Using Linux PSI to take job scheduling to the next level

About the Assignment Join NXP's High-Performance Computing (HPC) team and contribute to the optimization of job scheduling for Electronic Design Automation (EDA) workloads. Traditional schedulers consider factors such as job duration, CPU usage, and memory requirements, but storage I/O performance is becoming increasingly important. Using Linux Pressure Stall Information (PSI), you will investigate how CPU, memory, and I/O pressure impact workload performance and explore how this data can be used to improve job scheduling decisions across NXP's distributed datacenter environment. Your Responsibilities Investigate how Linux PSI metrics can be collected and associated with individual jobs within the IBM LSF scheduling environment. Analyze CPU, memory, and I/O pressure data from EDA workloads. Identify workload patterns and resource requirements using PSI measurements. Explore the relationship between I/O pressure and NFS storage performance. Develop and validate a basic model for predicting job I/O requirements. Provide recommendations on how PSI-based insights can improve scheduling decisions across local and remote datacenter resources. Document findings and present recommendations to the HPC engineering team. Your Qualifications Currently pursuing a Bachelor's or Master's degree in Computer Science, Software Engineering, Electrical Engineering, Data Science, or a related field. Basic knowledge of Linux operating systems and command-line tools. Interest in High-Performance Computing (HPC), distributed systems, and infrastructure engineering. Experience with scripting or programming languages such as Python, Bash, or similar. Strong analytical and problem-solving skills with an interest in data analysis. Familiarity with Linux performance monitoring tools is a plus. Self-driven, curious, and comfortable working on research-oriented assignments. Good communication skills and ability to work independently and within a team. What You'll Learn High-Performance Computing environments and job scheduling concepts. Linux performance monitoring and Pressure Stall Information (PSI). Data analysis and predictive modeling techniques. Datacenter infrastructure, storage systems, and NFS performance optimization. How data-driven insights can be used to improve large-scale engineering workflows. More information about NXP in the Netherlands... #LI-f5d0