NVIDIA Publishes Their Next-Gen Tegra 4 Code

NVIDIA released Linux kernel patches this morning for supporting their next-gen "Tegra 4" SoC under Linux. A few details were revealed within the code commits.

The set of nine patches for initial Linux kernel enablement of the new Tegra System-On-a-Chip provides the minimal support necessary for the Linux kernel to boot up into a shell console while the rest of the enablement code will come later.

The patches refer to the new Tegra SoC as the "Tegra 114" with the development/evaluation boards being called "Dalmore" and "Pluto" for this Cortex-A15 MP platform. This is the hardware that's coming to market as the Tegra 4 "Wayne" generation.

The Linux NVIDIA Tegra support does allow for a single kernel image to handle NVIDIA's Dalmore T114, Pluto T114, and Cardhu T30. The Cardhu is the current-generation NVIDIA Tegra 3 reference board.

The patches can be found on the kernel mailing list. With the merge window on Linux 3.8 closing soon, this work likely won't be merged into the mainline Linux kernel until the Linux 3.9 series in H1'2013. This initial NVIDIA Tegra 4 hardware support just amounts to a couple hundred lines of new code on top of the existing Tegra kernel code.

This next-generation Tegra 4 "Wayne" platform consists of a quad-core ARM Cotex-A15 processor plus a low-power companion core, 72 GPU cores, hardware-based VP8 encoding/decoding support, OpenGL 4.x support, and is to be manufactured on a 28nm process.

It's reported that the Tegra 4 will be about six times faster than the Tegra 3 or twenty times faster than the Tegra 2. This really isn't a surprise since the ARM Cortex-A15 is so damn fast and a huge upgrade over the Cortex-A9 as found with the current NVIDIA ARM hardware. I have been very pleased with the ARM Cortex-A15 Linux performance as tested up to this point using a Samsung Exynos 5 Dual as found in the Samsung Chromebook.