Success-Case

To Empower Scientific Study, NTNU Opens Center for Cloud Computing

Apr 07,2021 by GIGABYTE

High performance computing has a critical role to play in modern-day scientific research. The College of Science at National Taiwan Normal University anticipated the importance and rapid development of HPC. It purchased GIGABYTE servers to establish the Center for Cloud Computing on its campus, with an eye towards completing research projects more quickly and cultivating professionally trained experts in the field.

The College of Science at National Taiwan Normal University is a prestigious institute of science education, famed for having some of the best talent and most comprehensive sets of equipment in the country.

Producing Weather Simulations and Forecasts in a Flash with HPC Systems

High performance computing (HPC) is one of the main tools of modern scientific study. Researchers at the College of Science at National Taiwan Normal University (NTNU) use the computing power of high performance servers to run numerical simulations, develop mathematical models, and solve complicated scientific formulas. These tools help them get to the root of the problem more quickly, allowing them to execute carefully planned research projects of the utmost importance. For example, the Department of Earth Sciences at NTNU would not be able to get by without the frequent analysis of meteorological data, a process that involves turning a sea of information into accurate simulations of catastrophic weather events. They can predict just about anything, from the amount of rain during monsoon seasons and extreme precipitation events, to the path a typhoon takes and how it will affect rainfall in areas with differing geography in Taiwan. All of this can be done with an immense amount of data in a very short period of time to produce useful, informative predictions.

Learn More:
《Glossary: If you don't know what HPC is, find out here》
《Glossary: If you don't know what Big Data is, find out here》
《Decoding the Storm with GIGABYTE’s Computing Cluster》

Radar data and other information gathered from meteorological observations can be used by the Center for Cloud Computing to make weather simulations and predictions. (Images provided by Dr. Chung-Chieh Wang, the Department of Earth Sciences, NTNU)

The Center for Cloud Computing Expedites Scientific Discoveries

To handle research projects that are very data-intensive, some faculty members set up computing clusters in their own labs to handle the workload. The lack of resources means one set of equipment only has the capacity to handle one research project at a time, and not very efficiently at that. Other professors hire third-party providers to do the heavy lifting, but it takes time to find such vendors and schedule computing services from them, so important research may end up being delayed.

In order to alleviate these problems, the College of Science at NTNU decided to establish the Center for Cloud Computing to support the research projects of its faculty and students. They did so with servers and professional consultation from GIGABYTE.

Glossary:

《If you don't know what Cloud Computing is, find out here》

《If you don't know what Computing Cluster is, find out here》

Unlike the rudimentary computing clusters assembled by individual professors, the Center for Cloud Computing is able to schedule a large number of data processing tasks through the internet and automatically divide them into smaller workloads. The workloads are then assigned to computing clusters made up of multiple servers, which work together to complete the calculations and send back the results. This is a much faster way of dealing with data-intensive projects, and it allows system resources to be shared among the dozens of research teams in the Departments of Mathematics, Physics, Chemistry, Earth Sciences, Life Science, Computer Sciences & Information Engineering, to the mutual satisfaction of all involved.

GIGABYTE high density servers provide optimal performance and flexible scalability.

Unlocking the Servers' Potential with Leading Structural Design

GIGABYTE drew upon its years of experience and accumulated knowledge in the research and development of server technologies to offer a more comprehensive proposal than competing providers. Taking into account the limited space the campus has for server racks, GIGABYTE optimized the structural design of an HPC system by using servers from different comprehensive product lines: the H270-H70 High Density Server makes up the CPU node, the G190-H44 GPU Server with big data analysis capabilities makes up the GPU node, and the R180-F28 Rack Server makes up the storage portion. Each product line is able to fulfill its assigned role while providing competitive computing power.

"GIGABYTE has experience with and in-depth knowledge of everything, from the preliminary planning of the HPC system, to post-deployment management and maintenance. They were a big help in building and fine-tuning a HPC computing cluster on our campus," says a professor at NTNU's College of Science.

GIGABYTE chose the most suitable components, printed circuit board materials, and high-performance power delivery designs to optimize the integrated hardware architecture. This allowed the system to achieve low-latency and high-bandwidth data transmission with excellent reliability. GIGABYTE also adjusted software settings such as the BIOS and BMC to improve overall performance. The system underwent a stress test, during which multiple users logged in at the same time to check the solution's reliability.

What's more, GIGABYTE provides a complete range of after-sales tech support and services. Based on the research team's actual user experience, GIGABYTE can further adjust and fine-tune the overall system to optimize CPU and memory usage, allowing the servers to demonstrate the full extent of their computing prowess and stability.

Learn More:

《More information about GIGABYTE's High Density Servers》

《More information about GIGABYTE's GPU Servers》

《More information about GIGABYTE's Rack Servers》

《Glossary: If you don't know what a Baseboard Management Controller (BMC) is, find out here》

Starting in 2016, NTNU has already offered six semesters of HPC management courses in hopes of training more experts in the field.

NTNU Trains HPC Experts with Solutions and Support from GIGABYTE　　　

In order to ensure NTNU can reap the benefits of their Center for Cloud Computing for years to come and with minimal hassle, GIGABYTE compiled a compendium of relevant professional knowledge, and provided training courses about HPC system management, maintenance techniques, and troubleshooting to endow students and staff with the professional ability to keep the system running smoothly. One good thing led to another, and it wasn't long before NTNU began officially offering courses on "HPC Management" based on the systemized materials provided by GIGABYTE, with an eye towards cultivating more experts in the field of HPC management.

Continuous Development of HPC Set to Make a Lasting Impact

According to the latest scientific studies, using HPC to run complicated and sophisticated numerical simulations can overcome various technological barriers and grant us a deeper understanding of the "knowns", as well as bold predictions about the "unknowns". For this reason, HPC is an indelible part of pushing scientific progress forward.

The Center for Cloud Computing at NTNU's College of Science not only helps researchers uncover new scientific knowledge, it also proactively trains a fresh class of new talent to systematically engage in the research and development of related areas of study. This is in the spirit of the principle GIGABYTE has long believed in: cutting-edge technology can help "Upgrade Your Life" by giving people the tools they need to unlock their potential, bringing forth boundless opportunities in different fields of scientific research.

Learn More:

《How Can High Performance Computing Render Art and Imagination?》

《GIGABYTE’s GPU Servers Help Improve Oil & Gas Exploration Efficiency》

《As Good as Gold - High Performance Computing Accelerates Oil Extraction》