ChapelCon '24 Keynote: A Case for Parallel-First Languages in a Post-Serial, Accelerated World

This is Paul Sathre's keynote talk from ChapelCon'24, recorded live on June 7, 2024.
Slides available at: chapel-lang.org/ChapelCon/202...
Abstract: Parallel processors have finally dominated all scales of computing hardware, from the personal and portable to the ivory tower datacenters of yore. However, dominant programming models and pedagogy haven't kept pace, and languish in a post-serial mix of libraries and language extensions. Further, heterogeneity in the form of GPUs has dominated the performance landscape of the last decade, penetrating casual user markets thanks to data science, crypto and AI booms. Unfortunately GPUs' performance remains largely constrained to expert users by the lack of more productive and portable programming abstractions. This talk will probe questions about how to rethink and democratize parallel programming for the masses. By reflecting on lessons learned from a decade and a half of accelerated computing, I'll show where Chapel 2.0 fits into the lineage of GPU computing, can capitalize on GPU momentum, and lead a path forward.
Bio: Paul Sathre is a Research Software Engineer in the Synergy Lab and NSF Center for Space, High-performance, and Resilient Computing (SHREC) at Virginia Tech. His research interests encompass systems software and tools and programming systems, particularly with respect to democratizing access to high-performance computing. More recently, his work has focused on the intersection of computational co-design with portable and productive languages, tools, and libraries for heterogeneous computing.