Live migration with GPU-P

Author: Manika Dhiman

azure-local/manage/gpu-manage-via-partitioning.md

@@ -5,7 +5,7 @@ author: alkohli
 ms.author: alkohli
 ms.topic: how-to
 ms.service: azure-local
-ms.date: 03/26/2025
+ms.date: 04/29/2025
 ---
 
 # Manage GPUs using partitioning (preview)
@@ -33,7 +33,7 @@ Consider the following limitations when using the GPU-P feature:
 
 - Partitions are autoassigned to the VMs. You can't choose a specific partition for a specific VM.
 
-- Currently, GPU partitioning on Azure Local doesn't support live migration of VMs. But VMs can be automatically restarted and placed where GPU resources are available if there's a failure.
+- GPU partitioning on Azure Local supports live migration. However, the host and VMs must be on NVIDIA virtual GPU software version 18 and later. For more details, see [Microsoft Azure Local - NVIDIA Docs](https://docs.nvidia.com/vgpu/18.0/grid-vgpu-release-notes-microsoft-azure-stack-hci/index.html).
 
 - You can partition your GPU using Azure Command-Line Interface (CLI). We recommend that you use Azure CLI to configure and assign GPU partitions. You must manually ensure that the homogeneous configuration is maintained for GPUs across all the machines in your system.
 

azure-local/whats-new.md

@@ -45,6 +45,7 @@ This release has the following features and improvements:
 
 - **Azure Local VM changes**:
     - **Data disk expansion**: With this release, you can expand the size of a data disk attached to an Azure Local VM. For more information, see [Expand the size of a data disk attached to an Azure Local VM](./manage/manage-arc-virtual-machine-resources.md).
+    - **Live VM migration with GPU partitioning (GPU-P)**: You can now live migrate VMs with GPU-P. These VMs must be on the latest NVIDIA virtual GPU v18 drivers to enable live migration with GPU-P. For more information, see [Microsoft Azure Local - NVIDIA Docs](https://docs.nvidia.com/vgpu/18.0/grid-vgpu-release-notes-microsoft-azure-stack-hci/index.html).
     - **Documentation changes**: An article describing a [Comparison of the management capabilities of VMs on Azure](./concepts/compare-vm-management-capabilities.md) was released recently.
 
 - **Update improvements**: