> For the complete documentation index, see [llms.txt](https://nx.docs.scailable.net/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://nx.docs.scailable.net/ai-manager-v6.1.4/ai-manager-plugin/configure-the-plugin/runtime-selection.md).

# Runtime selection

Detect and select the AI Manager runtime used for inference on a device.

## Detect Available Runtimes

When the plugin first starts, it auto-detects the available runtimes for the server. This takes a moment; the Integrations tab shows its progress.

<figure><img src="https://2982163321-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FKGfC0TFSmuMDjbpLErO3%2Fuploads%2Fgit-blob-b9887a173d328772588ff6df8f083c272ac74441%2Fclient-ai-manager-detecting-accelerators.png?alt=media" alt="AI Manager panel showing the plugin detecting available AI accelerators on the device, with a prompt to wait and click Refresh"><figcaption><p>Detecting the available AI accelerators on the device</p></figcaption></figure>

If an accelerator is found, the plugin then installs the matching acceleration library before it can offer that runtime as an option.

<figure><img src="https://2982163321-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FKGfC0TFSmuMDjbpLErO3%2Fuploads%2Fgit-blob-ef1f3bc6b1b1012a8f434751046f80ded01b49d2%2Fclient-ai-manager-installing-acceleration-library.png?alt=media" alt="AI Manager panel showing the plugin installing the acceleration library, with a prompt to wait and click Refresh"><figcaption><p>Installing the acceleration library for the detected accelerator</p></figcaption></figure>

Click **Refresh** once these messages have had time to resolve.

## Choose a Runtime

After detection completes, if more than one AI-capable device was found, a list of the available runtimes to choose from is shown:

<figure><img src="https://2982163321-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FKGfC0TFSmuMDjbpLErO3%2Fuploads%2Fgit-blob-5faca0bd6cda2cd9a2978f7c1bd46c9136818110%2Fclient-ai-manager-runtime-selection-multiple-devices.png?alt=media" alt="AI Manager panel prompting to select a runtime, with Generic CPU and Intel buttons after multiple AI-capable devices were detected"><figcaption><p>Runtime selection: choose the acceleration library for your hardware</p></figcaption></figure>

The default "Generic CPU" runtime is usually available; depending on the hardware platform or installed [AI Accelerators](/ai-manager-v6.1.4/ai-accelerators-support/supported-ai-accelerators.md), other options also appear. Currently Intel OpenVino, NVIDIA, Hailo, MemryX and more are available. A runtime that still needs its acceleration library installed shows a tooltip saying so:

<figure><img src="https://2982163321-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FKGfC0TFSmuMDjbpLErO3%2Fuploads%2Fgit-blob-5ff61320a46c26712dcf6e27d5a9fc4a65ef10b7%2Fclient-ai-manager-runtime-install-tooltip.png?alt=media" alt="Runtime selection buttons with a tooltip on the Intel button reading Install this acceleration library on your server"><figcaption><p>Selecting a runtime installs its acceleration library if needed</p></figcaption></figure>

## Cloud-Managed Devices: Waiting for Device Parameters

For a device using a pipeline managed via the AI Manager Cloud, the plugin may first need to fetch that device's parameters from the cloud before a runtime can be selected:

<figure><img src="https://2982163321-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FKGfC0TFSmuMDjbpLErO3%2Fuploads%2Fgit-blob-13530e0a767cbcc94569d4f485feed1ecdf30cb4%2Fclient-ai-manager-waiting-for-device-parameters.png?alt=media" alt="AI Manager panel waiting for the device&#x27;s parameters from the cloud, with a link to the Intel accelerator support documentation"><figcaption><p>Waiting for the device's parameters from the cloud</p></figcaption></figure>

This panel links to the accelerator-specific setup documentation relevant to the detected hardware (in this example, [Intel Support](/ai-manager-v6.1.4/ai-accelerators-support/intel-support.md)). Click **Refresh** once the device's parameters are available:

<figure><img src="https://2982163321-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FKGfC0TFSmuMDjbpLErO3%2Fuploads%2Fgit-blob-87b1cc6c5f76ff915ef64e8aa0efb67a32724c9e%2Fclient-ai-manager-waiting-for-device-parameters-refresh.png?alt=media" alt="AI Manager panel still waiting for device parameters, with the cursor over the Refresh button"><figcaption><p>Click Refresh once the device's parameters have loaded</p></figcaption></figure>

Once a runtime is active, the **Device Overview** reports it directly:

<figure><img src="https://2982163321-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FKGfC0TFSmuMDjbpLErO3%2Fuploads%2Fgit-blob-02c0f1aea04f3e3fa7139ce220e0cd7c8a99f5ea%2Fclient-ai-manager-device-active-intel-runtime.png?alt=media" alt="AI Manager panel with Device Active on, AI Pipeline set to Demo Object Detection Model, and Device Overview reporting the active Intel runtime"><figcaption><p>Device Overview reporting the active Intel runtime</p></figcaption></figure>