View GPU Instance Monitor

You can view monitoring data for GPU-type CES instances to assess their GPU, CPU, memory, disk, and network resource usage.

​

View Monitoring Data

1. Log in to the GPU Cloud console and go to OBSERVABILITY > Monitor.
2. Select GPU Instance under Resource Type.
3. Select a Region and the target instance.
4. Select a time range.
5. View monitoring charts. GPU Instance monitoring includes the following Metrics groups:

   Group	Metrics	Description
   GPU	GPU compute utilization (%), GPU memory usage (GiB), GPU temperature (℃), GPU power consumption (W)	View the status of GPU compute, memory, temperature, and power consumption.
   Resource Monitoring	CPU utilization (%), system memory (%), load average (1m), memory page faults	View the instance CPU, system memory, load, and memory page faults.
   Disk	Disk usage (GiB), disk utilization (%), inode utilization (%), disk throughput (MiB/s)	View file system capacity, disk utilization, inode utilization, and disk read/write throughput.
   Network	Inbound network traffic (Mbps), outbound network traffic (Mbps), TCP connections	View inbound network traffic, outbound network traffic, and the number of TCP connections.

   Disk Metrics may display multiple curves by mount point or device, such as /, /boot, /boot/efi, or a specific disk device name. If some charts display No Data, confirm that the instance status, task runtime, and time range are correct.
6. To obtain the latest data, click Refresh.

When viewing metrics, pay close attention to trends in GPU compute utilization, GPU memory usage, CPU utilization, system memory, disk usage, and network traffic. When troubleshooting task issues, it is recommended to assess them together with task logs and alarm history.