Bridging Global Infrastructure to Address Local Challenge with HPC - Asst. Prof. Dr. Worawan Diaz C.

Summary

Central message

The lecture argues that high-performance computing (HPC) is most valuable when it connects global computing infrastructure to specific local needs. The “bridge” is not just the supercomputer: it also requires people who can connect community problems, domain expertise, data, software, and computing resources into a usable solution.

Main ideas

• What HPC does: Supercomputers perform calculations far faster than ordinary computers. Their performance is commonly measured in FLOPS—floating-point operations per second—because many scientific and engineering problems rely on calculations involving real numbers.
• Why HPC matters: Large-scale simulations help researchers study processes that are too complex or slow to investigate directly. The lecture uses molecular simulations related to COVID-19 as an example, describing simulations as a way to examine phenomena, test factors, and support forecasting.
• HPC is part of a wider ecosystem: The speaker traces the development of computing centers from early scientific facilities to national and regional centers, and describes today’s combination of HPC, AI infrastructure, cloud services, and emerging quantum systems. She stresses that machines alone are insufficient: industry, researchers, government, and local communities must also be connected.
• Integration is the key role: An “integrator” helps carry a problem from its local context through data collection, modeling, computing, and delivery of results—potentially as a website or app. This work is collaborative: computing specialists need to work with people who understand the relevant science or community need.
• Local problems can benefit from global infrastructure: Examples discussed include flood forecasting in Bangladesh, agricultural mapping in Norway, disease research in Uganda, and air-quality forecasting in Thailand. These applications show how local data and expertise can be combined with HPC to support decisions and services.
• The aim is practical benefit: Projects should address real needs—such as disaster response, food production, health, pollution, or energy use—and make results useful to people and decision-makers. The speaker also connects HPC work with sustainable development goals and energy efficiency.

Project examples

• Electric-vehicle travel planning: A student project combined route planning that accounts for charging stations with AI-generated, tour-guide-style narration.
• Building energy use: Researchers combined a physical building-energy model with agent-based simulation of people’s behavior. They modeled policies such as when air conditioning may be used or when buildings should close, while accounting for spaces that cannot simply be shut down.
• Local plants and asthma treatment research: Computing students supported biological researchers investigating local plants as possible ingredients for asthma sprays. The work illustrates the need to combine computing with biology, traditional medicine, and pharmaceutical expertise. The spray was still in development.
• Minority-language tools: A student developed an open-source translation system for Mon, addressing the lack of speech-to-text and text-to-speech support described in the lecture.
• Disaster and evacuation modeling: A demonstration showed how simulations can model traffic and people’s movement during incidents or disasters.
• Data-center cooling: A Singaporean startup used HPC-based computational fluid dynamics simulations to study airflow and temperature in data centers, helping develop and test cooling designs.

Practical route for students to become HPC integrators

1. Build technical foundations. Study computer architecture, networking, operating systems, parallel and distributed computing, parallel programming, GPU programming, data science, and—if interested—quantum computing.
2. Get hands-on access. The speaker recommends competitions, seasonal schools, and training programs as ways to learn HPC without necessarily paying for access. She describes a regional competition that provides access to several computing systems and training materials.
3. Learn the HPC workflow:
   • Activate an account through the computing center’s portal.
   • Create an SSH public/private key pair and register the public key.
   • Connect to the system using SSH; HPC systems commonly use Linux.
   • Prepare a batch script specifying requested resources, such as nodes and processors.
   • Submit the job and let the scheduler run it.
   • Download the output and visualize or analyze it on a personal computer.
4. Move from learner to practitioner. Practice Linux, remote access, file transfer, shell scripting, MPI, job submission, Python environments, PyTorch, and performance evaluation. Benchmarking competitions can provide structured practice.
5. Learn to facilitate and mentor. Join HPC schools and communities, observe how centers and users work together, and share what you learn with others.
6. Develop a local project with partners. Identify a real local challenge, find lecturers and domain experts, form a multidisciplinary team, and seek support or funding from universities, computing organizations, or industry partners.
7. Make the work reproducible and shareable. Document the project and its artifacts so others can understand, repeat, and reuse the work. Presenting at conferences can help students gain research experience and feedback.
8. Build community and contribute. Student chapters and other communities can provide training, peer support, access to opportunities, and connections to industry and research networks.

Challenges raised in the Q&A

• Environmental impact: The speaker says energy and sustainability are increasingly being addressed in HPC education and operations. She also notes that improving code performance can reduce job run times and may reduce energy use.
• Deploying AIoT systems: Deployment involves more than the HPC model. Sensors must be calibrated to local conditions, data from different sources must be cleaned and reconciled, and workloads may need to be divided among edge devices, cloud systems, and HPC. A successful project may require expertise in electronics, low-level programming, cloud and full-stack development, security, and HPC. The speaker’s advice is to form a diverse team and build partnerships.

Speakers and sources featured

• Assistant Professor Dr. Worawan Diaz Carbalo — keynote speaker; her name is rendered inconsistently in the auto-generated subtitles.
• Priscilia Edward (“Cecil”) — moderator and event host.
• Dr. Aries — chairperson of the organizing committee; delivered the welcome remarks. The subtitles also render the name as “Ariatia.”
• Q&A participants: Kenya Nurma Fabla and Fadli Rammania, identified in the subtitles as questioners.
• Embedded audiovisual sources: An NVIDIA video about COVID-19 molecular simulation and Omniverse; a Barcelona Supercomputing Center/quantum-computing news segment; a brief Minecraft-related YouTube clip; a KADIF Korea–ASEAN digital-infrastructure promotional video; and a video about Singapore’s Cool Logic data-center cooling work. The narrators in these clips are not identified in the subtitles.