Dynamic forecasting of the completion time of a computational experiment in a Desktop Grid (Valentina Litovchenko, ISPRASOPEN-2019)
Материал из 0x1.tv
- Валентина Литовченко
In this talk, we consider dynamic forecasting of the completion time of a computational experiment in a Desktop Grid. A statistical approach is proposed based on a linear regression model with the calculation of the confidence interval, taking into account accumulation of statistical error and changing the forecast. Based on the developed approach forecasting algorithm and program module for software platform Desktop Grid BOINC. We present experimental results based on data from the RakeSearch volunteer computing project.