Abstract: The signal processing of different application scenarios puts forward the requirements of generalization and virtualization to the realization platform. In response to these needs, this paper presents a signal processing platform based on heterogeneous computing. This design adopts a massively parallel processing architecture and a peer-to-peer heterogeneous resource organization form, uses ROCE interfaces and exchange units to interconnect computing resources such as FPGA, DSP, and CPU; at the same time, it uses dynamic reconfigurable technology to virtualize FPGA resources. Besides, this design uses the docker solution for software virtualization to fully complete system integration and resource scheduling. Finally, the prototype machine of the heterogeneous general signal processing platform is verified through typical communication tasks.