Our PADAL workshop series started in Lugano-Switzerland and first workshop compiled an exploratory report contributed by all 30+ participants to define challenges and open research areas on data locality. The combined outcome of the second workshop held in Berkeley-USA, and the third workshop held in Kobe-Japan resulted in a survey journal article published in IEEE TPDS (DOI: 10.1109/TPDS.2017.2703149). While the fourth workshop, held in Chicago-USA, had a machine learning focus and the fifth workshop, held in INRIA Bordeaux- France had an extreme heterogeneity focus.
The theme for this year’s workshop will be centered around machine imbalance that leads to data locality issues in applications and how we can address this issue with heterogeneous vs homogeneous acceleration, memory-centric computing, CPU-free computing, computing in place, disaggregated computing and with other solutions. This topic is timely because current processor architectures are increasingly imbalanced when it comes to their flops/words ratios. Diverse solutions to address the problem are needed in all aspects of computing including architectures, programming models, tools, solvers, and platforms.
Workshop: September 4-6, 2023
The workshop will be held at the Koç University, Istanbul, Turkey
Anshu Dubey, Argonne National Laboratory, and University of Chicago, USA
Didem Unat, Koç University, Istanbul, Turkey
John Shalf, Lawrence Berkeley National Laboratory, California, USA
Emmanuel Jeannot, INRIA Bordeaux, France