Overview

            COE Highlights

            Training Highlights

            oneAPI Ecosystem Highlights

            Marketing Highlights

Center of Excellence (COE) Highlights

• Knowledge Sharing for Aurora Machine Learning (ML) workloads: Organized a series of periodically held meetings between stakeholders from ALCF, the Intel ML development teams, the Intel-COE, and the Intel performance tools team, to share performance characterizations of various ML workloads from the ESP Machine Learning program. Based on interest expressed by the participants, these meetings are now held on a weekly basis. A training module explaining the use of VTune for ML workloads is being developed for use at ATPESC 2021.

Training

• Self-Paced Training: Join the 1,100 developers who have experienced Data-Parallel C++ through the DPC++ Essentials Learning Path, with hands-on Jupyter Notebooks and Intel® DevCloud.

• Aurora Early Adopter Series: Performance, Portability, and Productivity by Rakshith Krishnappa, Intel, Ben Odom, Intel on June 31 https://www.alcf.anl.gov/events/performance-portability-and-productivity

oneAPI Ecosystem

• The 3rd oneAPI Dev Summit was delivered on June 22/23^rd at ISC featuring 31 speakers from UNN, CERN, ANL, KTH, Stuttgart, UC London, UoBristol, Heidelberg, Codeplay, Leibniz Super Computing, Umass Lowell, KIT, UUIC, ORNL, Tangent Labs.  All sessions recorded and posted here.   Attendees participated from Karlsruhe Institute of Technology, Hewlett Packard Enterprise, GP Technologies Inc, L&T Technology Services, University of Florida, University of Texas, GNU/Weeb, Zuse Institute Berlin, Los Alamos National Laboratory, Georgia tech, inesc id Lisboa, Indiana University, Trigyn Technologies, Xilinx, Virginia tech, Instituto Nazionale di Fisica Nucleare, AI-Vets, Wayne University, Ansys, Siemens Healthineers, Codeplay, University of Bristol, Leukemia Research Association, SINTEC, Disney, University of Texas at Austin, University of Tennessee, University of Heidelberg.

• Community driven proof points: Univ of Tennessee/ Karlruhe Institute of Technology published “Porting a Sparse Linear Algebra Math Library to Intel GPUs” which was accepted in HETEROPAR’2021 and available in ARXIV, as well as new Dev Mesh Projects Arhat and A DPC++ Backend for the OCCA Portability Framework.  A Detect Spoofing with Fast Facial Recognition was published by Alessandro de Oliveira Faria of Oiti Technologies.  The open-source oneAPI based code integrates into existing banking and retail systems to detect facial biometrics.  oneAPI success story - “Ensure Social Media Credibility Using Artificial Intelligence” published by Professor Durga Toshniwal and her students from IIT Roorkee where they evaluate the vulnerability of social media users who may be implicit soft targets to such misinformation.

• Building Global Community: The Ministry of Education Teaching Learning Centre at Ramanujan College in India, in collaboration with Open Technology Forum organized an online two-week refresher course on next generation CS technologies which included oneAPI hands-on training sessions.  Delivered India  workshop on oneAPI AI Analytics Toolkit in collaboration with Analytics India Magazine with  254 developer attendees. Wuhan University of Technology delivered the first 2021 oneAPI on-campus workshop with 200 students participating in the hands-on lab session. XiDian University is pilot teaching DPC++ in their HPC class with students using the devcloud and Jupyter Notebook course module. After this pilot, learnings and enhanced lab materials will be built for fall 2021 class.

• Published next 3 episodes in the CERNopenlab vodcast series on oneAPI :

• Episode 3: interview with Bartlomiej Borzyszkowski focused on using oneAPI to detect anomalies in neural networks.
• Episode 4: Bob Duffy interviewed Viola Cavallini and Laura Cappelli about using oneAPI to reduce cost and increase productivity.
• Episode 5: Interview with Silke Donayre Holtz about her work using oneAPI at CERN openlab deploying generative models as well as comparing CPU’s & GPU’s to help increase system compatibility on the Large Hadron Collider.

Marketing Highlights

• Aurora software development: Packaging and early hardware: 
  https://alcf.anl.gov/news/aurora-software-development-packaging-and-early-hardware
• Porting a computational chemistry code to exascale architectures: 
  https://www.alcf.anl.gov/news/porting-computational-chemistry-code-exascale-architectures
• Porting ATLAS experiment codes to exascale architectures: 
  https://www.alcf.anl.gov/news/porting-atlas-experiment-codes-exascale-architectures
• Let’s Talk Exascale Code Development: WDMAPP—XGC, GENE, GEM:
  https://www.exascaleproject.org/lets-talk-exascale-code-development-wdmapp-xgc-gene-gem/

• HPC roundtable: Advances in computing and its impact on scientific research: 
  https://www.scientific-computing.com/webcast/hpc-roundtable-advances-computing-and-its-impact-scientific-research
• Argonne and Oak Ridge National Laboratories Award Codeplay Software to Further Strengthen SYCL Support Extending the Open Standard Software for AMD GPU:
  https://www.alcf.anl.gov/news/argonne-and-oak-ridge-national-laboratories-award-codeplay-software-further-strengthen-sycl

• Inspiring the Next Generation of Scientists & Coders:

https://connectedsocialmedia.com/19527/inspiring-the-next-generation-of-scientists-coders/

• Argonne scientists collaborate to create first full-core pin-resolved nuclear reactor model:
  https://www.alcf.anl.gov/news/argonne-scientists-collaborate-create-first-full-core-pin-resolved-nuclear-reactor-model
• ECP/ALCF Twitter Chat Live

https://twitter.com/search?q=ExascaleChat&src=typeahead_click

• Technology Innovations Already Make the Aurora Exascale Supercomputer a National Asset

https://itpeernetwork.intel.com/technology-innovations-already-make-the-aurora-exascale-supercomputer-a-national-asset/#gs.5pppso