论文部分内容阅读
目的:讨论凝聚态材料和蛋白质分子电子结构研究进展对超级计算机上高性能计算的要求。方法:从密度泛函理论出发,结合计算技术,分析影响计算速度和计算精度的主要因素。结果:计算机单个结点内存的大幅度增加可使该类计算速度提高约10倍,在此条件下每个结点CPU数目的增多可几乎线性地增大计算加速比;而增加数据的存储有效位数也可大大加快计算的收敛速度,从而减少计算量。结论:满足上述要求的高性能并行计算机将大大促进材料科学和生命科学中的理论计算研究工作。
OBJECTIVE: To discuss the progress made in the electronic structure of condensed matter and protein molecules for high performance computing on supercomputers. Methods: Based on the density functional theory, combined with the calculation technique, the main factors affecting the calculation speed and accuracy are analyzed. As a result, a substantial increase in the memory of a single node in the computer can increase the computational speed by about 10 times. Under this condition, the increase in the number of CPUs per node can increase the computational speedup almost linearly; while increasing the storage of data is effective The number of bits can also greatly speed up the convergence of the calculation, thereby reducing the amount of computation. Conclusion: High-performance parallel computers that meet the above requirements will greatly facilitate theoretical computational research in materials science and life sciences.