The Condor GR4-RTX3000 supports four 3G-SDI inputs, four 3G-SDI outputs, and one DisplayPort (4K UHD) output. This card is designed with the NVIDIA RTX 3000 (TU106) with 1920 CUDA parallel processing cores, 240 Tensor cores, and 30 RT cores. This single card offers video capture and display, GPGPU computing, Artificial Intelligence (AI), deep learning capabilities, and features dedicated H.265/H.264 encode and decode engines for low-latency GPGPU computing.
When integrated into a High Performance Embedded Computing System (HPEC), this card is capable of capturing raw video data and other metadata from up to four sensors to perform activities such as image enhancement, image analysis, video stitching, remote sensing, and target acquisition, target tracking, and several other data processing tasks, all with extremely low latency.
The Condor GR4-RTX3000 has 6 GB GDDR6 graphics memory, NVIDIA GPUDirect RDMA for real-time video streaming, and delivers up to 5.3 TFLOPS floating-point performance with CUDA and OpenCL support. The Condor GR4-RTX3000 has 50% more compute power than previous generations and offers multiple-precision modes such as FP64, FP32, FP16, INT8, INT4, and INT1.
The card also supports SDI VANC KLV metadata insertion and extraction, making it ideal for sensor processing applications involving low latency video capture/display such as unmanned systems, Signal Intelligence (SIGINT), Electronic Warfare (EW), and other Intelligence, Surveillance, Reconnaissance (ISR) markets.
This card is designed to operate in harsh environments and tested to MIL-STD-810 standards. The card is available in level 4 conduction cooled with thermally efficient heatsink technology and rear I/O.