Accurate Blind Predictions of OpenFOAM Energy Consumption Using the LBM Prediction Model

The ability to predict the energy consumption of an HPC task, varying the number of assigned nodes, can lead to the ability to assign the correct number of nodes to tasks, saving large amount of energy. In this paper we present LBM, a model capable of predicting the resource usage (applicable to different resources, such as completion time and energy consumption) of programs, following a black box approach, where only passive measures of the running program are used to build the prediction model, without requiring its source code, or static analysis of the binary. LBM builds the predicting model using other programs as benchmarks. We tested LBM predicting the energy consumption of pitzDaily, a case of the OpenFOAM CFD suite, using a very low number of benchmarks (3), obtaining extremely precise predictions.