S-1-28 Independent Platform Software Development and Application of High Throughput and High Concurrent Material Computing

Independent Platform Software Development and Application of High Throughput and High Concurrent Material Computing

Wu Zhang1*, Wenhao Zhu1, Jianyue Ni1, Lingti Kong2, Hong Zhu2, Qingyan Xu3, Chongyu Wang3

1 Shanghai University, Shanghai, 200444, China

2 Shanghai Jiao Tong University, Shanghai, 200240, China

3Tsinghua University, Beijing, 100084, China

 

ABSTRACT: High throughput and high concurrent material computing is one of the most important parts of materials genome engineering. Its essence is to explore how to apply "high throughput screening", "chemical component doping" and "data association analysis" to material computer simulation to guide new materials design. Material computing platforms have been established throughout the years, including MP, AFlow, NoMad, MatCloud etc. However, these platforms mainly focus on isolated computing resources rather than integrating with the discrete computing resources. Our research team has developed a high-throughput concurrent computing framework and software system called MIP with independent intellectual property rights and integrated with multiscale computing algorithms. MIP has been deployed in multiple super computers including the National High Performance Computing Center (Wuxi) realizes the concurrent calculation of 102 nickel base superalloy tasks with the processing and analysis of the results.

 

Figure 1. Architecture of MIP

Keywords: Materials genome engineering; high throughput and high concurrent computing; domestic supercomputer; material computing information platform.

Brief Introduction of Speaker
Wu ZHANG

Dr. Wu ZHANG, professor of School of Computing, Shanghai University, Materials Genome Institute and Shanghai Institute of Applied Mathematics and Mechanics. He has devoted himself to the research of high-performance computing and applications, and has published more than 200 papers. He established the high performance computing center of Shanghai University and served as the first director of the center in 2007. SuperLBM, an application software of CFD based on Lattice Boltzmann Method, has been jointly developed with Commercial Aircraft Company of China, and used in the design of domestic commercial aircraft since 2009. Dr. Zhang is currently a member of Numerical Software Working Group (WG2.5), the IFIP.