NCSI: Development of Datasets of Collected Metrics Charactering the Performances of Common Scientific Applications in Different Scientific Domains on Modern HPC Platforms

Status	Completed
Mentor Name	Zhiyong Zhang
Mentor's XSEDE Affiliation	Campus Champion
Mentor Has Been in XSEDE Community	4-5 years
Project Title	Development of Datasets of Collected Metrics Charactering the Performances of Common Scientific Applications in Different Scientific Domains on Modern HPC Platforms
Summary	The position involves using common HPC tools to collect performance metrics, such as MPI communications, accesses of memory/cache hierarchies, I/O across different storage devices, on a variety of modern HPC platforms for a large number of commonly used scientific applications in different scientific domains. The collected datasets and metrics can serve as benchmarks and guides for scaling and optimizations of the applications and for the optimal configurations of computational resources, for general scientific user communities, and potentially for the acquisition of resources at campus data centers and for better informed development and configurations of hardware for the optimal uses of scientific applications.
Job Description	The student tasks include, independently and/or under guidance, (1) compiling programs and running applications on a variety of HPC resources; (2) developing scripts to automate the collection and interpretation of performance metrics; (3) optionally developing and setting up databases for storage and retrieval of the collected data; and (4) potentially developing web based portals for uploading, retrieval, and access of data.
Computational Resources	Students will have access to computational resources at Stanford and XSEDE allocations.
Contribution to Community	The work will be potentially useful for large number of users in different scientific disciplines making use of XSEDE computational resources, developers of scientific applications, as well as for the informed decisions for acquisitions of resources by research groups and campus computing centers. The project will also contribute to the development of future professionals with in depth knowledges of the hardware features of modern computational resources and how these features will impact the performances of the scientific applications.
Position Type	Apprentice
Training Plan	The staff at the host institution will mentor the students to be able to work effectively on the proposed projects. We plan to leverage help from the support teams from Intel/AMD/NVidia. The students will have the opportunity to learn from and collaborate with interested users of the scientific applications at Stanford and potentially other institutions.
Student Prerequisites/Conditions/Qualifications	Strong interests in and motivation for using HPC for research and education. Strong python programming skills are highly desirable and are important for the success of the tasks and learning objectives of the position. Ability, or the ability to quickly learn, to understand bash programs and to program in the bash programming language. Interests and backgrounds in chemistry/computational chemistry are highly desirable.
Duration	Semester
Start Date	01/10/2022
End Date	05/04/2022