---
name: vpc-managing-gpus
title: Managing GPUs and accelerators for VPC
description: The GPU-enabled family of profiles provides on-demand, cost-effective access to GPUs and accelerators. GPUs and accelerators help to accelerate the processing time that is required for compute intensive workloads such as AI, machine learning, inferencing, and more. To use the GPUs and accelerators, make sure that you install the appropriate driver and associated toolkit for your workloads.
last-updated: 2026-07-31
---

> ## Documentation Index
> The table of contents for this documentation set is at https://cloud.ibm.com/docs/vpc?format=markdown
> The index for all IBM Cloud docs is at: https://cloud.ibm.com/docs/llms.txt
> Use these files to discover more information as needed.

# Managing GPUs and accelerators for VPC
{: #managing-gpus}

The GPU-enabled family of profiles provides on-demand, cost-effective access to GPUs and accelerators. GPUs and accelerators help to accelerate the processing time that is required for compute intensive workloads such as AI, machine learning, inferencing, and more. To use the GPUs and accelerators, make sure that you install the appropriate driver and associated toolkit for your workloads.
{: shortdesc}


## Configuring a virtual server instance with an NVIDIA GPU
{: #provision-gpu-on-vsi}

1. Provision a [virtual server instance](https://cloud.ibm.com/docs/vpc?topic=vpc-creating-virtual-servers&format=markdown) by choosing an NVIDIA [GPU profile](https://cloud.ibm.com/docs/vpc?topic=vpc-profiles&interface=ui&format=markdown#gpu) in the Profile field. Stock and custom operating system images are supported.
2. Install the NVIDIA GPU driver for your virtual server instance's image and GPU profile. The following table describes minimum driver and CUDA software version levels for Linux and Windows operating systems. For more information, see NVIDIA's [Download drivers](https://www.nvidia.com/en-us/drivers/){: external} page. For an overview of drivers for NVIDIA data center products, see [NVIDIA Data Center Drivers](https://docs.nvidia.com/datacenter/tesla/drivers/index.html#){: external}.

   | GPU | NVIDIA driver | CUDA version |
   |---------|---------|---------|
   | A100    | 550  | [12.4](https://developer.nvidia.com/cuda-12-4-0-download-archive?target_os=Linux){: external}  |
   | L4      | 550  | [12.4](https://developer.nvidia.com/cuda-12-4-0-download-archive?target_os=Linux){: external}  |
   | L40s    | 550  | [12.4](https://developer.nvidia.com/cuda-12-4-0-download-archive?target_os=Linux){: external} |
   | V100    | 535  | [12.2](https://developer.nvidia.com/cuda-12-2-0-download-archive?target_os=Linux){: external} |
   | H100    | 550  | [12.4](https://developer.nvidia.com/cuda-12-4-0-download-archive?target_os=Linux){: external} |
   | H200    | 570  | [12.8](https://developer.nvidia.com/cuda-12-8-0-download-archive?target_os=Linux){: external}  |
   | B300 [Select availability]{: tag-green} | 590  | [13.1](https://developer.nvidia.com/cuda-13-1-0-download-archive?target_os=Linux){: external}  |
   {: caption="NVIDIA drivers and CUDA version for Linux" caption-side="bottom"}
   {: #linux-gpu-nvidia-drivers}
   {: tab-title="Linux"}
   {: tab-group="NVIDIA GPU drivers"}
   {: class="simple-tab-table"}
   {: summary="GPUs and minimum NVIDIA drivers and CUDA versions"}

   | GPU | NVIDIA driver | CUDA version |
   |---------|---------|---------|
   | A100    | 538  | [12.2](https://developer.nvidia.com/cuda-12-2-0-download-archive?target_os=Windows){: external}  |
   | L4      | 538  | [12.2](https://developer.nvidia.com/cuda-12-2-0-download-archive?target_os=Windows){: external}  |
   | L40s    | 538  | [12.2](https://developer.nvidia.com/cuda-12-2-0-download-archive?target_os=Windows){: external}  |
   | V100    | 535  | [12.2](https://developer.nvidia.com/cuda-12-2-0-download-archive?target_os=Windows){: external} |
   | H100    | N/A  | N/A  |
   | H200    | N/A  | N/A  |
   | B300    | N/A  | N/A  |
   {: caption="NVIDIA drivers and CUDA version for Windows 2019, 2022, 2025" caption-side="bottom"}
   {: #windows-2019-2022-gpu-nvidia-drivers}
   {: tab-title="Windows 2019, 2022, 2025"}
   {: tab-group="NVIDIA GPU drivers"}
   {: class="simple-tab-table"}
   {: summary="GPUs and minimum NVIDIA drivers and CUDA versions"}

   | GPU | NVIDIA driver | CUDA version |
   |---------|---------|---------|
   | A100    | 529  | [12.0](https://developer.nvidia.com/cuda-12-0-0-download-archive?target_os=Windows){: external}  |
   | L4      | 529  | [12.0](https://developer.nvidia.com/cuda-12-0-0-download-archive?target_os=Windows){: external}  |
   | L40s    | N/A  | N/A  |
   | V100    | 535  | [12.0](https://developer.nvidia.com/cuda-12-0-0-download-archive?target_os=Windows){: external} |
   | H100    | N/A  | N/A  |
   | H200    | N/A  | N/A  |
   | B300    | N/A  | N/A  |
   {: caption="NVIDIA drivers and CUDA version for Windows 2016" caption-side="bottom"}
   {: #windows-2016-gpu-nvidia-drivers}
   {: tab-title="Windows 2016"}
   {: tab-group="NVIDIA GPU drivers"}
   {: class="simple-tab-table"}
   {: summary="GPUs and minimum NVIDIA drivers and CUDA versions"}

3. Install associated toolkit for your workload. See NVIDIA's [CUDA toolkit downloads](https://developer.nvidia.com/cuda-downloads){: external} page.

For detailed instructions to complete Steps 2 and 3, other GPU tools, and examples, see [How to Use V100-Based GPUs on IBM Cloud VPC](https://www.ibm.com/products/tutorials/how-to-use-v100-based-gpus-on-ibm-cloud-vpc){: external}.

For a Linux-focused guide on installing the NVIDIA drivers, see the [NVIDIA Driver Installation Guide](https://docs.nvidia.com/datacenter/tesla/driver-installation-guide/index.html){: external}.

If you want to automate the installation of the drivers, you can use the [User data](https://cloud.ibm.com/docs/vpc?topic=vpc-user-data&format=markdown) section of the virtual server. By using the user data field, you can input a script that issues the commands to install the NVIDIA drivers.
{: tip}

## Configuring a virtual server instance with an NVIDIA B300 GPU
{: #provision-b300-gpu-on-vsi}

NVIDIA HGX B300 accelerated virtual server profiles are available for select customers. Create a [support case](https://cloud.ibm.com/docs/support?topic=support-open-case&interface=ui&format=markdown) if you are interested in purchasing and using this offering.
{: preview}

When you use the B300 GPU profile, the guest operating system needs an update. If the guest OS is not modified, you might see errors such as, "NVRM: This PCI I/O region assigned to your NVIDIA device is invalid:"

To configure the guest OS for the B300 GPU:

1. These commands must be run as root. Sudo to root.
   ```sh
   sudo -i
   ```
   {: pre}

2. Edit the grub configuration file. The following example uses vi.
   ```sh
   vi /etc/default/grub.d/50-cloudimg-settings.cfg
   ```
   {: pre}

3. Add the following kernel parameters to the `GRUB_CMDLINE_LINUX_DEFAULT` line:
   ```text
   pci=assign-busses pci=realloc pci=nocrs pci=big_root_window
   ```
   {: pre}

   The updated line should look like this:
   ```text
   # CLOUD_IMG: This file was created/modified by the Cloud Image build process
   GRUB_CMDLINE_LINUX_DEFAULT="nofb console=ttyS0 console=tty1 pci=assign-busses pci=realloc pci=nocrs pci=big_root_window"
   ```
   {: screen}

4. Update grub to apply the changes.
   ```sh
   update-grub2
   ```
   {: pre}

5. Restart the virtual server.
   ```sh
   reboot
   ```
   {: pre}

## Configuring a virtual server instance with an Intel Gaudi 3 AI Accelerator
{: #provision-gaudi-3-on-vsi}

1. Provision a [virtual server instance](https://cloud.ibm.com/docs/vpc?topic=vpc-creating-virtual-servers&format=markdown) by choosing the Intel® Gaudi® 3 AI Accelerator [instance profile](https://cloud.ibm.com/docs/vpc?topic=vpc-accelerated-profile-family&interface=ui&format=markdown#gaudi-3-vsi-profiles) in the Profile field. Stock and custom operating system images are supported.
2. Install the Intel Gaudi 3 AI Accelerator software and drivers for your virtual server. To download the drivers, see [Intel Gaudi Driver and Software Installation](https://docs.habana.ai/en/latest/Installation_Guide/Driver_Installation.html){: external} page.

## Integrating drivers into a custom image from volume
{: #scaling-custom-image}

1. Provision a virtual server instance with a GPU and install the drivers.
2. Create an image from the virtual server instance stock image boot volume. For more information, see [Creating an image from a volume](https://cloud.ibm.com/docs/vpc?topic=vpc-create-ifv&interface=ui&format=markdown).
3. Repeat the Image from volume process to deploy across multiple instances.

## Configuring a bare metal server with an AMD Instinct MI300X Accelerator
{: #configure-bare-metal-server-with-amd-mi300x}

1. [Create a bare metal server](https://cloud.ibm.com/docs/vpc?topic=vpc-creating-bare-metal-servers&interface=ui&format=markdown) by choosing the AMD Instinct™ MI300X Accelerator [bare metal server profile](https://cloud.ibm.com/docs/vpc?topic=vpc-bare-metal-servers-profile&interface=ui&format=markdown) in the Profile field. Stock and custom operating system images are supported.
2. Install the necessary drivers for your bare metal server. To download the drivers, see [Installing ROCm and machine learning frameworks](https://rocm.docs.amd.com/en/latest/){: external} page.

## Next steps
{: #managing-gpus-next-steps}

For more information, see the [NVIDIA driver documentation](https://docs.nvidia.com/datacenter/tesla/index.html){: external}.