In a remarkable stride toward global AI innovation, Taiwan now houses one of the world’s most advanced and efficient supercomputing facilities. Announced on November 19, 2024, the partnership between ASUS and Ubilink has elevated Taiwan’s status on the global stage. The newly established facility has positioned the nation in the Top 31 on the 500 list of global supercomputers and secured an impressive 44th spot on the Green500 list, boasting a performance of 45.82 PFLOPS. This accomplishment is not just a technological feat but a testament to Taiwan’s commitment to sustainable and efficient AI computing.
A Vision Transformed into Reality
The ambitious project, driven by Ubilink a joint venture by Foxlink, Ubitus, and Shinfox Energy has made history. The state-of-the-art facility, supported by 128 ASUS ESC N8-E11 servers equipped with 1,024 NVIDIA HGX H100 GPU cards, redefines the standards for AI computing.
“This strategic partnership with ASUS has turned our vision into reality,” said Michael Chen, General Manager of Ubilink. “The combination of ASUS’s advanced software solutions and comprehensive services enabled us to achieve extraordinary performance milestones in just three months.”
Let’s delve deeper into what makes this project groundbreaking.
The Powerhouse of AI Computing
Performance Like No Other
This facility delivers an incredible 45.82 PFLOPS, setting a new benchmark for AI computing in Taiwan. Ranked among the best globally, it is the largest supercomputing center in the country.
Energy-Efficient Design
Ranking high on the Green500 list, the center is powered by renewable energy, aligning with the global shift toward sustainability. Customers can opt for packages entirely reliant on green energy, ensuring environmental responsibility.
Cutting-Edge Infrastructure
The facility utilizes 128 NVIDIA HGX H100 servers with 1,024 GPU cards, creating a computing powerhouse capable of handling the most complex AI applications. The infrastructure supports high-demand tasks such as smart manufacturing, large-scale AI training, and live streaming.
Efficiency Beyond Competition
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 an Rmax of 45.82 PFLOPS with an efficiency of 66.08%. This is 1.23X more efficient than competitors under similar conditions, highlighting the superior infrastructure designed by ASUS.
Speedy Execution, Robust Deployment
The entire supercomputing facility was constructed in just 3 months. This rapid deployment was made possible by the ASUS Infrastructure Deployment Center (AIDC), which meticulously managed infrastructure, cooling systems, and automation tools.
Key features include:
- Liquid-Cooling System: Ensures efficient power consumption and thermal management for thousands of GPUs.
- Auto-Detection & Recovery Tools: Advanced tools minimize hardware failures and downtime, ensuring seamless operations.
- Integrated Development Platform: Simplifies the deployment and optimization of AI applications, accelerating innovation.
Transforming Industries, Driving Global Impact
The ASUS-Ubilink supercomputing project is not just a national achievement but a global milestone. It aims to revolutionize industries through AI-powered solutions. Applications include:
- Smart Manufacturing: Leveraging AI to enhance industrial processes and efficiency.
- Training Large Language Models: Supporting AI advancements in natural language processing and more.
- AI Cloud Services: Offering rental and subscription-based services to democratize access to high-performance computing.
Challenges Met with Excellence
Building such a vast and high-performance infrastructure posed significant challenges. From managing power consumption to ensuring uninterrupted operation, every aspect required precision. ASUS’s expertise ensured that the facility met these challenges head-on, creating a stable and efficient ecosystem for AI computing.
Key Challenges included:
- Ensuring stability across 1,024 GPUs to avoid costly interruptions.
- Managing cooling systems for such an energy-intensive setup.
- Implementing advanced automation to maintain system reliability.
Role of ASUS: Pillar of AI Innovation
As the exclusive hardware supplier for this ambitious project, ASUS has proven its commitment to pushing the boundaries of AI technology. Known globally for its reliable and high-performance servers, ASUS delivered:
- Fastest Equipment: Cutting-edge servers designed for peak performance.
- Professional Support: A dedicated team ensuring smooth deployment and operation.
- Integrated Software Services: Tools that simplify the optimization of AI applications.
This partnership showcases the power of collaboration in driving technological progress.
Future-Ready and Globally Significant
The ASUS-Ubilink supercomputing facility is a leap forward for AI innovation. Its emphasis on sustainable energy and efficient infrastructure sets a new standard for supercomputing centers worldwide.
As industries increasingly rely on AI, facilities like this will play a pivotal role in enabling high-demand applications. Whether it’s enhancing manufacturing efficiency or powering AI-driven solutions, the global impact of this center is undeniable.
A Milestone in AI Computing
The ASUS-Ubilink supercomputing project is a shining example of what strategic partnerships can achieve. With its groundbreaking performance, green energy focus, and industry-transforming capabilities, it is not just a milestone for Taiwan but a beacon of innovation for the world.
This achievement underscores ASUS’s vision of driving AI-powered progress while prioritizing sustainability and efficiency. For businesses and researchers looking to harness the power of AI, this facility offers unparalleled opportunities to innovate and thrive.