GRA116V

The GRA116V is a rugged 3U VPX GPGPU card designed for high-performance embedded computing (HPEC) in mission-critical environments. 

The RTX3000 variant is based on NVIDIA’s Turing architecture and features the Quadro RTX 3000 GPU with 1920 CUDA cores, 240 Tensor cores, and 30 RT cores. It supports up to 5.3 TFLOPS of FP32 performance and includes 6 GB of GDDR6 memory. This card is optimized for video capture, encoding/decoding (H.265/H.264), and real-time processing with support for four 3G-SDI inputs and outputs, plus one DisplayPort output. 

The A2000 variant upgrades to NVIDIA’s Ampere architecture with the RTX A2000 GPU, offering 2560 CUDA cores, 80 Tensor cores, and 20 RT cores. It delivers up to 9.3 TFLOPS FP32 performance and includes 8 GB of ECC-enabled GDDR6 memory. This model supports PCIe Gen 4 for faster data throughput and is tailored for AI inferencing, deep learning, and real-time video processing. Like the RTX3000 variant, it supports four 3G-SDI inputs and outputs, but with enhanced performance and efficiency for modern ISR and C5ISR applications. 

At the top end, the A4500 variant features the powerful RTX A4500 GPU, also based on Ampere, with 5888 CUDA cores, 184 Tensor cores, and 46 RT cores. It boasts 16 GB of ECC GDDR6 memory and up to 17.66 TFLOPS FP32 performance. This card is ideal for compute-intensive AI workloads, real-time ray tracing, and multi-sensor video processing. It supports four 3G-SDI inputs and outputs, plus one DisplayPort output, and is designed for the most demanding embedded systems in defence and aerospace environments.

Software Support

Windows and Linux drivers are provided for x86 based platforms. The GRA116V also supports the NVIDIA software stack including the CUDA, CUDA-X, Open CL, H.264/H.265 along with NVIDIA GPUDirect RDMA technology.