Through the Gemini AI Console to build GPU management nodes that facilitates enterprise organizations to collaborate on cross AI projects. Make underlying construction infrastructure simplified, so that time and human resources can be focused on core algorithms to help companies to dig out better business opportunities from massive data more efficiently.
Data scientists and developers can quickly and easily open a cluster environment with a large amount of pre-loaded data and AI tools by Gemini AI console portal. With Gemini's unique GPU partitioning technology, the GPU utilization rate reaches its peak!
Gemini AI Console is an artificial intelligence management platform specially designed for collaborative sharing of multiple users, multiple teams, and multiple workloads.
Maximize GPU resources in Kubernetes. Derive business value from deep learning training and inferential predictions.
With automated features and functions, it liberates IT , architects and data scientists from the dilemma of manual scheduling management.
Pure software solutions, as long as CUDA-capable GPUs are available.
Users can use the split GPU without changing any program.
It has the ability to control the isolation and independence of container resources to ensure that the resources of individual containers are not interfered by other containers.
The user can regulate the minimum/maximum QoS. It can also flexibly and automatically increase the GPU quota, which also reduces the overhead of personnel management.
Kubernetes can execute more machine learning containers at the same time, reduce preemption, and reduce queuing time.
The total utilization can approach the sum of the individual container utilizations.
Jupyter development services created through AI Console all have built-in Jupyter to Job plug-in function, which can dispatch GPU tasks through Jupyter and browse the task log, which is helpful for debugging and tracking.
"Our requirement is to distribute GPU resources fairly and to facilitate management. After the introduction of Gemini’s system, AI artificial intelligence teaching will be simplified. Teachers and students can focus on the establishment and training of AI models without spending a lot of time on the operation of the system. It can speed up stepping into the application areas of AI artificial intelligence and machine learning."More...
Takming University of Science and Technology Computer Center