From Serverless to Supercomputers: Systems, Performance, and the HPC-Cloud Convergence
Marcin Copik ETH Zürich
Course Summary |
About the lecturer |
Location and schedule |
Cloud computing and high-performance computing (HPC) have long developed in parallel, with different programming models, resource management strategies, and performance expectations. Today these worlds are converging - cloud providers offer HPC-grade hardware and networks, HPC systems adopt elastic and serverless workflows, and modern AI workloads push both in the same direction. This course offers a systems-oriented introduction to that convergence, starting from foundations and building toward current research problems. The course begins with modern cloud architectures and the serverless (Function-as-a-Service, FaaS) model, then turns to methods for measuring and modeling performance and cost in these environments, and closes with open questions at the HPC-cloud boundary: programming models for scientific computing on serverless, performance portability, GPU resource management, and the systems infrastructure emerging around AI workloads. The course is designed to be accessible to PhD students without prior background in cloud computing or HPC. Lectures will be accompanied by hands-on exercises that run locally on a laptop (Linux, macOS, WSL). Students who want to try them on a real cloud platform can set up a free-tier account in advance (AWS, Google Cloud, or Azure). Our benchmarking suite's documentation provides an overview of the supported platforms and their setup: https://github.com/spcl/serverless-benchmarks/blob/master/docs/platforms.md
About the lecturer:Marcin Copik is a researcher at the Scalable Parallel Computing Laboratory (SPCL) at ETH Zürich. His research is at the intersection of cloud computing and high-performance computing, spanning serverless systems, HPC-cloud convergence, performance analysis, and more recently the systems infrastructure for emerging AI workloads. He is the primary author and maintainer of SeBS, the Serverless Benchmark Suite. He received his PhD from ETH Zürich in 2024, advised by Torsten Hoefler. His work has been recognized with the ACM SIGHPC Doctoral Dissertation Award (2025), the ACM-IEEE George Michael Memorial HPC Fellowship (2022), the Microsoft Research PhD Fellowship (2021), and Gold Medals at the ACM Student Research Competition at the ACM/IEEE Supercomputing Conference.
Location and schedule:
|
|
|
||||||||||||||||