Feature Request
As it is done with nvdia GPUs would be nice to have the possibility to query metrics from AMD devices using the rocm-smi [0] utility.
Proposal:
Write a plugin that parses the JSON output provided by rocm-smi executable and extracts metrics to be sent via Telegraf.
Current behavior:
Currently I could not find any effort in that direction.
Desired behavior:
Use case:
Very convenient in all those scenario where ROCm [1] is deployed as computing environment for GPGPU applications.
Some pointers:
[0] https://rocmdocs.amd.com/en/latest/index.html
[1] https://rocmdocs.amd.com/en/latest/Installation_Guide/ROC-smi.html
Feature Request
As it is done with nvdia GPUs would be nice to have the possibility to query metrics from AMD devices using the
rocm-smi[0] utility.Proposal:
Write a plugin that parses the JSON output provided by
rocm-smiexecutable and extracts metrics to be sent via Telegraf.Current behavior:
Currently I could not find any effort in that direction.
Desired behavior:
Use case:
Very convenient in all those scenario where ROCm [1] is deployed as computing environment for GPGPU applications.
Some pointers:
[0] https://rocmdocs.amd.com/en/latest/index.html
[1] https://rocmdocs.amd.com/en/latest/Installation_Guide/ROC-smi.html