Berkeley Lab computer scientists are leading three projects to develop software technologies for next-generation supercomputers under the Department of Energy’s Exascale Computing Project (ECP). Lab researchers also provide support to four other ECP software projects.
“The Pagoda Project: Lightweight communication and global address space support for exascale applications,” is led by Scott Baden, head of the Computer Languages and Systems Software Group in the Computational Research Division (CRD). The Pagoda project is developing UPC++ and GASNetsoftware libraries for supporting efficient asynchronous communication on high-performance applications running on supercomputers. These libraries can provide significant reductions in communication costs of moving data, which are often a performance bottleneck in scientific applications. The software will be used by other Exascale Computing Project efforts to meet the challenges posed by computing exascale systems expected by the end of the next decade.
“ExaHDF5: Delivering Efficient Parallel I/O on Exascale Computing Systems,” is led by Suren Byna, staff scientist in CRD’s Scientific Data Management Group. Hierarchical Data Format version 5 (HDF5) is the most popular high-level I/O library for scientific applications to write and read data files. The HDF Group released the first version of HDF5 in 1998 and since then it has been used by numerous applications, not only in a wide range of scientific domains, but also in finance, aerospace and more. HDF5 is the most-used library for performing parallel I/O on existing HPC systems at the Department of Energy’s leadership computing facilities.
NASA gives HDF5 software the highest technology readiness level (TRL 9), which is given to actual systems “flight proven” through successful mission operations. In the ECP project, the team is developing various parallel HDF5 features to address efficiency, fault-tolerance,and other challenges posed by data management and parallel I/O on exascale architectures. Many of the funded exascale applications and co-design centers require HDF5 for their I/O, and enhancing the HDF5 software to handle the unique challenges of exascale architectures will play an instrumental role in the success of the ECP.
“Factorization Based Sparse Solvers and Preconditioners for Exascale,” is led by Xiaoye “Sherry” Li, leader of CRD’s Scalable Solvers Group. When researchers try to solve science and engineering problems, they often create systems of linear equations that need to be solved. As science and engineering goals become more ambitious, the number of equations keeps growing, requiring large parallel computers for their solution. Software libraries known as solvers provide mathematical tools that can be applied to problems in areas such as fusion energy, accelerator physics, and analysis of X-ray scattering data.
This project is developing novel techniques for solving large sparse linear systems, and implementing them in two state-of-the-art software packages: SuperLU and STRUMPACK. The resulting mathematical software will enable many multphysics and multiscale simulations with higher resolution and scale, making effective use of the emerging parallel machines. In systems of linear equations that are sparse, where many of the matrix entries are zero and ideally need neither to be stored nor operated on, the strength of SuperLU (which was developed by Li) is that it can automatically determine which matrix entries are zeros and can be ignored, allowing the computer to focus its calculations on the other entries and solve the problem faster.
Additionally, Berkeley Lab staff will contribute their expertise to these four ECP software projects:
“Extreme-scale Scientific Software Development Kit for the Exascale Computing Project: xSDK4ECP,” Sherry Li, head of CRD’s Scalable Solvers Group, is a member of the team.
“The ADIOS Framework for Scientific Data on Exascale Systems,” John Wu, leader of the Scientific Data Management Group in CRD, is a member of the team.