Abaco Systems' GRA116OP high-performance GPGPU delivers outstanding video and graphics output flexibility with the latest graphics processing unit (GPU) technologies. Based on the NVIDIA® Ampere™ architecture using the NVIDIA A2000™ platform, the 3U VPX VITA 65 form factor graphics output is ideal for rugged defence, aerospace and industrial applications.
The GRA116OP combines general-purpose computing on graphics processing units (GPGPU) computing, AI inferencing, deep learning, sensor processing and data analytics with highly flexible graphics and video I/O configurations. Its GPU node is equipped with 8 GB GDDR6 graphics memory with ECC, 2560 CUDA® cores.
The new high-performance GPGPU also includes 80 Tensor cores (3rd gen), 20 RT cores (2nd gen), and delivers up to 9.3 TFLOPS floating-point 32 performance with CUDA™ and OpenCL™ support. Additionally, the platform provides a 128-bit wide memory bus for 192 GB/s peak memory bandwidth.
To minimise video latency, real-time video and data streaming are achieved with NVIDIA GPUDirect™ RDMA. The GPU also delivers real-time performance for encoding applications with dedicated H.265 and H.264 encode and decode engines.
Graphics/video outputs are available in four configuration options:
The GRA116OP also helps reduce bottlenecks and increase system performance with the latest Gen 4 PCIe fabric interface supporting both x8 lanes and x4 lanes and optimising expansion plane bandwidth to adjacent single board computer (SBC) boards.
Air- and conduction-cooled versions of the GRA116OP are available. The GRA116OP can consume from 35W to 95W depending on the factory configuration. Windows and Linux drivers are provided for x86 based platforms.
More Information.
The GRA116OP combines general-purpose computing on graphics processing units (GPGPU) computing, AI inferencing, deep learning, sensor processing and data analytics with highly flexible graphics and video I/O configurations. Its GPU node is equipped with 8 GB GDDR6 graphics memory with ECC, 2560 CUDA® cores.
The new high-performance GPGPU also includes 80 Tensor cores (3rd gen), 20 RT cores (2nd gen), and delivers up to 9.3 TFLOPS floating-point 32 performance with CUDA™ and OpenCL™ support. Additionally, the platform provides a 128-bit wide memory bus for 192 GB/s peak memory bandwidth.
To minimise video latency, real-time video and data streaming are achieved with NVIDIA GPUDirect™ RDMA. The GPU also delivers real-time performance for encoding applications with dedicated H.265 and H.264 encode and decode engines.
Graphics/video outputs are available in four configuration options:
- x2 DisplayPort, x2 Single-Link DVI-D outputs
- x4 Single-Link DVI-D outputs
- x4 DisplayPort outputs
- x2 Single Link DVI-D, x2 3G-SDI
The GRA116OP also helps reduce bottlenecks and increase system performance with the latest Gen 4 PCIe fabric interface supporting both x8 lanes and x4 lanes and optimising expansion plane bandwidth to adjacent single board computer (SBC) boards.
Air- and conduction-cooled versions of the GRA116OP are available. The GRA116OP can consume from 35W to 95W depending on the factory configuration. Windows and Linux drivers are provided for x86 based platforms.
More Information.