Workload Properties

This page is for Self-Hosted, version v3.0 (default). For other versions, use one of these documentation indexes: - v3.0 (default): https://docs.nvidia.com/run-ai/self-hosted/v3.0/llms.txt - v2.26: https://docs.nvidia.com/run-ai/self-hosted/v2.26/llms.txt - v2.25: https://docs.nvidia.com/run-ai/self-hosted/v2.25/llms.txt - v2.24: https://docs.nvidia.com/run-ai/self-hosted/v2.24/llms.txt

Workload properties define the behavioral and scheduling characteristics of a workload submitted to the NVIDIA Run:ai platform. These properties such as type, category, and priority determine how workloads are classified, monitored, and scheduled.

• Type - Defines the functional behavior of a workload, such as training or inference, and serves as the foundation for how it is classified within the platform. Each workload type includes a default category (used for monitoring) and a default priority (used for scheduling within a project).
• Category - Represents the role or purpose of a workload, such as Build, Train, Deploy. Each workload type is assigned a default category to ensure consistent classification across the platform for monitoring purposes. See Monitor workloads by category.
• Priority - Determines the scheduling order of workloads within a project and whether a workload can be preempted. Higher-priority workloads are scheduled first and may become non-preemptible, preventing interruption by lower-priority workloads. See Workload priority control.