Benjamin Gaignard
February 17, 2021
Reading time:
One of the platforms we have been working on at Collabora is VeriSilicon's Hantro Codec. This video IP is present on a number of popular SoCs (Rockchip, i.MX8, Microchip) and is marketed as a small and power efficient device, but it also has a feature that makes it specially attractive for open source developers: it is a stateless accelerator.
Stateless devices do not need firmware to operate, making them more robust and better suited for open source platforms, where having full control over the system is desirable. In this case, the support is split in two: a kernel driver (which is provided by a Video4Linux2 Hantro driver), and a userspace component (which can be provided by frameworks such as GStreamer and FFMPEG).
Our recent efforts on the Hantro kernel driver have resulted in the addition of H.264 decoding support and multiple performance improvements. Continuing this work, we are now introducing High Efficiency Video Coding (HEVC), also known as H.265, decoding support on NXP's i.MX8 M chipset.
Unlike the currently supported codecs (JPEG, MPEG-2, VP8 and H.264), HEVC doesn't rely on the G1 hardware block but on the second video processor unit: the G2.
For this first step, the driver supports the basic HEVC features up to level 5.1. Enhance features like 10-bits depth per sample with 4:2:0 chroma sampling, scaling or tile decoding could be added later. Another possible evolution is to take benefit of the hardware capability to use compressed buffers to limit the memory bandwidth consumption.
Supporting HEVC on the Hantro driver will help mature the HEVC V4L2 stateless API enough to be able to remove it from the staging directory. Getting HEVC as a stable API is an important requirement for the userland stacks that are relying on it.
More work is in the pipeline in the kernel side and in GStreamer, to improve CODEC support. For instance, we are working on VP9 decoding and V4L2 hardware accelerated encoding. Want to know more? Get in touch!
22/04/2026
Using our XR expertise, Collabora created a standalone XR experience for our 1% for the Planet partner, SOMAR, to showcase the direct impact…
17/04/2026
BitNet-style ternary brings LLM inference to ExecuTorch via its Vulkan backend, enabling much smaller, bandwidth-efficient models with portable…
23/03/2026
PanVK’s new framebuffer abstraction for Mali GPUs removes OpenGL-specific constraints, unlocking more flexible tiled rendering features…
02/03/2026
Get the recap of Nicolas Frattaroli's FOSDEM talk detailing Rockchip’s mainline progress, including Vulkan 1.4 and NPU support as a vital…
02/12/2025
As an active member of the freedesktop community, Collabora was busy at XDC 2025. Our graphics team delivered five talks, helped out in…
24/11/2025
LE Audio introduces a modern, low-power, low-latency Bluetooth® audio architecture that overcomes the limitations of classic Bluetooth®…