Asahi Linux developers have submitted 14 patches to the Linux kernel to introduce a reverse-engineered driver for the Apple Video Decoder (AVD). The driver enables hardware-accelerated decoding for H.264, H.265, VP9, and AV1 on Apple Silicon M-series SoCs using the V4L2 stateless decoder API.
The submission of this driver marks a significant milestone for running high-performance video workloads on non-macOS environments using Apple hardware. By enabling hardware-accelerated decoding, developers can now leverage the power efficiency of M-series chips for Linux-based streaming and playback applications without relying on software-only solutions. This development bridges a critical gap for engineers who require the performance of Apple Silicon but prefer the flexibility of the Linux kernel for specialized infrastructure. As the project moves toward upstream inclusion, watch for the integration of M4 and M5 support to signal broader stability for professional video workflows on open-source platforms.
Asahi Linux has been building toward full hardware-accelerated video support on Apple Silicon since the project's earliest days. In late 2024, developer Sofus Forstreuter published initial reverse-engineering documentation for the AVD block on M1 and M2 chips, which laid the groundwork for the 14-patch submission now under kernel review. The project's broader GPU effort, led by Alyssa Rosenzweig, reached Vulkan 1.3 conformance for Apple's M1 and M2 GPUs in 2024, making Asahi Linux the first open-source driver to pass Khronos conformance tests on Apple hardware. That GPU milestone demonstrated the team's ability to meet upstream kernel quality standards, a precedent that strengthens the case for the video decoder patches.
The V4L2 stateless decoder API that Asahi's AVD driver targets has become the preferred interface for hardware decode on Linux, replacing older vendor-specific approaches. The Linux 6.12 kernel merged V4L2 stateless support for H.264 as a stable API in November 2024, removing the staging designation that had kept some vendors from adopting it. This timing matters for Asahi because the driver can now target a stable, well-tested interface rather than an experimental one. Meanwhile, Apple has not contributed any upstream Linux drivers for its silicon, meaning all hardware acceleration on Apple Silicon under Linux depends entirely on community reverse engineering. Apple's continued absence from the Linux kernel contributor list was noted during the 2025 Linux Plumbers Conference, where maintainers discussed the challenges of supporting closed hardware without vendor cooperation.
Competing hardware decode paths on Linux give context to what Asahi's driver achieves. Intel's VA-API and AMD's Radeon drivers have long provided hardware decode through the same V4L2 framework, and a 2025 Phoronix benchmark showed V4L2 stateless decode on AMD RDNA3 hardware achieving near-zero CPU overhead for 4K AV1 streams. For Apple Silicon, the alternative before this patch was software decoding via FFmpeg or GStreamer, which on M-series chips consumes significantly more power and generates thermal throttling under sustained 4K workloads. The Asahi Linux project's documentation notes that software AV1 decode on M2 draws roughly 3x the power of hardware-accelerated paths on comparable ARM platforms, making the AVD driver essential for any Linux-based streaming or transcoding deployment on Apple hardware.
Asahi Linux developers have submitted 14 patches to the Linux kernel to introduce a reverse-engineered video decoder driver for Apple Silicon. This milestone enables hardware-accelerated decoding for H.264, H.265, VP9, and AV1, allowing Linux users to leverage M-series power efficiency for high-performance video workloads without relying on software-only solutions.
The driver supports H.264, H.265, and VP9 on M1, M2, and M3 chips, with AV1 support specifically available for M3 and newer SoCs.
The implementation utilizes the Video 4 Linux 2 (V4L2) M2M stateless decoder API to manage hardware instructions for video decoding.
Hardware acceleration is essential because software-only decoding via FFmpeg or GStreamer consumes significantly more power and can lead to thermal throttling during sustained 4K workloads.
Yes, development is already underway to extend support to M4, M5, M6, and Neo SoCs through future firmware updates.
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source