China’s Iluvatar CoreX’s first AI computing supernode squeezes most GPUs in 1 system

What happened

Iluvatar CoreX introduced the Tiangai 300 IXL144, a supernode system that integrates 144 GPUs, 18 CPUs, and 10.1 terabytes of memory into a single cabinet. This system is designed for large-scale AI model training and inference and is part of a trend to scale computing power amid US export restrictions on advanced chips.

Why it matters

The Tiangai 300 IXL144 is significant for AI developers and businesses as it offers a scalable and efficient solution for AI model training and inference, especially in the context of restricted access to advanced chips from Nvidia. The system's cost efficiency and integration of multiple components make it a valuable resource for expanding computing capabilities.

• The Tiangai 300 IXL144 represents a major advancement in AI computing infrastructure.
• It offers a cost-effective solution for large-scale AI model training and inference.
• The system's integration of multiple components enhances efficiency and scalability.
• It is a response to US export restrictions on advanced chips and rising AI demand.

The bigger picture

The Tiangai 300 IXL144 has the potential to significantly reduce costs for AI operations and improve computational efficiency, making it a valuable asset for businesses and developers looking to scale their AI capabilities in a competitive market.

Iluvatar CoreX's Tiangai 300 IXL144 supernode system marks a significant milestone in AI computing, offering a scalable and efficient solution for large-scale AI operations.

Technical details

• The system integrates 144 GPUs and 18 CPUs into a single cabinet.
• It includes around 10.1 terabytes of memory.
• The flexible single-cabinet architecture allows for efficient resource management.
• The system is designed for large-scale AI model training and inference.

FAQ