Skip to content

Overview

Product Introduction

1 Product Overview

GPU bare metal is a high-performance computing resource specifically designed for scenarios such as deep learning, machine learning, and high-performance computing (HPC). Bare metal servers offer dedicated performance of physical machines, making them suitable for environments that demand exceptional computational capabilities and resource isolation. The platform supports multiple resource types and provides a unified resource management interface, enabling users to efficiently create, manage, and monitor computing resources.

2 Product Advantages

  • High Performance: Bare metal servers deliver the exclusive performance of physical machines, ideal for high-performance computing and large-scale data processing.

  • Strong Flexibility: Virtual machines support on-demand resource allocation, enabling rapid creation, destruction, and scaling to meet elastic business needs.

  • Unified Management: The platform integrates resources from multiple vendors and offers a unified console, simplifying the resource management process.

  • Support for Multiple Scenarios: Supports a variety of application scenarios, including AI training, scientific computing, and big data analysis.

3 Application Scenarios

  • AI Training and Inference: Bare metal servers provide powerful GPU computing capabilities, ideal for training and inference of deep learning models.

  • High-Performance Computing: Bare metal servers are suitable for scientific computing, simulation, and other scenarios that require extreme computational performance.

  • Elastic Business Deployment: Virtual machines are well-suited for web applications, microservices architectures, and other businesses that require rapid scaling and flexible deployment.

  • Big Data Processing: Both bare metal and virtual machines can be utilized for big data analysis and data mining, catering to various computational demands across different scales.

4 Usage Limitations

  • Bare Metal Servers:

  • Only support annual and monthly billing models; pay-as-you-go billing is not available.

  • Dynamic configuration adjustments (such as CPU, memory, storage, etc.) are not supported.

  • Certain features (such as custom images and VNC remote connection) may be subject to restrictions.