Deploy Inference Workloads with NVIDIA NIM

This guide explains how to deploy a GenAI model from NVIDIA NIM as a native inference workload via the NVIDIA Run:ai UI.

An inference workload provides the setup and configuration needed to deploy your trained model for real-time or batch predictions. It includes specifications for the container image, data sets, network settings, and resource requests required to serve your models.

The inference workload is assigned to a project and is affected by the project’s quota.

To learn more about the inference workload type in NVIDIA Run:ai and determine that it is the most suitable workload type for your goals, see Workload types and features.

Before You Start

• Make sure you have created a project or have one created for you.

• Make sure Knative is properly installed by your administrator.

• To access NGC, make sure you have an NGC account with an active NGC API key. To obtain a key, go to NGC → Setup → API Keys, then generate or copy an existing key. In NVIDIA Run:ai, store the key either as a shared secret (created by an administrator) or a user credential (created under User settings).

Note

• Selecting the Inference type is enabled by default. If you do not see it in the menu, contact your administrator to enable it under General settings → Workloads → NIM models.

• Docker registry URL for inference workloads - For Knative-based inference workloads, Docker Hub credentials must be configured using https://index.docker.io/v1/ as the registry URL. Credentials configured with docker.io result in 401 Unauthorized errors for Knative-based inference workloads due to differences in how image digests are resolved during image pull. See Credentials for more details.

Workload Priority and Preemption

By default, inference workloads in NVIDIA Run:ai are assigned a Very high priority and are non-preemptible. These defaults ensures that inference workloads, which often serve real-time or latency-sensitive traffic, are guaranteed the resources they need. The default very-high priority ensures the fastest possible scheduling, and the default non-preemptible value guarantees that the workload will not be disrupted by other workloads once it starts running.

You can override the defaults by configuring priority and preemptibility. For more details, see Workload priority and preemption.

Advanced Setup Form

The Advanced setup form allows you to fine-tune your workload configuration with additional preferences, including environment details, image settings, workload priority, and data sources. This gives you greater flexibility, helping you adapt the workload to your specific requirements. After completing the initial setup you can either create the workload as is or use the dropdown next to CREATE INFERENCE and select Advanced setup.

Within the form, you have two options:

• Load from an existing setup - You can select an existing setup to populate the workload form with predefined values and customize any of the populated fields for a one-time configuration. These changes will apply only to this workload and will not modify the original setup. If needed, you can reset the configuration to the original setup at any time.

• Provide your own settings - Manually fill in the workload configuration fields. This is a one-time setup that applies only to the current workload and will not be saved for future use.

Creating a NIM Inference Workload

• To create an inference workload, go to Workload manager → Workloads.

• Click +NEW WORKLOAD and select Inference from the dropdown menu.

• Within the new form, select the cluster and project. To create a new project, click +NEW PROJECT and refer to Projects for a step-by-step guide.

• Select a template or Start from scratch to launch a new workload quickly. You can use a workload template to populate the workload advanced setup form with predefined configuration values. You can still modify the populated fields before submitting the workload. Any changes you make will apply only to current workload and will not be saved back to the original template.

• Select NVIDIA NIM from Inference type

• Enter a unique name for the workload. If the name already exists in the project, you will be requested to submit a different name.

• Click CONTINUE

Setting Up an Environment

Note

• NGC catalog is disabled by default. If unavailable, your administrator must enable it under General settings → Workloads → NGC catalog.

• To select an image from the NGC private registry, your administrator must configure it under General settings → Workloads → NGC private registry.

Initial setup

1. Set the model name by selecting a model from the dropdown list or entering the model name

2. Set how the model profile should be selected. A NIM model profile sets compatible model engines and criteria for engine selection, such as precision, latency, throughput optimization, and GPU requirements. Profiles are optimized to balance either latency or throughput, with quantized profiles (e.g., fp8) preferred to reduce memory usage and enhance performance.

   • Automatically (recommended) - NIM is designed to automatically select the most suitable profile from the list of compatible profiles based on the detected hardware. Each profile consists of different parameters that influence the selection process.

   • Manually - Enter profile name or unique hash identifier.

3. Set how to access the NGC catalog:

   • If you choose to not allow access to NGC, load the model from a local model-store. Go to Setting up data & storage.

   • Select credential

     • Source - Select Shared secret or My credentials.

     • Type - Select NGC API Key or Generic secret.

     • Credential name - Select an existing credential from the dropdown:

       • If you select My credentials, you can also create a new credential directly from the dropdown. The credential is saved under User settings → Credentials.

       • If you select Generic secret, make sure to select an existing Docker registry credential from the Credential name dropdown to pull NVIDIA NIM images from the NGC catalog.

   • Provide a key

     • Enter your API key.

     • Select an existing Docker registry credential from the Credential name dropdown to pull NVIDIA NIM images from the NGC catalog.

Note

The Docker registry credential must include the following:

• Username: $oauthtoken

• Password: <NGC API key>

• Docker registry URL: nvcr.io

1. Set whether inference serving endpoints can be accessed externally or only by internal cluster traffic:

   • External (Public access) - Available only if your administrator has configured Knative to support external access. See the Inference requirements section for more details.

   • Internal only - Endpoint is accessible only within cluster traffic.

2. Set an inference serving endpoint

   • Select HTTP or gRPC and enter the corresponding container port

   • Modify who can access the endpoint. See Accessing the inference workload for more details:

     • By default, Public is selected giving everyone within the network access to the endpoint with no authentication

     • If you select All authenticated users and service accounts, access is given to everyone within the organization’s account that can log in (to NVIDIA Run:ai or SSO).

     • For Specific group(s), enter group names as they appear in your identity provider. You must be a member of one of the groups listed to have access.

     • For Specific user(s) and service account(s), enter a valid user email or name. If you remove yourself, you will lose access.

Advanced setup

Setting Up Compute Resources

Note

• This form is available in both the initial and advanced setup forms. If you configure it during the initial setup, the values will be populated into the advanced setup. Any changes made during the advanced setup will override the initial configuration and be applied to the workload.

• GPU memory limit is disabled by default. If unavailable, your administrator must enable it under General Settings → Resources → GPU resource optimization.

Load from existing setup

1. Click the load icon. A side pane appears, displaying a list of available compute resources. Select a compute resource from the list.

2. Optionally, customize any of the compute resource's predefined fields as shown below. The changes will apply to this workload only and will not affect the selected compute resource.

3. Alternatively, click the ➕ icon in the side pane to create a new compute resource. For step-by-step instructions, see Compute resources.

Provide your own settings

Manually configure the settings below as needed. The changes will apply to this workload only.

Configure compute resources

1. Set the number of GPU devices per pod (physical GPUs).

2. Enable GPU fractioning to set the GPU memory per device using either a fraction of a GPU device’s memory (% of device) or a GPU memory unit (MB/GB):

   • Request - The minimum GPU memory allocated per device. Each pod in the workload receives at least this amount per device it uses.

   • Limit - The maximum GPU memory allocated per device. Each pod in the workload receives at most this amount of GPU memory for each device(s) the pod utilizes. This is disabled by default, to enable see the above note.

3. Set the CPU resources

   • Set CPU compute resources per pod by choosing the unit (cores or millicores):

     • Request - The minimum amount of CPU compute provisioned per pod. Each running pod receives this amount of CPU compute.

     • Limit - The maximum amount of CPU compute a pod can use. Each pod receives at most this amount of CPU compute. By default, the limit is set to Auto which means that the pod may consume up to the node's maximum available CPU compute resources.

   • Set the CPU memory per pod by selecting the unit (MB or GB):

     • Request - The minimum amount of CPU memory provisioned per pod. Each running pod receives this amount of CPU memory.

     • Limit - The maximum amount of CPU memory a pod can use. Each pod receives at most this amount of CPU memory. By default, the limit is set to Auto which means that the pod may consume up to the node's maximum available CPU memory resources.

4. Set extended resource(s)

   • Enable Increase shared memory size to allow the shared memory size available to the pod to increase from the default 64MB to the node's total available memory or the CPU memory limit, if set above.

   • Click +EXTENDED RESOURCES to add resource/quantity pairs. For more information on how to set extended resources, see the Extended resources and Quantity guides.

5. Set the minimum and maximum number of replicas to be scaled up and down to meet the changing demands of inference services:

   • If the number of minimum and maximum replicas are different, autoscaling will be triggered and you'll need to set conditions for creating a new replica. A replica will be created every time a condition is met. When a condition is no longer met after a replica was created, the replica will be automatically deleted to save resources.

   • Select one of the variables to set the conditions for creating a new replica. The variable's values will be monitored via the container's port. When you set a value, this value is the threshold at which autoscaling is triggered.

6. Set when the replicas should be automatically scaled down to zero. This allows compute resources to be freed up when the model is inactive (i.e., there are no requests being sent). Automatic scaling to zero is enabled only when the minimum number of replicas in the previous step is set to 0.

7. Set the order of priority for the node pools on which the Scheduler tries to run the workload. When a workload is created, the Scheduler will try to run it on the first node pool on the list. If the node pool doesn't have free resources, the Scheduler will move on to the next one until it finds one that is available:

   • Drag and drop them to change the order, remove unwanted ones, or reset to the default order defined in the project.

   • Click +NODE POOL to add a new node pool from the list of node pools that were defined on the cluster. To configure a new node pool and for additional information, see Node pools.

8. Select a node affinity to schedule the workload on a specific node type. If the administrator added a ‘node type (affinity)’ scheduling rule to the project/department, then this field is mandatory. Otherwise, entering a node type (affinity) is optional. Nodes must be tagged with a label that matches the node type key and value.

9. Click +TOLERATION to allow the workload to be scheduled on a node with a matching taint. Select the operator and the effect:

   • If you select Exists, the effect will be applied if the key exists on the node.

   • If you select Equals, the effect will be applied if the key and the value set match the value on the node.

Setting Up Data & Storage

Note

• Advanced setup - Data volumes is enabled by default. If unavailable, contact your administrator to enable it under General Settings → Workloads → Data volumes.

• Advanced setup - If Data volumes is not enabled, Data & storage appears as Data sources only, and no data volumes will be available.

• S3 data sources are not supported for inference workloads.

Initial setup

Select the data source that will serve as the model store. If the model is already stored on the selected data source it will be loaded from there automatically. Otherwise, it will be stored on the selected data source during the first workload deployment:

1. Click the load icon. A side pane appears, displaying a list of available data sources. Select a data source from the list.

2. Optionally, customize any of the data source's predefined fields as shown below. The changes will apply to this workload only and will not affect the selected data source:

   • Container path - Enter the container path to set the data target location.

3. Alternatively, click the ➕ icon in the side pane to create a new data source. For step-by-step instructions, see Data sources.

Advanced setup

Setting Up General Settings

Note

• The following general settings are optional.

• General settings are available via the Advanced setup form.

1. Set whether the workload may be interrupted by selecting Preemptible or Non-preemptible:

   • Non-preemptible workloads use the project's available GPU quota and will not be interrupted once they start running.

   • Preemptible workloads may be interrupted if resources are needed for higher-priority workloads.

2. Set the workload priority. Choose the appropriate priority level for the workload. Higher-priority workloads are scheduled before lower-priority ones.

3. Set the workload initialization timeout. This is the maximum amount of time the system will wait for the workload to start and become ready. If the workload does not start within this time, it will automatically fail. Enter a value between 5 seconds and 720 minutes. If you do not set a value, the default is taken from Knative’s max-revision-timeout-seconds.

4. Set the request timeout. This defines the maximum time allowed to process an end-user request. If the system does not receive a response within this time, the request will be ignored. Enter a value between 5 seconds and 10 minutes. If you do not set a value, the default is taken from Knative’s revision-timeout-seconds.

5. Set annotations(s). Kubernetes annotations are key-value pairs attached to the workload. They are used for storing additional descriptive metadata to enable documentation, monitoring and automation.

6. Set labels(s). Kubernetes labels are key-value pairs attached to the workload. They are used for categorizing to enable querying.

Completing the Workload

1. Decide if you wish to fine-tune your setup with additional preferences including environment details and data sources. If yes, click on the dropdown next to CREATE INFERENCE and select Advanced setup. Follow the advanced setup steps above.

2. Before finalizing your workload, review your configurations and make any necessary adjustments.

3. Click CREATE INFERENCE

Managing and Monitoring

After the inference workload is created, it is added to the Workloads table, where it can be managed and monitored.

Accessing the Inference Workload

Access to the inference serving API depends on how the serving endpoint access was configured when submitting the inference workload:

• If Public access is enabled, no authentication is required.

• If restricted access is configured:

  • Authentication is performed either by a user (with a username/password), user access keys or service accounts (with client credentials).

  • Authorization to access the endpoint is enforced based on user, service account or group membership.

  • Users relying on SSO should authenticate using their user access keys.

Follow the below steps to obtain a token:

1. Use the Tokens API with:

   • grantType: password for specific users or groups

   • grantType: client_credentials for service accounts and user access keys

2. Use the obtained token to make API calls to the inference serving endpoint. For example:

   #replace <serving-endpoint-url> and <model-name> (e.g. "meta-llama/Llama-3.1-8B-Instruct")
   curl <serving-endpoint-url>/v1/chat/completions \
     -H "Content-Type: application/json" \
     -H "Authorization: Bearer <your-access-token>" \
     -d '{
       "model": "<model-name>",
       "messages": [{
         "role": "user",
         "content": "Write a short poem on AI"
       }]
     }'

Using CLI

To view the available actions, see the inference workload CLI v2 reference.

Using API

To view the available actions for creating an inference workload, see the Inferences API reference.