Nvidia’s Jetson Nano (Fig. 1) has confirmed to be extraordinarily in style in supporting machine-learning (ML) functions to the sting. The compact kind issue (70 × 45 mm with a 260-pin edge connector) and low energy necessities enable it to suit into drones, cell gadgets, and an array of different low-power functions the place synthetic intelligence (AI) can a profit. This could vary from figuring out objects in video streams from a surveillance system to impediment avoidance in an automatic supply robotic.
1. The Jetson XAVIER NX is pin-compatible with the unique Jetson Nano.
The brand new Nvidia Jetson XAVIER NX (Fig. 2) is plug-compatible with the unique Jetson Nano, however it brings the efficiency and performance of the higher-end Jetson TX2. The six 64-bit Carmel Arm cores can all work with the 384 cores within the Nvidia Volta GPU. It additionally has a pair of Nvidia deep-learning accelerators (NDLA) that may be utilized. The system options eight GB of 128-bit LPDDR4x DRAM and 16 GB of eMMC flash storage.
2. The Jetson XAVIER NX runs quicker than its older sibling, the Jetson Nano, operating on par with the bigger Jetson TX2.
The Jetson XAVIER NX has loads of power-management assist in addition to 800-MHz, 10-W and 1100-MHz, 15-W working modes. The previous permits a pair of CPU cores to run at 1.5 GHz or 4 at 1.2 GHz. The latter bumps the frequencies to 1.9 GHz and 1.four GHz, respectively.
The newer module can deal with a dozen MIPI CSI-2 D-PHY inputs for a complete throughput of 30 Gb/s. This interprets to half-a-dozen digital camera streams with as much as 36 digital channels. Video encode assist tops out supporting a pair of 4K video streams at 30 frames/s.
The opposite peripheral assist is much like the Jetson Nano and the Jetson TX2. The Jetson XAVIER NX has a 1-Gb/s Ethernet port, PCIe Gen three x4 and x1, and HDMI 2 as much as 2Kp50 or DPI 1.four as much as 4Kp120 or two eDP 1.four 4Kp120 channels. There’s a USB three.zero port and a pair of USB 2.zero ports together with SDIO, SPI, UAR, I2C, I2S, CAN and GPIO ports.
The Jetson XAVIER NX doesn’t fairly hit the efficiency of the higher-end Jetson AGX XAVIER. The latter delivers 32 TOPS utilizing INT8, whereas the Jetson XAVIER NX is available in at 21 TOPS. Nonetheless, it’s extraordinarily spectacular efficiency given the compact kind issue and energy necessities.
The Jetson Nano was a preferred goal for provider boards just like the Gumstix Jetson Nano Growth Board quickly to be launched by Gumstix (Fig. three). It sports activities a pair of Raspberrry Pi vertical digital camera connectors in addition to a Raspberry Pi-compatible 40-pin header. It additionally has an RJ-45 Ethernet jack, USB three.zero port, an HDMI port and an M.2 Key E wi-fi card socket.
three. Gumstix does Nvidia one higher on improvement boards by offering extra digital camera inputs.
The Gumstix Jetson Nano Snapshot Board (Fig. four) helps 4 Jetson modules, offering every with its personal four-camera inputs. One of these provider board targets embedded video functions.
four. The Gumstix Jetson Nano Snapshot Board targets embedded digital camera functions by internet hosting 4 modules every with 4 digital camera inputs.
AVerMedia’s EN713-AAE9-0000 provider board (Fig. 5) targets finish functions like video-surveillance programs. It consists of an eight-port 10/100 Ethernet change with power-over-Ethernet (PoE) assist. The board additionally exposes the Gbit Ethernet energy, and maintains two USB three.zero ports and a 4Kp60 HDMI output. Different peripheral interfaces embody a UART, two I2C ports, 5 GPIOs, a Micro-B USB 2.zero for restoration assist, and a mPCIe socket with USB 2.zero assist.
5. AVerMedia’s provider board targets surveillance programs with PoE digital camera assist.
The Jetson XAVIER NX and Jetson Nano present a scalable, compact modular kind issue that packs a variety of compute energy tailor-made for AI inference functions. The Jetson Nano has been confirmed on this area, and the added efficiency that the Jetson XAVIER NX brings will solely broaden the capabilities and variety of functions that may be addressed.
The opposite benefit of this household is that the corporate’s software program assist spans this household in addition to the opposite Nvidia GPU-based hardware by means of the cloud-based playing cards and programs. In consequence, functions will be moved simply from platform to platform, serving to simplify improvement and deployment.