Abstract: A graph sampling and random walk acceleration method and system based on GPU. The graph data is read from the storage medium through the CPU and converted into CSR format and then output to the GPU. The GPU works based on the defined working mode: Real-time generation of alias table and sampling; or offline judgment whether there is a pre-generated alias table and sampling, which executes the alias method efficiently and in parallel, and can significantly improve the performance of graph data processing on the same hardware platform, including improving the sampling throughput and reducing the overall running time.

Abstract: An adaptive, large graph-oriented unified memory management method is proposed; according to the order of priorities of different types of graph data structure in a graph computing application, whether the current GPU memory is full is sequentially determined by means of GPU memory checking; whether the size of current graph data exceeds the available memory capacity of the GPU is determined by means of data overflow checking; and then a unified memory management policy is configured; the method uses different graph algorithms based on the characteristics of graph data structure and the size of the available GPU memory; the method can significantly improve the performance of processing large graphs, which exceed the video memory capacity under a unified memory architecture, including improving GPU bandwidth utilization, reducing the number of page faults and the overhead of processing page faults, and speeding up the execution of graph computing programs.

Abstract: The invention relates to a nanosecond-scale power resource allocation method and system for microservices. In order to eliminate the macro-control delays, it uses an auto power budgeting module to distribute the limited power resource to each microservice according to its unique power-performance model. On the other side, it deploys an agile power regulation module to bypass the slow power control procedures at the server system layer. By directly invoking the voltage regulation circuit on the processor, this can remove the micro-execution delays. It also leverages a transparent mapping mechanism to coordinate the work of the auto power budgeting module and the agile power regulation module to achieve the optimal allocation of power resources. The present invention can exploit the intra-application variability brought by microservices, thus further break the energy-saving limitation of traditional data centers.

Abstract: The invention relates to a nanosecond-scale power resource allocation method and system for microservices. In order to eliminate the macro-control delays, it uses an auto power budgeting module to distribute the limited power resource to each microservice according to its unique power-performance model. On the other side, it deploys an agile power regulation module to bypass the slow power control procedures at the server system layer. By directly invoking the voltage regulation circuit on the processor, this can remove the micro-execution delays. It also leverages a transparent mapping mechanism to coordinate the work of the auto power budgeting module and the agile power regulation module to achieve the optimal allocation of power resources. The present invention can exploit the intra-application variability brought by microservices, thus further break the energy-saving limitation of traditional data centers.

Abstract: A graph sampling and random walk acceleration method and system based on GPU. The graph data is read from the storage medium through the CPU and converted into CSR format and then output to the GPU. The GPU works based on the defined working mode: Real-time generation of alias table and sampling; or offline judgment whether there is a pre-generated alias table and sampling, which executes the alias method efficiently and in parallel, and can significantly improve the performance of graph data processing on the same hardware platform, including improving the sampling throughput and reducing the overall running time.