New VPX GPGPU leads in AI performance
To help UK-based aerospace, defence and space companies to stay ahead of market demand for advanced AI, embedded computing specialist Recab UK is now offering the updated C530 general-purpose graphics processing unit (GPGPU) from Aitech. The latest version of the C530 is a powerful AI-enabled 3U VPX GPGPU, offering high compute performance in a compact and rugged footprint.
The C530 features a choice of NVIDIA Quadro GPUs, both based on NVIDIA’s Turing architecture for improved efficiency and performance in demanding applications. The architecture also offers enhanced CUDA features such as independent thread scheduling and hardware accelerated Multi Process Service (MPS) with address space isolation for multiple applications.
Aitech’s C530 is suited to the computational requirements of in-field AI applications, real-time image and video processing and numerous other high data-intensity operations. Alongside its choice of high-performance GPUs, the board is OpenVPX compliant and features four video output channels, PCIe x16 Gen 3 host interface and support for Windows and Linux operating systems. The board is also vibration and shock resistant and is available in air-cooled and conduction-cooled versions.
“There is a growing demand for high performance computing and advanced AI in embedded systems,” explained Martin Frederiksen, Managing Director of Recab UK. “We’ve found from our extensive experience in the aerospace and defence sectors that these markets in particular are looking for increasingly sophisticated embedded computing. It’s here that the computing capabilities of GPGPUs, particularly the C530, are the optimum solution.
“One of the challenges in the sector is balancing the high computing power required for applications like real-time video feed analysis and deep-learning AI with ruggedised designs. The C530 is designed for these demanding environments and, coupled with our industry experience, we are able to support companies with full embedded system design that meets project requirements and is compliant to relevant military standards.”
The C530 GPGPU is available with either an RTX3000 GPU, which features 1920 CUDA cores for parallel processing and 6 GB DDR6 memory that support up to 5.3 Tera floating-point operations per second (TFLOPS), or T1000, which boasts 896 CUDA cores and 4GB DDR5 to enable 2.6 TFLOPS.
Real-time image and video processing and deep learning applications are best suited to the higher performance RTX3000 model. This is because the GPU features 30 ray tracing (RT) cores for enhanced real-time rendering, as well as 240 Tensor cores for greater computing capacity and AI inferencing.
The C530 VPX board is available as part of a bundle with other Aitech 3U VPX boards such as the C875 and C877. It is also one of the 3U VPX boards that features in Aitech GPGPU systems, such as the A196 high-performance embedded computer.