Font Size: a A A

Research On Pipelining Of Data Penetration Migration And Its Monitoring Method

Posted on:2022-08-28Degree:MasterType:Thesis
Country:ChinaCandidate:L Y WanFull Text:PDF
GTID:2518306560492634Subject:Computer technology
Abstract/Summary:PDF Full Text Request
For a long time,the contradiction between the increasing data demand of the processor and the inefficient data supply is the main factor limiting the performance of the processor.Academia and industry have focused on how to alleviate this contradiction and made attempts in different directions.Penetration data migration is one of these attempts.Different from the traditional prefetching scheme,penetration data migration emphasizes the purpose and stage of migration behavior,and advocates to adjust the data layout of storage space in real time according to the temporal and spatial characteristics of data,so as to create a situation of active convergence of data to the processor.However,the research on the process of data penetration and migration is still in a preliminary exploration stage,and the internal regularity of migrated data needs to be further studied and discussed.The research results of this paper are as follows:1.Analyze and summarize the penetration process,point out the relationship between the data penetration migration process and instruction execution,and put forward the solution to realize the pipelining of data penetration migration ——Percolation Pipeline;2.A compound pipeline with verification function is designed.Two instruction pipelines are coupled and interleaved to ensure the smooth development of the penetration process;3.The Percolation Pipeline and the Composite Pipeline with verification function are simulated by using Model Sim simulation tool.Simulation results show that the Penetration Pipeline structure can effectively realize the pipelining of data penetration migration process and ensure the correctness of data penetration migration process.
Keywords/Search Tags:Penetration Cache, Timely Localization, Penetration Data Migration, Pipeline, Coupled Execution
PDF Full Text Request
Related items