The analysis of a plane wave pseudopotential density functional theory code on a GPU machine

Density functional theory based calculations for high pressure research

“…Currently most codes parallelize well for a few hundred to a few thousand cores, depending on the calculation and a comparison of several codes has shown that it is not straightforward to adopt the programs to very large computers (Plummer et al, 2006). It requires a significant effort to port these codes to new hardware with tens of thousands or possibly millions of cores (Jia et al, 2013), and this has not happened yet for the codes which are currently widely used.…”

Computational Physics on Graphics Processing Units

“…Wand et al [59] and Jia et al. [60] have done an implementation for GPU clusters of a plane wave pseudo-potential code called PEtot. They were able to achieve speedups of 13 to 22 and parallel scalability up to 256 CPU-GPU computing units.…”

Parallel transport time-dependent density functional theory calculations with hybrid functional on summit

“…The application of the Fock exchange operator can be accelerated with a massive number of CPUs [9,35]. The GPU acceleration has been reported in software packages such as ABINIT [11], BigDFT [29], NWChem [34], Octopus [4], PWmat [20,21], Quantum ESPRESSO [30], VASP [12,18], to name a few. When approximation of the Fock exchange operator can be tolerated, the cost can also be reduced through other algorithmic approaches such as localization [6][7][8]38] and density fitting techniques [17].…”