This topic describes how to view the monitoring data of a GPU-accelerated compute optimized ECS instance in the CloudMonitor console and query the data by calling an API operation.
Prerequisites
- A GPU-accelerated compute optimized ECS instance is created. The required GPU driver is installed on the instance. For more information, see Create a GPU-accelerated instance that is not configured with a driver.
- The CloudMonitor agent is installed on the ECS instance. For more information, see Install and uninstall the CloudMonitor agent for C++.
- Monitoring charts that include GPU metrics are added. For more information, see Add a monitoring chart.
GPU metrics
GPU metrics can be viewed based on GPUs, instances, and application groups. The following table lists the GPU metrics.
| Metric | Unit | MetricName | Dimensions |
|---|---|---|---|
| (Agent)gpu_decoder_utilization | % | gpu_decoder_utilization | userId, instanceId, and gpuId |
| (Agent)gpu_encoder_utilization | % | gpu_encoder_utilization | userId, instanceId, and gpuId |
| (Agent)gpu_gpu_temperature | °C | gpu_gpu_temperature | userId, instanceId, and gpuId |
| (Agent)gpu_gpu_usedutilization | % | gpu_gpu_usedutilization | userId, instanceId, and gpuId |
| (Agent)gpu_memory_freespace | Byte | gpu_memory_freespace | userId, instanceId, and gpuId |
| (Agent)gpu_memory_freeutilization | % | gpu_memory_freeutilization | userId, instanceId, and gpuId |
| (Agent)gpu_memory_userdspace | Byte | gpu_memory_userdspace | userId, instanceId, and gpuId |
| (Agent)gpu_memory_usedutilization | % | gpu_memory_usedutilization | userId, instanceId, and gpuId |
| (Agent)gpu_power_readings_power_draw | W | gpu_power_readings_power_draw | userId, instanceId, and gpuId |
View GPU metric data in the CloudMonitor console
- Log on to the CloudMonitor console.
- In the left-side navigation pane, click Host Monitoring.
- On the Host Monitoring page, click the host name or click the Monitoring Charts in the Actions column of the host.
- Click the GPU Monitoring tab. On the GPUMonitor tab, view the monitoring charts for GPU metrics.
Query GPU metric data by calling an API operation
You can call the DescribeMetricList operation to query the GPU metric data of the ECS instance. For more information, see DescribeMetricList.
Note Set the
Namespace parameter to acs_ecs_dashboard and specify the MetricName and Dimensions parameters. For more information, see GPU metrics.