Nvidia's $ Nvidia Jetson Nano Is A Raspberry Pi Killer?



[ad_1]

Nvidia has been in the machine learning and AI game for a number of years. The company launched the Jetson TX1 "Supercomputer on Module" in 2015 as an integrated solution for robots, drones and autonomous vehicles that need to do a lot of visual computing. It's the beginning of a whole range of Nvidia "AI" products that have proven themselves. Nvidia says that there are hundreds of thousands of Jetson developers today. Although it is a viable solution for commercial enterprises, its selling price is $ 599, which often makes it too expensive for manufacturers, hobbyists and amateurs.

Today, all this has changed with the launch of Jetson Nano, a $ 99 computer development kit that paves the way for a Raspberry Pi type revolution, this time for machine learning.

The secret sauce in Nvidia's AI products is, of course, its GPUs. The Jetson TX1 used a 1024-GFLOP Maxwell GPU with 256 CUDA cores. The TX2 offers 1.3 TFLOP using a 256-core Pascal GPU and the high-end Jetson AGX Xavier exceeds 10 TFLOP with its 512-core Nvidia Volta GPU. But the Jetson AGX Xavier also passes the $ 1,000 mark! For the $ 99 Jetson Nano, Nvidia opted for a 128 CUDA graphics processor, based on the Maxwell architecture. It offers 472 GFLOPs.

The graphics processor is supported by a 64-bit four-core processor based on the Cortex-A57 processor, 4 GB of RAM, a video processor (capable of handling encoding up to 4K 30 frames per second or higher). 4K decoding at 60 frames per second) and support PCIE and USB 3.0.

The video capabilities of the Jetson Nano are impressive. The idea is not that you can watch a 4K video, but that the unit can handle multiple video streams (think of drones with multiple cameras) for object detection, tracking and obstacle avoidance . While 4K 60 frames per second looks good, the Jetson Nano is capable of decoding eight video / camera streams in Full HD at 30 frames per second! Once decoded, streams can be processed simultaneously by machine learning algorithms for tracking objects, and so on.

Read also: How to create your own digital assistant with Raspberry Pi

Jetson Nano comes in two forms. A module – which measures only 70 x 45 mm – for use in final production-ready designs, and a development kit that looks like a Raspberry Pi and offers a turnkey solution for developers and enthusiasts. The first comes with 16GB of eMMC embedded storage while the latter uses a microSD card.

Unlike previous versions of the Jetson platform, Nvidia provides two separate (but related) uses of the Jetson Nano. On the one hand, the development kit will be useful to commercial organizations that wish to create products with machine learning capabilities. The product can be designed using the development kit, and then the modules are used for the final product. This is how other Jetson cards and modules are used. The second use case is for enthusiasts and hobbyists who may never use the module version but want to create projects based on the development kit, much like Raspberry Pi.

To this end, Nvidia is ready to sell both modules and development kits, not only through wholesale distributors, but also in a larger market through more traditional outlets.

Raspberry Pi killer?

The Raspberry Pi uses a quad-core processor based on Cortex-A53 and comes with up to 1GB of RAM. While it may be fun to run simple Python scripts and other basic tasks, it can be a pain to use a desktop environment. The Jetson Nano is equipped with a quad-core Cortex-A57 processor and 4GB of RAM. This should mean that it should be at least twice as fast as the Raspberry Pi for learning tasks other than the machine. In addition, the extra RAM should allow it to run a more smooth desktop environment.

Related: Learn how to develop Android apps at DGiT Academy!

On top of that, the Jetson Nano comes with 40 GPIO pins, just like the Raspberry Pi. Although Nvidia does not specifically specify the compatibility of Raspberry Pi, it says that Jetson Nano is "compatible, with many peripherals and other add -ons ". There is also support for the Adafruit Blinka library and the Raspberry Pi V2 camera. . The card boots into a full Linux desktop environment via Linux4Tegra, derived from Ubuntu 18.04.

In other words, the Jetson Nano is similar to a Raspberry Pi, but better, stronger, faster! Add all the ML quality to the top and you have a game-changing potential.

Jetbot

To demonstrate the capabilities of the card, Nvidia launches JetBot, an open source artificial intelligence project based on Jetson Nano. It comes with a list of hardware, a hardware installation guide and tutorials. The idea is that anyone with some basic skills in Python should be able to build the little robot and learn everything about motor control, camera image acquisition and the training in AI by teaching JetBot to track objects, avoid collisions, etc.

Multiple devices, same software

One of the reasons for the success of the Raspberry Pi, compared to other Arm-based single-card computers, is that the software is still up-to-date. There are far too many cards that offer initial support for a version of Linux and the distribution is never updated or upgraded. No security patches, no new packages, and certainly no new kernel versions.

Nvidia understands this and is doing its best to keep its software up-to-date and relevant. Jetson TX1 supported Linux 3.10 and used Ubuntu 14.04. Over time, support for the 4.4 kernel was added, followed by kernel 4.9. Similarly, the Ubuntu base distribution has been upgraded from 14.04 to 16.04 and now to 18.04.

This means that Nvidia offers a unified development environment for all of its Jetson cards. You can start developing a project on Jetson Nano, but if you need more GPU power, upgrading to a more advanced Jetson card will result in little or no penalty from the software point of view.

It seems that the Jetson Nano could be a fantastic tip. Price is good, general computing performance is much better than Raspberry Pi, machine learning features (software and hardware) are excellent, and potential compatibility with existing sensors and hacks allows fans to use ( and improve) existing projects. I should have my hands on a board soon, so watch out for a full review here and on the Gary Explains YouTube channel.

Disclosure of Affiliates: We may receive compensation in connection with your purchase of products via links on this page. The remuneration received will never influence the content, topics or posts published in this blog. See our disclosure policy for more details.

[ad_2]
Source link