Nvidia today announced what it's calling the world's smallest supercomputer for artificial intelligence (AI) at the edge, the Jetson Xavier NX. The supercomputer joins Nvidia's Jetson product line, which also includes the Jetson AGX Xavier, Jetson TX2 series and Raspberry Pi rival Jetson Nano.
Jetson Xavier NX, which runs off a six-core SoC based on Arm's Carmel microarchitecture, is smaller than a credit card and has low power consumption, running off as little as 10 watts. Expect a variety of I/Os, including CSI, PCIe, I2C and GPIO support. The Jetson Xavier NX is also pin-compatible with the Jetson Nano.
Nvidia Jetson Xavier NX Specs
GPU | Nvidia Volta (384 CUDA cores, 48 Tensor cores, 2x NVDLA |
CPU | 6-core Carmel Arm 64-bit CPU (6MB L2 + 4MB L3) |
Video | 2x 4K30 Encode and 2x 4K60 Decode |
Camera | Up to 6 CSI cameras (36 using virtual channels); 12 lanes (3x4 or 6x2) MIPI CSI-2 |
Memory | 8GB LPDDR4x; 51.2 GBps |
Internet | Gigabit Ethernet |
OS | Ubuntu / Linux |
Dimensions | 70 x 45mm (2.8 x 1.8 inches) |
Nvidia claimed its Jetson Xavier NX topped all five benchmark metrics when measuring the performance of AI inference workloads when working with both data centers and at the edge. The new supercomputer uses the Nvidia JetPack software development kit. This makes it possible to implement more serious AI projects like deep learning tools, AI networks and more.
The new Jetson Xavier NX is set to release in March with a price tag of $399.00 (€360.48).