The NVIDIA Jetson TX1 is teraflop computing in a credit card size form factor. Nvidia today announced a new module called the Jetson TX1 that fits in your hand but can bring GPU-powered intelligence to development boards, drones, robotics, and other devices.
The module includes a 1-teraflop/second 256-core Maxwell-based GPU, 64-bit ARM A57 chips, 4GB of LPDDR4 RAM with 25.6GB/second bandwidth, 15GB of storage, 802.11 2×2 ac Wi-Fi, and 1GB Ethernet. It also comes with the Jetson Linux software development kit (SDK). The device measures 50 millimeters by 87 millimeters, making it a bit smaller than a credit card.
The new gadget can handle many kinds of machine learning workloads and thereby enable devices to run smarter applications. And that’s increasingly important.
“The ability for computers to learn, the ability for computers to write software itself and do seemingly amazing things, artificially intelligent things, is revolutionizing web services,” Nvidia cofounder and chief executive Jen-Hsun Huang said at a company press event in San Francisco today.
Nvidia sells GPUs to Microsoft, Amazon Web Services, Google, and IBM, among others, in order for them to run their own or other companies’ artificial intelligence applications. Partners in the launch of the new Jetson TX1 include the Massachusetts Institute of Technology, Herta, Kespry, Percepto, and Stereolabs.
The Jetson TX1 is the successor to the Jetson TK1 embedded hardware. Nvidia launched the Drive PX hardware for autonomous vehicles earlier this year.
The new Jetson TX1 developer kit, which comes with a development board, will become available in the U.S. for $599 ($299 for education customers) starting on November 12. The modules will become available for sale at a suggested price of $299 early next year, according to a statement.