The Nvidia Jetson AGX Xavier is said to deliver more than 20 times the performance of the Jetson TX2, yet is 10 times more energy efficient, consuming as little as 10W.
It is aimed at AI workloads in autonomous machines. Early adopters include Yamaha (mobility devices), Komatsu (construction and mining equipment), Denso (industrial IoT and robots), Fanuc (factory automation) and Oxford Nanopore (DNA sequencing equipment).
“Denso can leverage its long history in auto parts manufacturing to bring AI to factories, boosting productivity and efficiency while increasing workplace safety,” said executive director Katsuhiko Sugito.
Oxford Nanopore chief executive Gordon Sanghera said: “We’re using Jetson AGX Xavier for our MinIT handheld AI supercomputer, which is used to perform real-time analyses with the MinION, a powerful handheld DNA sequencer.
“MinIT can be nearly 10 times more powerful than standard laptops and brings portable, real-time sequencing to more people in more locations.”
Like its predecessors, the Nvidia Jetson AGX Xavier supports applications developed with the JetPack (autonomous machine) and DeepStream (streaming analytics) SDKs.
Jetson AGX Xavier delivers 32 TeraOPS (TOPS) of peak compute and 750Gbps of high-speed I/O in a 100x87mm form-factor.
It features an integrated 512-core Volta GPU with 64 Tensor Cores, an eight-core Nvidia Carmel ARMv8.2 64-bit CPU, 16GB RAM, dual Nvidia Deep Learning Accelerator engines, Nvidia Vision Accelerator engine, HD video codecs, 128Gbps of dedicated camera ingest, and 16 lanes of PCIe Gen 4 expansion.
Following the September release of the US$2499 Jetson AGX Xavier development kit, the Nvidia Jetson AGX Xavier module is now available, priced at US$1099 in quantities of at least 1000 units.
In related news, Nvidia announced a new 4GB version of the Jetson TX2 module.