Vision AI specialized company SeeLab (CEO Wooyoung Lee) announced on the 11th that it held a technology seminar on the topic of ‘AI infrastructure optimization’ for AST Global, a company specializing in AI infrastructure construction.
This seminar was designed to share ways to efficiently utilize and manage AI infrastructure resources, focusing on the AstraGo solution that improves GPU resource operation efficiency, and to introduce actual introduction cases and application effects.
AstraGo, introduced by SeeLab, is a GPU infrastructure management solution that maximizes GPU utilization during AI model learning and inference processes. It has automatic distribution, real-time monitoring, job scheduling, and distributed learning support functions essential for AI project operation, and it helps multiple users utilize optimized GPU resources simultaneously by linking with Kubernetes-based workloads. This can increase the productivity of AI developers and significantly reduce GPU usage costs.
In particular, by applying GPU partitioning (virtualization) technology, multiple workloads can be performed simultaneously on a single GPU, allowing companies to secure higher throughput with the same resources and reduce TCO (Total Cost of Ownership).
In addition, AstraGo supports perfect integration with major IT server environments such as HPE OneView, allowing integrated management of software and hardware as a single platform. It simplifies the entire operation and maintenance process, including server status monitoring, log analysis, resource optimization, and failure prediction, thereby simultaneously enhancing the stability and efficiency of AI infrastructure management. Through the latest updated RedFish integration service, a wide range of hardware can be monitored and controlled, and maintenance costs can be reduced and operational visibility increased based on detailed report indicators.
In this technical seminar, organized in collaboration with AST Global, participants experienced the powerful features of AstraGo firsthand and saw how to automate GPU resource management and apply Kubernetes-based distributed learning, realizing the potential for shortening project schedules and reducing costs.
Starting with this seminar, Cielab plans to hold regular technology sharing sessions to support domestic and foreign companies and organizations to effectively introduce and operate AI infrastructure.
Mr. Eugene Suh, manager of Cielab, said, “Based on our core strength of GPU optimization, we will actively target industries that require high-performance computation such as semiconductors, bio, and finance,” and “Through AstraGo, we will expand **PoC (Proof of Concept)** with global IT companies, domestic large corporations, and research institutes to increase our market share.”
- See more related articles
You must be logged in to post a comment.