In this issue
- Nvidia Jetson Nano, $99 AI computer
- Nvidia Linux4Tegra (L4T) 32.1
- Arduino 1.8.9 supports arm64
- “Arm on Arm” summit at Linaro Connect
- How usable is desktop Linux on ARM? (Linux Journal)
- LLVM toolchain 8.0.0
- Using multiple CPU cores with GNU Parallel
- Bootstrapping Windows 10 on arm64 devices
Nvidia Jetson Nano, $99 AI computer
The Jetson Nano is a $99 development system from Nvidia. It features 4 GB of memory, a quad-core A57 CPU, and most importantly for the applications envisioned a 128-core Maxwell GPU with software support for AI and machine learning applications like TensorFlow, PyTorch, Caffe, Keras, and MXNet.
Nvidia Linux4Tegra (L4T) 32.1
Linux4Tegra (L4T) 32.1 is a Linux distribution for the Jetson AGX Xavier, Jetson TX2, Jetson TX2i, and Jetson Nano. It is derived from Ubuntu 18.04, and uses a kernel based on Linux 4.9. L4T is distributed as part of their Jetpack system, which also includes software and hardware drivers.
A number of patches for the Tegra line have been upstreamed to the mainline Linux kernel. Work is underway to boot Fedora 30 on these systems.
Arduino 1.8.9 supports arm64
The Arduino software development kit has been updated in version 1.8.9 to include support for arm64, with the Raspberry Pi Model 3 and Nvidia Tegra systems the first targeted systems. This SDK enables developers to program their Arduino and update its software directly from their arm64 device.
“Arm on Arm” summit at Linaro Connect
Marvell, Arm, and Linaro are hosting an “Arm on Arm” summit at the Linaro Connect conference in Bangkok, Thailand. The goal of the event is to describe how developers can build and maintain a self-hosted development system for Arm development completely on Arm devices, without any need for emulation.
How usable is desktop Linux on ARM? (Linux Journal)
Bryan Lunduke offers an opinionated viewpoint on the usability of Linux on the desktop on Arm processors in this Linux Journal podcast.
LLVM toolchain 8.0.0
LLVM version 8.0.0 is out. This compiler family for C and C++ has a set of general improvements in this version across all architectures, and features new support for aarch64 (arm64) to mitigate against speculative attacks.
Using multiple CPU cores with GNU parallel
GNU Parallel is an effective and simple tool for using multiple cores and executing applications and pipelines in parallel. It is notably interesting on arm64 systems due to the relatively high core counts found in these machines.
A tutorial from John Westlund at Xensoft describes how GNU Parallel can even speed up single-threaded Linux commands, if the work those commands tackle can be appropriately divided up.
Bootstrapping Windows 10 on arm64 devices
Ben (imbushuo) describes the process of bringing Windows 10 for Arm onto new devices, including the Nvidia Tegra based Nintendo Switch. Getting the boot process exactly right is the big challenge, and he looks specifically at UEFI development.