Overview

Training Highlights

Intel/ALCF Marketing Highlights

oneAPI Ecosystem Highlights

COE Highlights

Training Highlights

• Technical Training Webinars: "Optimize Edge Compute Performance by Migrating CUDA to DPC++", presented by Jon Kim and Sofea Azrina had 172 live attendees, a satisfaction score of 84% and NPS of 58. Attendee comments included: “Helps keeps me up to date with new developments and tools” and “very informative and includes a demo”. " 

• The Intel® Certified Instructor Program expanded the course catalog to include OpenMP Offload Basics and now has 69 instructors certified to instruct on DPC++ Essentials and the first 9 instructors certified to also instruct on OpenMP Offload Basics from companies including ArrayFire, Locuz Enterprise Solutions, Paratera, AMAX, Bayncore and AlgoAsylum and universities including Paderborn University, Tianjin University, and Peking University.  Workshops by these instructors continue in May in China, Japan, Korea and Europe by instructors from Paratera, XLSoft, Escomsoft, Sarada, Banycore and more.

• The community-driven Intel Extreme Performance Users Group (IXPUG) will host this summer a Mid-Year Workshop on June 21, 2021 on heterogeneous computing experiences, with talks on computational work for COVID-19 remedies.  Register now (free)

Intel/ALCF Marketing Highlights

ALCF

• Aurora Software Development: Preparing Profiling Tools for Exascale:

https://www.alcf.anl.gov/news/aurora-software-development-preparing-profiling-tools-exascale

• Porting a lattice QCD code-suite to exascale architectures:

https://www.alcf.anl.gov/news/porting-lattice-qcd-code-suite-exascale-architectures

• Aurora Software Development: Compiler Readiness:

https://www.alcf.anl.gov/news/aurora-software-development-compiler-readiness

Intel/ALCF

• NWChemEx: Computational Chemistry Code for the Exascale Era

https://www.hpcwire.com/2021/04/29/nwchemex-computational-chemistry-code-for-the-exascale-era/

• Code Together: Driving Innovation Further Faster through Open Standards

https://connectedsocialmedia.com/19478/driving-innovation-further-faster-through-open-standards/

oneAPI Ecosystem Highlights

• Delivered the oneAPI DevSummit at IWOCL on April 26^th with 273 attendees and a 63 or “Great” NPS feedback score with attendees coming from a wide range of Labs, Universities, Partners and Customers including University of Cambridge, Microsoft, Oak Ridge National Laboratory, NEC EMEA, Synaptics, Molex, CERN, Siemens, Matrixx and XLsoft.  This “by developer for developer” conference featured 90% content delivered by external developers including from Codeplay, Xilinx, SENAI, Fujitsu, EURECOM, Creative Solutions Space, University of Salerno, Univ. of Pittsburgh, Univ. of Oregon, Heidelberg Univ., Universidade de Lisboa, UNSW Sydney, and Univ. of Chinese Academy of Sciences. Videos of all sessions are posted. 

• Next oneAPI Devsummit at ISC on June 22 - 23:  Agenda is published here and registration open here.  Speakers to include Erik Lindahl from GROMACs (oneAPI CoE), Ronal Keryell from Xilinx, David Hardy and Jaemin Choi NAMD (oneAPI CoE), Zheming Jin ORN, Penporne  Google, Dr. Iosif Meyerov UNN  (oneAPI CoE),  Dr. Yan Luo, U Mass Lowell, Niven Liber ANL and website is under construction.

• Community driven proof point published by Univ of SENA in Parallel Universe Magazine: Clícia Pinto, Lucas Batista, Pedro de Santana and Georgina Gozález published “Using oneAPI to Speed UP Finite-Difference Method: Migrating a CUDA-Based Stencil Computation to DPC++”  in the. The article describes a successful migration of a CUDA code to DPC++ using the compatibility tool.  

• oneAPI Innovator - Professor Hartwig Anzt has been added to the oneMKL Technical Advisory Board representing project Ginkgo and contributing to the future of the oneMKL Public Specification.  Prof Hartwig was initially curated up through grass roots community workshops and is now contributing to the oneAPI efforts on many fronts including product feedback, Ginkgo code porting, speakerships and now the TAB participation.

• New Vodcast series for the CERNopenlab – Intel oneAPI showcase has been launched featuring host Bob Duffy:
• Interview with openlab CTO Maria Girone who gives an overview of CERN, the tech and science behind the various oneAPI projects as well as an intro into the larger series – Watch here also: Twitter & LinkedIn.
• Episode 2 features and features an interview with openlab intern Marcin Swinarski on the work he did updating an inference engine for custom neural networks with oneAPI - Watch here also: Twitter & Linkedin.

• May edition of IDZ news (Video version) from host Stephanie Essin:  brings you some monthly updates, including a hot tip on a new podcast about visual analytics, plus details on Intel Inspector. Watch the full video here: https://intel.ly/3tYLla8

• The oneAPI DevSummit at IWOCL on April 26th ^.  This featured 90% content delivered by external developers. If you missed the DevSummit videos of all sessions are posted on the oneAPI and software. Intel sites.  273 external non-intel attendees participated and gave it a “Great” NPS feedback score of 63 with attendees coming from a wide range of Labs, Universities, Partners and Customers including University of Cambridge, Microsoft, Oak Ridge National Laboratory, NEC EMEA, Synaptics, Molex, CERN, Siemens, Matrixx and XLsoft. 

• oneAPI basics training series by Guy Tamir: Guy has produced 9 videos covering topics from “What is oneAPI” to toolkits, to deep neural networks to data science performance boosting.  Watch the series.

• Coffee Chats: Intel Vice President Pete Baker discusses Risk Analysis, Analytics and Artificial Intelligence on Intel Architecture in the Financial Services sector with Dmitri Goloubentsev of Matlogica and Sebastian Hahn of Quantifi. Watch here.

COE Highlights

To foster even more communication and stimulate collaboration, the COE launched the Argonne COE HPC/AI/Quantum-Computation seminar series at Intel. The series is aimed at highlighting, for the Intel audience, research done by scientists at ALCF, ANL, and the DOE. Invitees include all the engineering staff as most of the senior managers in the HPC program at Intel. For Intel engineers, the series offers exposure to DOE research covering a broad range of areas in science and technology. For senior managers, the series provides the ground level perspective on topical items that they might not normally get in executive sessions. The first session of the series was kicked off by ALCF’s Margaret Butler Fellow, Romit Maulik. Romit’s talk was titled: “In-situ scientific machine learning for computational physics with OpenFOAM”. The talk was extremely well received, and it drew an audience of 100+, including Intel VP Trish Damkroger and several members of her senior staff.