March 2021 Newsletter

Around the globe we are ever-inspired by the research and experiences shared in our IXPUG open-forum events and webinars. Check out the recent replays, podcasts, and more in this issue. Hear and learn through IXPUG’s free exchange of information with leading experts in HPC and AI.

EVENT REPLAYS

• Performance Optimizations for End-to-End AI Pipelines Meena Arunachalam Ph.D, Principal Engineer and Vrushabh Sanghavi, Sr. Software Engineer of the Machine Learning Product Architecture group at Intel Architecture, Graphics and Software discuss applying AI in application domains and simulations at-scale, fast experiments, prototyping, and model development at an accelerated pace that was never possible before.
• Migrating from CUDA-only to Multi-Platform DPC++ Steffen Christgau, research associate in the Algorithms for Innovative Architectures research group of the Supercomputing Department at Zuse Institute Berlin spoke about heterogeneous architectures, portable code, and open-source tools, joined by colleague Marius Knaust during the Q&A.
• IXPUG Workshop at HPC Asia 2021 Replay the keynote Advancing HPC Together and many other sessions from this half-day, open workshop on high performance computing applications, systems, and architecture with Intel technologies. View the full playlist HERE.
• IXPUG Annual Conference 2020 catch up on sessions from our 4-day meeting, available to replay via the full playlist HERE. Includes the keynotes by Michael Wong, distinguished engineer at Codeplay: The Landscape of Modern Parallel Programming Using Open Standards and How HPC Developers Using SYCL Need to Deal with Data Movement.

PODCASTS

• Let’s Talk Exascale Code Development: EQSIM Brian Homerding, Argonne National Laboratory and Houjun Tang, Lawrence Berkeley National Laboratory discuss simulating seismic wave propagation on HPC systems, optimizing code, and tools available to help researchers and developers to prepare their code for Aurora and other exascale systems.
• A Proving Ground for Open Standards – Andrew Lumsdaine, Chief Scientist at Northwest Institute for Advanced Computing, discusses how parallelism and heterogeneity are hand-in-glove for achieving performance, underscores the value of open standards, and provides ways developers can shape ISO C++, SYCL, DPC++, and oneAPI.
• At the OpenMP Forefront – Bronis de Supinski, CTO of Livermore Computing at Lawrence Livermore National Lab and Chair of the OpenMP Language Committee, discusses 5.1 and beyond, how the C++ ecosystem is evolving, and why Python in HPC.

Resources

• Article: CERN Accelerates Simulation Workloads with AI Researchers used the Intel® AI Analytics Toolkit to obtain higher performance for inferencing in Intel® Xeon® Scalable processors with Intel® Deep Learning Boost. For more from CERN researchers, see the lightning talk Mixed-Precision Arithmetic for 3DGAN to Simulate High Energy Physics Detectors from IXPUG Annual Conference 2020.
• Article: Accelerating Linear Models for Machine Learning: Linear Regression Has Never Been Faster Victoriya Fedotova, Machine Learning Engineer at Intel Corporation covers how to train ridge regression models using a version of scikit-learn that’s optimized for Intel CPUs, then compare the performance and accuracy of these models trained with the vanilla scikit-learn library.
• Article: Servers Are Becoming More Heterogeneous Jeff McVeigh, vice president and general manager of data center XPU products and solutions at Intel, shares his views on XPUs, oneAPI, and what data centers will look like 5-10 years from now.
• oneAPI specification: https://www.oneapi.com/documentation/ As part of the oneAPI industry initiative, oneAPI elements have been open sourced to make it easy for developers to take advantage of oneAPI in their projects.
• Open access book: Data Parallel C++: Mastering DPC++ for Programming Heterogeneous Systems using C++ and SYCL (free, includes code examples)
• Code walk-throughs:
  • DPC++ Foundations Code Sample Walk-Through
  • DPC++ Unified Shared Memory (USM) Code Sample Walk-Through
  • Compile and Run MPI Programs Using DPC++ Language
  • Migrating OpenCL™ Designs to DPC++

ABOUT IXPUG

• IXPUG is an independent users group whose mission is to provide a forum for the free exchange of information that enhances the usability and efficiency of scientific and technical applications running on HPC, AI, and data analytics systems using advanced Intel technologies. IXPUG is administered by representatives of member sites that operate large Intel-based HPC systems.
• Looking for presentations from our past Workshops, Meetings, and BOFs? Check the IXPUG Resources searchable repository of slides and recordings from all of our events.
• What extreme performance technologies do you want to learn more about in 2021? We want to hear from members! Please contact us at This email address is being protected from spambots. You need JavaScript enabled to view it..