Research

Modular and automated monitoring and analysis infrastructures for HPC systems

Holistic performance engineering and operational data analytics

Novel techniques and tools for performance evaluation and prediction

Performance modeling, benchmarking, prediction, and I/O characterization studies of emerging HPC workloads (DL, AI, Big Data, Workflows etc.)

Micro-benchmarking, application benchmarking – Workflows, Reproducibility, Comparability, Portability

High-performance I/O middleware libraries, I/O services, metadata and complex data management

Tuning and auto-tuning of emerging HPC applications and algorithms

High-level parallel programming languages and frameworks

Correlation between network communiation and data access patterns 

Innovative approaches for extreme-scale storage systems and multi-tier HPC architectures