Fecha de publicación: 21/07/2017 Fuente: Wipo "BigData" The invention provides an Arria10-based FPGA heterogeneous acceleration card. The card comprises an FPGA chip, an optical port module connected with the FPGA chip, a memory channel module, a PCIE host interface, a configuration module, a power conversion module, a clock module, an LED indication module, an on-board USB Blaster, and a Micro USB connector connected with the on-board USB Blaster; the power conversion module is further connected with the optical port module, the memory channel module and the PCIE host interface; and the clock module is further connected with the on-board USB Blaster through a JTAG debugging interface. According to the card, CPU+FPGA heterogeneous acceleration computing is adopted; the power consumption of an FPGA is smaller than 35W, and the energy consumption is more than 5 times that of a GPU; and the CPU+FPGA heterogeneous acceleration computing has shorter time delay and better acceleration performance, and will replace CPU+GPU computing in the field of big data and cloud computing in the internet era.