NVML integration

NVML integration

Integrating NVML with New Relic provides valuable insights into the GPU utilization and performance metrics of your applications and systems, facilitating resource optimization, performance bottleneck identification, and the maintenance of overall stability and efficiency in your environment.

After setting up the NVML integration with New Relic, see your data in a dashboard right out of the box.

Set up the NVML integration

Complete the following steps to set up the NVML integration:

Install the infrastructure agent

To use the NVML integration, you need to first install the infrastructure agent on the same host. The infrastructure agent monitors the host itself, while the nvml integration extends your monitoring with data specific to your gpu clusters.

Use NRI-Flex to capture metrics

Flex comes bundled with the New Relic infrastructure agent. You need to configure NRI-Flex for nvml and create a flex configuration file. Follow these steps:

Create a file named nvml-config.yml on the path below:
- for Linux, /etc/newrelic-infra/integrations.d
- for windows, C:\Program Files\New Relic\newrelic-infra\integrations.d\
Use the below snippet to update your configuration file named nvml-config.yml
```
integrations:
  - name: nri-flex
    # interval: 30s
    config:
      name: NVMLexample
      apis:
        - name: nvml
          file: <PATH_TO_METRIC_CSV_FILE>
```

Restart the infrastructure agent

Use the instructions in our infrastructure agent docs to restart your infrastructure agent. This is a basic command that should work for most people:

bash

$sudo systemctl restart newrelic-infra.service

View your nvml metrics in New Relic

Once you've completed the setup above, you can view your metrics using our pre-built dashboard template. To access this dashboard:

Go to one.newrelic.com > + Integrations & Agents.
Click on the Dashboards tab.
In the search box, type nvml.
Select it and click Install.
To instrument the nvml quickstart and to see metrics and alerts, you can also follow our NVML quickstart page by clicking on the Install now button.
Here's an example query to check the number of devices in GPU:
```
SELECT latest(temperature_gpu) FROM nvmlSample TIMESERIES
```

What's next?

To learn more about building NRQL queries and generating dashboards, check out these docs: