Nov. 11, 2021 — Rapid progress in HPC technologies is fueling breakthroughs in science across industries and beyond. HPC synergy with AI presents imperatives for more complex systems, algorithms, and diversified open architectures, while improving efficiency and reducing their carbon-footprint is a must. The number of supercomputers with GPU, FPGA and other accelerator cards in the TOP500 is increasing. Fugaku, powered by Arm-based processors, has continued to hold the top spot. On the other hand, new AI algorithms are now needed to train astronomical-sized models, and be commercialized efficiently in less performance-demanding and energy-consuming systems.HPC Connection Workshop (HPCC) Digital at SC21 on Nov. 15, 2021, world renowned experts and industry leaders will introduce the latest progress in two main open-source hardware technologies in ARM and RISC-V. They will elaborate on how the computation cost and the associated carbon foot-print for training trillion parameter neural networks can be drastically reduced by mimicking the brain’s sparsity. There will be also reports on the cutting-edge use of HPC and AI in different vertical applications including life science, material science, high energy physics, astrophysics, financial services, and so on. Featured speakers include:
During the 28th- Calista Redmond, CEO of RISC-V International, will talk about the mission to expand and engage RISC-V stakeholders, compel industry adoption, and increase visibility and opportunity for RISC-V within and beyond RISC-V International.
- Vikram Saletore, from Intel, will discuss how the confluence of HPC, and AI has begun to reshape the landscape in many verticals and will demonstrate customer use cases in health & life sciences, organic chemistry, astrophysics, and financial services.
- Brent Gorda, Senior Executive for HPC at ARM, will share a comprehensive overview on the current status of ARM in HPC, including V9 and V1 Neoverse IP.
- Anshumali Shrivastava, from Rice University, will discuss the first algorithm for training billion-parameter neural networks by mimicking the human brain’s sparsity, which exponentially reduce the computation cost and the associated carbon foot-print.
- Wang Qian, from Inspur, will go in details how her research team created the largest-to-date database for material science by using HPC to calculate the electronic charge densities of 17,418 inorganic materials, which can be used for diverse electronic structure analyses, and provide input for further machine-learning studies.
The 28th HPC Connection Workshop Digital will be online at 10 am CST on Nov. 15, 2021.
Agenda
- The Open Era of Computing Calista Redmond CEO, RISC-V International
- How AI and HPC Enable Different Verticals Vikram Saletore HPC-AI Architect, Principal Engineer, Intel Corporation
- 2021 Update on Arm in HPC Brent Gorda Senior Executive for HPC at Arm
- Scalable and Sustainable AI Acceleration for Everyone: Hashing Algorithms to Train Billion-parameter AI Models on a Commodity CPU, faster than Hardware Accelerators Anshumali Shrivastava Associate Professor in the Computer Science Department at Rice University
- High Performance Computing created Largest Material Science Database for Machine Learning Wang Qian HPC Application Engineer, Inspur Information
To register, visit this link.
About The HPC Connection Workshop
The HPC Connection Workshop is an international High Performance Computing event organized by the Asia Supercomputer Community and Inspur. This event takes place three times a year: during ASC in China, ISC in Germany, and SC in the USA. Top researchers and leading professionals from around the world gather at the workshops to discuss the latest developments and disruptive technologies in AI and supercomputing. Visit ASC’s website for videos of past highlights.
Source: Asia Supercomputer Community