identify potential issues or areas for improvement in the AV2 specification.
3. Performance Benchmarking: Establishes a baseline for AV2 decoder performance that can be improved upon as optimizations are developed.
4. Platform Preparation: Allows hardware manufacturers to begin preparing their platforms for AV2 support.

The timing of this release is particularly interesting, as it may indicate that AOMedia is nearing completion of the AV2 specification. The availability of a production-ready decoder suggests that the specification is mature enough for implementation, even if it hasn't been formally finalized.

For the broader computing community, the emergence of AV2 represents the next step in video compression technology. With improved efficiency over AV1, we can expect:

• Higher quality streaming at the same bitrates
• Lower bandwidth requirements for 4K/8K content
• Better performance on lower-end hardware
• Reduced storage requirements for video archives

The open-source nature of dav2d ensures that these benefits will be accessible to everyone, not just those tied to proprietary ecosystems. This aligns with VideoLAN's mission of promoting open multimedia standards and technologies.

Those interested in exploring dav2d can find the code at the official VideoLAN GitLab repository. As the decoder continues to mature, we can expect to see integration into popular media players and streaming applications, bringing AV2 decoding to mainstream users.

The progress of dav2d also highlights the importance of open-source development in the codec space. While proprietary codecs have historically dominated the market, open-source implementations like dav1d and now dav2d are proving that high-performance, efficient decoding can be achieved through collaborative development. This bodes well for the future of video technology, as it ensures that innovation remains accessible and transparent.