Checks whether the number of GPU cores for each Elastic Compute Service (ECS) instance within your Alibaba Cloud account is greater than or equal to a specified value.
Scenarios
You can use GPU-accelerated compute-optimized ECS instances to build AI applications based on frameworks such as TensorFlow. Such instances have lower computing capacity requirements for clients and are suitable for scenarios such as image processing and real-time online rendering for cloud-based gaming and AR/VR applications.
Risk level
Default risk level: low.
You can change the risk level as required when you apply this rule.
Compliance evaluation logic
- If the number of GPU cores for each ECS instance within your account is greater than or equal to a specified value, the evaluation result is compliant.
- If the number of GPU cores for an ECS instance within your account is less than a specified value, the evaluation result is non-compliant. For more information about how to correct the non-compliant configuration, see Non-compliance remediation.
Rule details
| Item | Description |
|---|---|
| Rule name | ecs-gpu-min-count-limit |
| Rule ID | ecs-gpu-min-count-limit |
| Tag | ECS and GPU |
| Automatic remediation | Not supported |
| Trigger type | Configuration change |
| Supported resource type | ECS instance |
| Input parameter | gpuCount |
Non-compliance remediation
For more information about how to upgrade an ECS instance for more GPU cores, see Overview of instance upgrade and downgrade.