NCSI

   

PPerfLab Performance Measurement


Shodor > NCSI > XSEDE EMPOWER > XSEDE EMPOWER Positions > PPerfLab Performance Measurement

Status
Completed
Mentor NameKaren L. Karavanic
Mentor's XSEDE AffiliationPortland Institute for Computational Science; XSEDE Empower mentor
Mentor Has Been in XSEDE Community4-5 years
Project TitlePPerfLab Performance Measurement
SummaryThe intern will assist the PPerfLab research group at Portland State as they develop state-of-the-art performance measurement tools for HPC platforms. We are developing a runtime detection framework for file system interference to allow performance tools to determine a correct diagnosis in the case where an unrelated job is causing the performance bottleneck.
Job DescriptionThe intern will perform testing and experiment runs as part of the File System Interference project. This includes installing and executing codes on PSU's Coeus Cluster, and on smaller testbeds available in the lab. The position requires familiarity with Linux and python, and some familiarity with MPI and parallel performance measurement tools. The intern will work with graduate students and Dr. Karavanic, and participate in group meetings.
Computational ResourcesWork will be performed on PSU's Coeus Cluster, and on smaller testbeds available in the lab.
Contribution to CommunityThe project goals are developing performance tools for use by HPC application developers. The project involves mentoring an undergraduate student interested in graduate study, and increases the chance they will pursue studies and research relevant to HPC.
Position TypeIntern
Training PlanThe student will receive additional training in parallel performance tools and data analysis techniques. This training will be done by graduate students in the group, and directed by Dr. Karavanic.
Student Prerequisites/Conditions/Qualificationsfamiliarity with Linux and python, and some familiarity with MPI and parallel performance measurement tools
DurationQuarter
Start Date01/03/2022
End Date03/18/2022

Not Logged In. Login