Principles and Practice of Scalable Systems

closed

A key focus of the design of modern computing systems is performance and scalability, particularly in light of the limits of Moore’s Law and Dennard scaling. To this end, systems are increasingly being implemented by composing heterogeneous computing components and continually changing memory systems as novel, performant hardware surfaces. Applications fueled by rapid strides in machine learning, data analysis, and extreme-scale simulation are becoming more domain-specific and highly distributed. In this scenario, traditional boundaries between hardware-oriented and software-oriented disciplines are increasingly blurred. Achieving scalability of systems and applications will therefore require coordinated progress in multiple disciplines such as computer architecture, high-performance computing (HPC), machine programming, programming languages and compilers, security and privacy, systems, and theory and algorithms. Cross-cutting concerns such as performance, correctness and accuracy, and heterogeneity must be taken into account from the outset in all aspects of systems and application design and implementation. The aim of the Principles and Practice of Scalable Systems (PPoSS) program is to support a community of researchers who will work symbiotically across the multiple disciplines above to perform basic research on scalability and correctness and accuracy of modern applications, systems, and toolchains built on heterogeneous architectures. The intent is that these efforts will foster the development of principles that lead to rigorous and reproducible artifacts for the design and implementation of large-scale systems and applications spanning the full hardware/software stack. Importantly, as described below, PPoSS specifically seeks to fund projects that span the entire hardware/software stack and that lay the foundations for sustainable approaches for implementing performant, scalable, and correct and accurate computing applications that run on heterogeneous platforms.

High