TapTechNews network friend Wang Miao reported that AMD recently announced the mass production of the Alveo V80 computing accelerator card. The card is aimed at memory bottleneck applications with large datasets and can flexibly optimize workloads through FPGA hardware, suitable for HPC, data analysis, finance, network security, and computational storage. The suggested retail price of the computing accelerator card on the AMD official website is $9495 (approximately 68744 RMB at current exchange rates). The Alveo V80 accelerator card has a full-height 3/4 length physical specification, uses dual PCIe 8-pin auxiliary power supply, has a maximum power consumption of 300W, and features passive (ducted) cooling design. The accelerator card is based on the 7nm AMD Versal XCV80 HBM series adaptive SoC, with a FPGA scale of 2574K LUT logic units and 10848 DSP compute logic slices. The Alveo V80 is equipped with two 16GB HBM2E memory stacks, providing a memory bandwidth of 819GB/s, and can also expand the memory capacity by an additional 32GB through the DDR4 slot onboard. The Alveo V80 accelerator card has four QSFP56 network interfaces on the IO panel, each supporting 200G networks. AMD claims that the Alveo V80 card provides more abundant interconnect bandwidth compared to GPUs that only support PCIe connections. The Alveo V80 accelerator card supports PCIe Gen4 x16 single port or Gen5 x8 dual port. The card also provides three MCIO expansion ports at the rear.