Design Of Edge Supercomputers For A Wide Range Of Application Scenarios

In the Internet of Everything(Io E)environment,the future computing requirements present a number of new characteristics,including pan-scenario,high-performance,green,realtime and dynamic tide.Currently,the centralised computing model based on the cloud computing model is unable to meet these new computing demands due to its high energy consumption and deployment limitations.In this context,the edge computing model has emerged.By deploying miniature computing nodes at the edge and cooperating with the large computing platforms of the cloud computing model,edge computing can build a multi-level,multi-node computing network to better meet the computing needs of the future digital society.In addition,Professor Chen Shiqing,an internationally renowned expert in supercomputing,also suggested that the future development form of supercomputing should be distributed and peripheral,and a distributed computing network should be built based on a new generation of information technology such as artificial intelligence and the Internet of Things.Under this development trend,it is necessary to explore the edge-based deployment solutions of supercomputing platforms to expand the application scope of supercomputing platforms,realise the productivity transformation of the powerful computing power of supercomputing platforms,and support the digital transformation of various industries.Based on the limitations of existing supercomputing platforms,this project proposes and designs a forward-looking new model of supercomputing edge deployment,taking into account the general trend of industrial development and the future demand for computing power.Considering the close relationship between edge supercomputers and contexts,the design and development process also requires extensive research on possible operational contexts to explore design requirements.Therefore,this paper introduces the context analysis method in the design and development process of edge supercomputers,and completes the design of edge supercomputers through four steps: context pre-study,context creation,requirement analysis,and design solution.The design pre-study refers to the study of the four elements of the context,including the supercomputing product,the relevant user,the application environment and the relevant tasks,in the early stage of design to provide material support for the subsequent context creation,so as to ensure the authenticity and completeness of the context creation.Scenario creation is based on the selection of typical use cases to simulate product use situations and the use of storytelling to demonstrate the flow of events related to supercomputers.Requirement analysis involves analysing the problematic situations in the above scenario stories to gain insight into user requirements and translate them into design requirements.Considering the complexity of the edge supercomputer system,functional analysis is introduced in the design solution stage to break down the overall functionality into a number of independent subfunctions,and to explore the best solution for the edge supercomputer through the divergence and convergence of technical solutions for each sub-function module,so that the transformation and solution of the design requirements can be more systematic and comprehensive.This research has resulted in a design solution for an edge supercomputer that is miniaturised,low power,highly mobile,integrated,environmentally tolerant and flexibly deployable,greatly improving the adaptability of the edge supercomputer,enabling it to flexibly adapt to different operational contexts and provide localised,real-time computing services to users,while reducing energy consumption and costs and improving sustainability.