Parallel quantum chemistry computation in a high-level data-parallel programming environment PROMISE

Quantum chemistry has become an essential tool in many research areas of chemical, physical and materials science, because by using massively parallel computers we can solve the large molecule problems, which supply the fundamental information for these researches. Whereas the efficient implementation and maintains of ab initio chemistry computational software on the advanced computer systems are still difficult. This requires to create the new methodologies for developing the next generation of this software. The object-oriented approach have been chosen for this evaluation. Up to now, we have developed a PROMISE, a high-level data-parallel programming Environment, and successfully employed in several applications such as 2 and 3 dimensional CFD, FEM, and a diversity of solvers. This paper will address the experiments of massive parallel integral-direct SCF quantum chemistry computations in PROMISE, in order to exploit this new methodology in developing new quantum chemistry code. Thereby, the new concept of using object-oriented programming is proposed for efficient evaluation of SCF calculation. The good parallel performance had been reached on CRAY T3E and PC cluster.