Keynote

We are happy to announce our keynote speaker is Dr. Albert-Jan Yzelman. Please find the details of the talk below.  

Title: Algebraic Programming

Abstract: We first recall the concept of the "humble programmer" who seeks to achieve high productivity, and contrast it to the concept of the “hero programmer” who seeks to achieve peak performance. Given that the complexity of programming novel architectures increases and that different systems increasingly rely on different heterogeneous components, humble programming models that automatically achieve both scalable and high performance regardless of the system details are becoming a necessity. 

This talk introduces the free and open-source Algebraic Programming (ALP) paradigm as one such a candidate solution. With ALP, programmers must annotate their code with algebraic information, which the compiler exploits to automatically optimise, to automatically parallelise, to detect programmer errors, and to reject code expressions that do not scale or violate algebraic laws. ALP/GraphBLAS is a C++ dialect of the GraphBLAS standard that implements this vision for generalised sparse linear algebra, which covers workloads ranging from classical numerical linear algebra to graph computations, machine learning, and AI. 

We detail how ALP/GraphBLAS achieves scale-out over hybrid shared- and distributed-memory architectures, and present extensions that cover generalised dense linear algebra as well as vertex-centric programming. We touch on the importance of interoperability with industry-standard frameworks like Spark, and demonstrate the high performance that the framework achieves on recent architectures—even when used from external distributed frameworks. Particular attention will be spent on our novel nonblocking execution work which allows optimisation across linear algebra primitives, so enabling –and achieving– multiple factor speedups versus the state of the art.

Bio: Albert-Jan N. Yzelman is a Research Scientist and Expert at the Computing Systems Laboratory in the Huawei Zürich Research Center, and previously held Senior and Principal research positions at Huawei Paris. He obtained his doctorate from Utrecht University, and has held post-doctoral positions at KU Leuven and the Intel ExaScience Labs. His research interests center around paradigms for irregular and parallel computing, focusing on easy to use, yet high-performance, scalable, as well as portable programming principles and associated system design.