QMC=Chem: A Quantum Monte Carlo Program for Large-Scale Simulations in Chemistry at the Petascale Level and beyond

摘要

In this work we discuss several key aspects for an efficient implementation and deployment of large-scale quantum Monte Carlo (QMC) simulations for chemical applications on petaflops infrastructures. Such aspects have been implemented in the QMC=Chem code developed at Toulouse (France). First, a simple, general, and fault-tolerant simulation environment adapted to QMC algorithms is presented. Second, we present a study of the parallel efficiency of the QMC=Chem code on the Curie machine (TGCC-GENCI, CEA France) showing that a very good scalability can be maintained up to 80 000 cores. Third, it is shown that a great enhancement in performance with the single-core optimization tools developed at Versailles (France) can be obtained.