NVIDIA: New Features, Support For Linux For Tegra

NVIDIA has quietly released new versions of its Linux 4 Tegra software platform that's currently targeting their "Cardhu" and "Ventana" platforms.

Released last week was NVIDIA's Linux 4 Tegra R16.2, even after the R16.1 release barely got any publicity at all. The NVIDIA downloads area is still reflecting the old L4T R15 release from June, but while stumbling across the page today after wanting to load up new software on the NVIDIA Tegra 3 "Cardhu" tablet, I was pleased to see R16 is actually available.

The NVIDIA Linux For Tegra R16.1 release introduced full support for the ARM hardfp/armhf ABI while deprecating the ABI support for softfp/armel. This is for both the drivers and sample file-system. With R16, the sample file-system has been updated to Ubuntu 12.04 LTS rather than the old softfp version of Ubuntu 11.04. This is nice to see finally happen as there were Tegra hardfp drivers back in May but only now is an "official" sample file-system based upon hard floating-point binaries for ARM now out from NVIDIA. Hard floating performance is huge.

Back in August I shared some Tegra 3 vs. ODROID-X (Exynos 4) benchmarks along with my original NVIDIA Tegra 3 Cardhu tablet benchmarks from Ubuntu Linux. New benchmarks of NVIDIA Tegra 3 Cardhu will be forthcoming with the new 12.04 packages, especially with comparing to the Samsung Exynos 5 Dual that sports an ARM Cortex-A15. It won't be until the Tegra 4 next year that NVIDIA moves from Cortex-A19 to Cortex-A15.

The R16.2 release provides additional kernel stability improvements, the Tegra X11 driver now supports the X video ABI version 13, the NVIDIA Tegra FUSE driver has been renamed to "tegra_fuse", and the EGL/GLES libraries have versioned ELF SONAME entries.

Linux For Tegra can be downloaded from developer.nvidia.com.