Taipei, Taiwan, December 10, 2024 – The recently released U.2 NVMe hybrid flash unified storage device, EonStor GS 5000U, by Infortrend Technology, Inc., is a creation directed toward making the AI model training ultra-fast. It provides up to 50GB throughput, 1.3 million IOPS, and a latency of 0.3 ms, which is capable enough to meet all the specifications for AI workloads, as this 2U 24-bay solution enables efficient data processing and scalability for research institutes across the globe.
The alignment also expects connectivity with a capacity of 200GbE and provides seamless integration with any Lustre file system. This would allow researchers to access large datasets more easily for processing. By adding JB 3090 enclosures, it can reach a maximum throughput of more than 140GB/s while holding around 4PB. EonStor GS 5000U ensures that AI jobs can make maximum efficiency of computing resources, acknowledging the heavy read-write activity for AI training.
“EonStor GS 5000U is designed to accelerate AI workloads, speeding up model training and maximizing compute efficiency. This solution has been successfully implemented in several prominent research institutes around the world to tackle complex AI tasks, enabling breakthroughs in research and development,” said Frank Lee, Senior Director of Product Planning at Infortrend Technology.
Infortrend continues to innovate, equipping industries with tools to push the boundaries of AI.
Source: https://www.infortrend.com/in/news/20241210/527
Latest Stories:
OpenAI o1 Launches with Advanced Features and Improved Accuracy
Extreme Platform ONE: AI-Powered Networking for Unmatched Efficiency
Remote Launches AI-Driven Tool to Revolutionize Global Hiring
What is your reaction?
