Compute Overview

The Compute Platform in Matrice.ai provides users with the flexibility to create instances for various tasks such as data processing, model training, and deployment. Users have the autonomy to select instances that best fit their workload requirements. For smaller tasks like dataset management or annotation, instances with lower specifications are sufficient, whereas model training and deployment often require instances with higher capacities of RAM, VRAM, and storage.

Matrice.ai supports all major cloud computing services, including AWS, GCP, OCI, and Lambda Labs. Users can choose an instance provided by any of these cloud providers according to their preference and project needs.

Types of Compute Instances

1. On-Demand Instances

On-demand instances are publicly accessible compute resources that can be allocated to users on a first-come, first-served basis. These instances are shared among all users and are suitable for tasks that do not require dedicated resources.

2. Dedicated Instances

Dedicated instances are reserved specifically for a user and offer exclusive access. These instances ensure consistent availability and performance, making them ideal for critical tasks like model training and deployment where dedicated resources are required.

Warning: Ensure that you select the instance type based on your specific task and resource requirements. Choosing an instance that is underpowered for high-resource tasks (e.g., model training) or overpowered for smaller tasks (e.g., dataset management) can lead to inefficient use of resources and potential delays.