Delivering 1.23X Greater Efficiency and Over 45 PFLOPS
ATLANTA, GA & TAIPEI, TAIWAN / ACCESSWIRE / November 20, 2024 / ASUS and Ubilink today announced the successful establishment of a supercomputing facility that has positioned Taiwan in the top 31 on the global TOP500 list of supercomputers and secured a prestigious 44th ranking on the Green500 list, delivering 45.82 PFLOPS. This ambitious project, equipped with 128 ASUS ESC N8-E11 servers, each with either NVIDIA HGX H100 servers for a total of 1,024 GPU cards, is set to redefine the standards of AI computing. As the largest data center in Taiwan, the achievement by Ubilink – a company created by Foxlink, Ubitus and Shinfox Energy – and assisted by ASUS exemplifies the power of strategic collaboration and enables extraordinary performance milestones in the field of artificial intelligence. This world-class data center is also remarkable for handing customers the freedom to choose packages that draw only on renewable energy.
“The strategic partnership with ASUS has transformed our vision into reality, allowing us to build a world-class supercomputing facility that empowers the next wave of AI innovation in Taiwan and beyond,” said Michael Chen, General Manager of Ubilink. “The company’s advanced software solutions and comprehensive services have facilitated extraordinary performance achievements in a remarkable time frame of just three months.”
Highlights of this groundbreaking supercomputing project include:
World-class facility: The new green-energy AI computing center will feature 45.82 PFLOP/s, making it the largest of its kind in Taiwan.
Innovative solutions: The data center will offer public cloud services, AI-computing-power rentals and subscription-based services for training large language models, with a focus on smart manufacturing and livestreaming.
Advanced infrastructure: Equipped with 128 NVIDIA HGX H100 servers and 1,024 GPU cards, this facility is set to redefine the standards of AI computing.
Rapid deployment: The construction of the data center involved the ASUS Infrastructure Deployment Center (AIDC) tool for meticulous management of infrastructure, cooling systems and advanced automated tools, ensuring reliability and efficiency.
Superior computing efficiency: The ASUS-Ubilink project uses NVIDIA HGX H100 SXM5 80GB InfiniBand NDR400 servers with a total of 128 nodes of ESC N8-E11. The ASUS Solution Performance Team measured the performance and achieved an Rmax of 45.82 PFLOPS, with an efficiency (Rmax / Rpeak) reaching 66.08%. As seen in the table below, under the same GPU and high-speed network card conditions, the ASUS infrastructure solution is more efficient, delivering 1.23X the overall performance of competitors.[i]
ASUS, as the infrastructure supplier for this project and similar large-scale endeavors globally, delivers unmatched computing power and stability, reinforcing its commitment to driving AI innovation.
Learn more about this success story: https://servers.asus.com/stories/Ubilink-supercomputing-center.
Availability & Pricing
ASUS server solutions are available worldwide. Please contact your local ASUS representative for further information.
Press Contact:
Olivia Liu
Product Marketer
[email protected]
Notes to Editors:
ASUS Server website: https://server.asus.com
ASUS AI Server: https://www.asus.com/event/ai-server-solutions
ASUS Business website: https://www.asus.com/business
ASUS LinkedIn: https://www.linkedin.com/company/asus/posts
ASUS Business LinkedIn: https://www.linkedin.com/showcase/asus-business
ASUS Pressroom: https://press.asus.com
ASUS Global Facebook: https://www.facebook.com/asus
ASUS Global X (Twitter): https://www.x.com/asus
[i] Based on data from Top500.org. All the data used in this analysis is sourced from the published information of the TOP500 list, November, 2024. We assess the performance of each data center by comparing the computing efficiency, represented as the ratio of Rmax to Rpeak. For more details please visit: ASUS HGX H100 AI Servers Deliver 1.23X Boost in Performance Efficiency | ASUS Servers
[ii] Brand A uses NVIDIA HGX H100 SXM5 80GB, InfiniBand NDR400 with a total of 190 nodes
[iii] Brand B uses NVIDIA HGX H100 SXM580 GB, InfiniBand NDR400 with a total of 64 nodes
[iv] Efficiency is measured by Rmax / Rpeak, with a higher efficiency indicating better computing performance.
Contact Information
Brian Ngo
Senior Product Marketing Manager
[email protected]
SOURCE: ASUS
View the original press release on newswire.com.
