What benefits does upgrading from 800Gbps to 3.2Tbps bring?

Question

Accepted Answer

The upgrade to a 3.2Tbps AI DCI Superhighway offers significant enhancements in speed, latency, and efficiency for AI infrastructure and research.

The transition from an 800Gbps to a 3.2Tbps AI Data Centre Inter-connect (DCI) Superhighway provides substantial improvements for AI infrastructure providers and institutions. This upgrade enables more efficient scaling of AI training and inference operations, while also empowering organizations like universities and research laboratories to conduct complex scientific simulations by pooling computing power across Hong Kong. The new Superhighway utilizes advanced hollow core fibre, reducing latency by approximately 30% compared to traditional fibre optics.
Enhanced Capacity for AI Operations

The 3.2Tbps AI DCI Superhighway significantly boosts the capacity for AI workloads, allowing AI infrastructure providers to efficiently scale their training and inference operations. This increased bandwidth is crucial for handling the growing demands of next-generation AI applications that require seamless access to distributed computing resources across multiple data centres.
Facilitation of Complex Scientific Simulations

Academic and research institutions can leverage the upgraded Superhighway to run sophisticated scientific simulations. By enabling the pooling of computing capacity across Hong Kong, it allows these organizations to overcome the limitations of individual facilities and access the necessary computational power for advanced research endeavors.
Support for Advanced Industry Standards and Reduced Latency

This next-generation network supports critical industry standards such as Remote Direct Memory Access (RDMA), RDMA Over Converged Ethernet (RoCEv2), Multipath Reliable Connection (MRC), and Ultra-Ethernet Transport. Furthermore, its deployment on advanced hollow core fibre is designed to achieve about 30% lower latency than conventional fibre, a key factor for real-time AI processing.
Improved Efficiency and Synchronization

The combination of high bandwidth and ultra-low latency is designed to keep thousands of Graphics Processing Units (GPUs) across different data centres synchronized. This synchronization minimizes delays in data transmission, thereby reducing the time and cost associated with compute-intensive tasks. Additionally, the Superhighway aims to lower the power-per-bit, contributing to improved energy efficiency in AI operations.

Reading history

Hide →

This history is cleared when the page is closed.

What benefits does upgrading from 800Gbps to 3.2Tbps bring?

Loading

Was this article helpful?

Helpful

Not helpful

Report an issue

2

0

0

Share with friends