Accelerator cards for inference computing at the edge

Cards based on VPU, FPGA, and TPU for high-performance computing at the edge. - Low latency, high throughput, and low power consumption – these are the core characteristics of the accelerator cards based on FPGA, VPU, or custom ASIC. Operating System: Linux 64-bit, Windows 64-bit / Ubuntu 64-bit, CentOS Framework / Topology: Tensorflow Lite / AlexNet, GoogleNet, MobileNet, MTCNN, Squeezenet, Tiny Yolo, Yolo, ResNet, etc. Main Chip: Coral Edge TPUs™ / Intel Movidius Myriad / Intel Arria Slot: PCIe x4 Gen2 / PCIe x4 / PCIe x8 Power Consumption: 15 Watts / 25 Watts / up to 60 Watts