High Performance Computation Achieved Through Advanced GPGPU
- Integration of RedHawk Linux for hard RT processing
- COTS SFF GPGPU for military (A176) and industrial (A177) applications
- Compact, low-power, SWaP-optimized with TFLOP performance
Chatsworth, Calif. March 14, 2019 – As leaders in GPGPU computing technologies, Aitech and Concurrent Real-Time have partnered to integrate RedHawk Linux RTOS with two of Aitech’s powerful NVIDIA Jetson TX2-based SFF supercomputers. The military-grade A176 Cyclone and the industrial-focused A177 Twister enhance real-time computing for mission-critical applications.
Using the increased power-to-performance ratio offered by the Jetson TX2 and the advanced development environment of RedHawk Linux, the two units enable reliable deep learning capabilities in a growing number of artificial intelligence (AI) environments.
“Guaranteeing real-time processor performance, low process-dispatch latencies and minimal process run-time jitter are challenging in a real-time operating environment,” said Dan Mor, GPGPU and HPEC product line manager at Aitech. “Using NVIDIA’s CUDA driver, combined with Concurrent Real-Time’s RedHawk Linux distribution, we can overcome many of these computing obstacles.”
“Innovations, such as incorporating the Linux RedHawk RTOS into the Jetson AI platform, helps advance real-time computing power beyond what’s been traditionally expected,” said Murali Gopalakrishna, head of product management for autonomous machines at NVIDIA.
RedHawk offers field-proven, open-source Linux technology, advanced GUI debugging and analysis tools, and multi-core architecture support, all of which are ideal for high performance embedded computing (HPEC) systems.
“RedHawk is the ideal RTOS for complex embedded projects such as intelligent devices, advanced robotic systems and other autonomous applications,” said Ken Jackson, Concurrent Real-Time president and CEO. “With its unrivaled combination of ultra-low latency, high determinism and advanced debugging tools, RedHawk delivers the guaranteed performance required to support an emerging wave of Jetson-based products.”
Mor continued, “By combining Aitech’s hardware expertise with these advanced software resources, our CUDA-accelerated GPGPU-based supercomputers can reliably provide a hard real-time environment on many qualified hardware configurations, even in the presence of a heavy system load.”
The compact A176 Cyclone is a military-grade, fanless, conduction-cooled, rugged GPGPU SFF supercomputer that measures only 5 square inches, while providing 1 TFLOP of parallel processing, drawing less than 17 Watts and weighing less than 2.2 lbs.
The unit features multiple hardware I/O and software options (1553, ARINC 429, camera link/HD-SDI or composite frame grabbers) as well as internal microSD storage. These attributes increase a system’s ability to easily scale to more complex, compute intensive applications, enabling fast integration that saves development time and money.
Offering the same functionality as the A176, but designed for less rugged industrial environments, the A177 also handles complex computational requirements, such as managing multiple data and video streams. It can also reach 1 TFLOPS, while consuming only 17 Watts and weighing less than 2.2 lbs.
These low-cost, low-power, SWaP-optimized (size, weight and power) SFF GPGPUs use CUDA, deep learning acceleration capabilities and the real-time Linux software to ensure hard real-time performance, while several demanding embedded applications are running.
Typical uses include high data throughput and processing, such as image and data manipulation, application of digital filters, image and frame object edge detection, image recognition and data analysis, in a number of military (A176) and industrial (A177) applications.
Need more information? Contact an Expert