Using BOINC Desktop Grid for High Performance Memory Detecti

Memory DetectionHuiping Yao, Lei Zhao*, Ying Li, Jiwen YangSchool of Computer Science and TechnologySoochow UniversitySuzhou 215006, China{20074227065072, zhaol, ingli, jwyang}@Abstract—Memory detection computation, data intensiveapplications, demands large computing resources. Mainframesare usually used to accelerate its computation. However, thissolution seems to be costly. In this paper, we present analternative approach of using desktop grid, using PC computers’idle CPU time, to realize parallel processing of memory detection.In order to demonstrate the effectiveness of this approach, aserial of simulation experiments were done. The results show thatit is a feasible approach for memory detection computation.Index Terms—BOINC, desktop grid, memory detection, highperformance computingI.INTRODUCTIONIn organizations(business, university, etc.), the percentageof computation time, when PC computers are busy, is less than5%.The rest are idle[1]. Desktop grid[2] appears to use theseidle resources to meet large-scale computation. BOINC[3], oneof the desktop grid platforms, is designed to supportapplications that have large computation requirements, storagerequirement, or both. The main requirement of BOINC is thatthe application be divisible into a number of jobs that can bedone independently.Memory detection, indispensable in semiconductor test,analyzes the data files from testers and gives the correspondingresults. Memory detection computation belongs to the large- scale computing. Mainframes are usually used to executememory detection. However, the mainframes are quiteexpensive.The new approach, presented in this paper, using BOINC inmemory detection computation is feasible for reasons below: