WOLFHPC 2016

Sixth International Workshop on
Domain-Specific Languages and High-Level Frameworks for High Performance Computing

November 13, 2016

Half-day workshop in conjunction with

SC16: The International Conference for High Performance Computing, Networking, Storage, and Analysis

Salt Lake City, Utah, USA

Held in cooperation with ACM SIGHPC

| Home | Advance Program |

Workshop location: 251-F (Salt Palace Convention Center)

Please provide workshop feedback here

Advance Program

Session 1

9:00	Opening remarks
9:00-10:00	Keynote Talk: Prof. Robert A. van de Geijn [webpage] University of Texas at Austin DSLs for DLA: Past the BLAS [Talk]

Break: 10:00-10:30

Session 2:

10:30-11:00	Simon A. F. Lund, Mads R. B. Kristensen and Brian Vinter "Automatic mapping of array operations to specific architectures" [Talk][Online slides] [Project website] [Benchmark suite]
11:00-11:30	Navjot Kukreja, Mathias Louboutin, Felippe Vieira, Fabio Luporini, Michael Lange and Gerard Gorman "Devito: automated fast finite difference computation" [Talk]
11:30-12:00	Johannes Spazier, Steffen Christgau and Bettina Schnor "Efficient parallelization of MATLAB stencil applications for multi-core clusters" [Talk]
12:00-12:30	Charles Yount, Josh Tobin, Alexander Breuer and Alejandro Duran "YASK--yet another stencil kernel: a framework for HPC stencil code-generation and tuning" [Talk] [Software]
12:30	Closing remarks

Keynote Talk

Prof. Robert A. van de Geijn , The University of Texas at Austin

Title: DSLs for DLA: Past the BLAS

Abstract: The linear algebra software community was among the first to recognize and embrace standard interfaces for high-performance computing, most notably the Basic Linear Algebra Subprograms (BLAS). These early forms of Domain Specific Languages (DSLs) supported portable high performance and a certain level of readability. More recently, the FLAME project has demonstrated how abstraction in notation can translate into an elegant DSL for dense linear algebra computations while facilitating systematic (and even mechanical) generation of proven-correct, high-performance algorithms. After almost two decades of foundational research that started as a pedagogical exercise, this DSL now enables a new DLA software stack. Benefiting from hindsight, we share insight.

Bio: Robert van de Geijn is professor of Computer Science and core member of the Institute for Computational Engineering and Sciences. He received his Ph.D. in Applied Mathematics from the University of Maryland, College Park, in 1987.

Prof. van de Geijn's interests are in linear algebra, high-performance computing, parallel computing, and formal derivation of algorithms. He heads the FLAME project, a collaboration between UT Austin, Universidad Jaume I (Spain), RWTH Aachen University (Germany), and Carnegie Mellon University. This project pursues foundational research in the field of linear algebra libraries and has led to the development of the open-source BLAS-like Library Instantiation Software (BLIS) framework and the libflame library. One of the benefits of these libraries lies with their impact on the teaching of numerical linear algebra, for which van de Geijn received the UT President's Associates Teaching Excellence Award. He has published several books and more than 100 refereed publications, and is co-instructor of the Massive Open Online Courses (MOOCs) titled "Linear Algebra - Foundations to Frontiers" (LAFF, offered by edX) and "LAFF-On Programming for Correctness" (to be offered by edX in Spring 2017).