Close this search box.

Aitech’s Rugged GPGPU AI Supercomputer Now Offers Quicker Processing of Critical Functions

Enhanced NVIDIA AGX Xavier-based A178 offers double RAM, new high-speed interfaces


Technology Highlights:

  • NVIDIA Jetson AGX Xavier SoM offers TFLOP processing
  • Rugged, compact, low power, SWaP-C-optimized
  • 32 GB LPDDR4x RAM; double the previous version
  • High speed interfaces for exceptional data handling: 10GB Ethernet, USB 3.0, DisplayPort @ 4K

Chatsworth, Calif. October 2020 – Aitech Systems, a leading provider of rugged boards and system level solutions for defense, space and industrial programs across the globe, has released an upgraded, qualified version of its high performance, compact A178. Designed for intense data processing in extreme environments, the rugged GPGPU AI supercomputer reliably operates in the harsh conditions found throughout mobile, remote, military and autonomous platforms. The A178 system is ideal for applications such as training simulation, situational awareness, AI computing, image and video processing, moving maps and much more.

Dan Mor, GPGPU product line manager for Aitech, noted, “High speed interfaces that can quickly capture and process data, then deliver it to the next decision point, are critical to avoid bottlenecks typical of the high processing demands in today’s rugged AI applications.  The new A178 includes not only the needed interfaces, like 10GbE and USB 3, but up to 32 GB of RAM for even better performance in these high throughput scenarios.”

Already one of the smallest of Aitech’s extensive AI SFF systems, the new A178 packs even more performance into a compact, rugged form factor. It uses the NVIDIA Jetson AGX Xavier System-on-Module that features the Volta GPU with 512 CUDA cores and 64 Tensor cores to reach 32 TOPS INT8 and 11 TFLOPS FP16. 

The upgrades to the A178 were designed to help meet the demand for standalone, compact, GPGPU-based small form factor (SFF) systems that are both rugged and SWaP-C-optimized. The low power unit offers a high level of energy efficiency, while providing all the power necessary for AI-based local processing right where it’s need, next to the sensors.   

The advanced computation abilities of the new system include two dedicated NVDLA (NVIDIA Deep-Learning Accelerator) engines that provide an interface for deep learning applications, making it ideally suited for distributed systems. The system can accommodate up to three expansion modules, such as an HD-SDI frame grabber, composite frame grabber or NVMe SSD. A variety of expansion modules are available upon request.

Four high definition HD-SDI inputs and eight composite inputs handle multiple streams of video and data simultaneously at full frame rates. Interfaces include Gigabit and 10GB Ethernet, DisplayPort output handling 4K resolution, USB 3.0 & 2.0 as well as DVI/HDMI output, UART serial and CANbus, among others.

Need more information? Contact an Expert
Download Hi-res Image