A Reverse-Engineered Tegra Video Decode Driver Steps Closer To Mainline
A video decoder driver for the NVIDIA Tegra is closer to the mainline kernel, but is focused on the (older) Tegra 2 chips.
Dmitry Osipenko has been leading the charge to reverse-engineer video decoding support for NVIDIA Tegra SoCs and his tegra-vde driver continues to be revised for hopeful acceptance into the mainline Linux kernel.
This driver is initially proposed for the kernel's staging area and when paired with a libvdpau-tegra user-space driver written by Dmitry, is capable of providing accelerated video decoding for the Tegra 2 SoC. This video decode driver has been tested on devices like the Toshiba AC100 and Acer A500 though should work with any Tegra 2 (Tegra20) device.
The latest version of the tegra-vde driver published today can be found on the kernel mailing list while the user-space VDPAU driver for interfacing with tegra-vde can be found on GitHub under the grate-driver project.
While NVIDIA is more friendly towards open-source support with their newer Tegra SoCs with contributions to Nouveau and more, we haven't seen them do much in the space of open-source video acceleration aside from the VDPAU library itself being open-source albeit for desktop GPUs they are no longer pushing that as much as cuvid/nvdecode for the future.
Dmitry Osipenko has been leading the charge to reverse-engineer video decoding support for NVIDIA Tegra SoCs and his tegra-vde driver continues to be revised for hopeful acceptance into the mainline Linux kernel.
This driver is initially proposed for the kernel's staging area and when paired with a libvdpau-tegra user-space driver written by Dmitry, is capable of providing accelerated video decoding for the Tegra 2 SoC. This video decode driver has been tested on devices like the Toshiba AC100 and Acer A500 though should work with any Tegra 2 (Tegra20) device.
The latest version of the tegra-vde driver published today can be found on the kernel mailing list while the user-space VDPAU driver for interfacing with tegra-vde can be found on GitHub under the grate-driver project.
While NVIDIA is more friendly towards open-source support with their newer Tegra SoCs with contributions to Nouveau and more, we haven't seen them do much in the space of open-source video acceleration aside from the VDPAU library itself being open-source albeit for desktop GPUs they are no longer pushing that as much as cuvid/nvdecode for the future.
9 Comments