Name: Khaled Ibrahim
Pronouns:
Biography:
Khaled Ibrahim is a scientist with the Computational Research Division. He joined Lawrence Berkeley Laboratory in Jan. 2009. He obtained his PhD in computer engineering from North Carolina State University in 2003. His research interests include performance analysis, modeling and tuning for High Performance Computing Simulation, and Deep-Learning Workloads. He is also interested in communication runtime optimizations for irregular communication patterns.
Institution/Lab: Lawrence Berkeley National Laboratory
Website:
SRP Collaboration Topic/Title: HPC Workflows Performance Modeling and Tuning
Field or research area: Performance Modeling, tuning, and Optimization
Please select all the topical areas that apply to your project:
Computer Science (i.e., architectures, compilers/languages, networks, workflow/edge, experiment automation, containers, neuromorphic computing, programming models, operating systems, sustainable software); High-Performance Computing; Machine Learning and AI
Brief Abstract:
In the performance and algorithms group, we tackle various applications and workflow performance optimization problems, using refactoring techniques of these codes to leverage DOE supercomputing machines efficiently. We aim to enable the development of cutting-edge solutions to tackle computationally challenging problems. Our SRP visitors are expected to engage in ongoing research efforts within our group to engage in an experience in developing performance modeling and tuning methods. We also encourage application developers with performance challenges to engage with us to apply our developed methods in improving the performance of their code in leading DOE computational environments. We also encourage application developers with performance challenge to engage with us to apply our developed methods in improving the performance of their code in leading DOE computational environments.
Desired relevant skills, background, or interests:
Skills in performance modeling, profiling, and tuning. Familiarity with HPC programming models and/or with DL and ML frameworks.
Other comments:
Do any special requirements apply? In-Person Only
Other, specify:
Keywords:
AI workloads Performance Portability ML Methods
Lightning Talk Title: Performance for AI Workloads and ML Techniques for Performance Portability