Revised per SME suggestions.

Author: LouisBerner

articles/databox-online/azure-stack-edge-gpu-overview-gpu-virtual-machines.md

@@ -7,7 +7,7 @@ author: alkohli
 ms.service: databox
 ms.subservice: edge
 ms.topic: conceptual
-ms.date: 08/08/2024
+ms.date: 08/15/2024
 ms.author: alkohli
 #Customer intent: As an IT admin, I need to understand how to deploy and manage GPU-accelerated VM workloads on my Azure Stack Edge Pro GPU devices.
 ---
@@ -50,7 +50,7 @@ This extension supports the following OS distro, depending on the driver support
 | Red Hat Enterprise Linux | 7.4 |
 
 > [!NOTE]
-> Ubuntu version 18.04 LTS GPU extension has been deprecated. The GPU extension is no longer supported on GPU VMs running on Azure Stack Edge devices. If you plan to utilize the Ubuntu version 18.04 LTS distro, see steps for manual GPU driver installation at [CUDA Toolkit 12.1 Update 1 Downloads](https://developer.nvidia.com/cuda-12-1-1-download-archive?target_os=Linux&target_arch=x86_64&Distribution=Ubuntu&target_version=18.04&target_type=deb_local).
+> Ubuntu 18.04 LTS GPU extension has been deprecated. The GPU extension is no longer supported on Ubuntu 18.04 GPU VMs running on Azure Stack Edge devices. If you plan to utilize the Ubuntu version 18.04 LTS distro, see steps for manual GPU driver installation at [CUDA Toolkit 12.1 Update 1 Downloads](https://developer.nvidia.com/cuda-12-1-1-download-archive?target_os=Linux&target_arch=x86_64&Distribution=Ubuntu&target_version=18.04&target_type=deb_local). You may need to download the CUDA signing key before the installation. For detailed steps to install the signing key, see [Troubleshoot GPU extension issues for GPU VMs on Azure Stack Edge Pro GPU](azure-stack-edge-gpu-troubleshoot-virtual-machine-gpu-extension-installation.md#in-versions-lower-than-2205-linux-gpu-extension-installs-old-signing-keys-signature-andor-required-key-missing).
 
 ## GPU VM deployment
 
@@ -69,15 +69,15 @@ Before you deploy GPU VMs on your device, review the following considerations if
 
 - **Create a GPU VM followed by Kubernetes configuration on your device**: In this scenario, the GPU VM creation and Kubernetes configuration will both be successful. Kubernetes won't have access to the GPU in this case.
 
-- **Configure Kubernetes on your device followed by creation of a GPU VM**: In this scenario, the Kubernetes will claim the GPU on your device and the VM creation will fail as there are no GPU resources available.
+- **Configure Kubernetes on your device followed by creation of a GPU VM**: In this scenario, the Kubernetes claims the GPU on your device and the VM creation will fail as there are no GPU resources available.
 
 #### For 2-GPU device
 
 - **Create a GPU VM followed by Kubernetes configuration on your device**: In this scenario, the GPU VM that you create will claim one GPU on your device and Kubernetes configuration will also be successful and claim the remaining one GPU.	
 
-- **Create two GPU VMs followed by Kubernetes configuration on your device**: In this scenario, the two GPU VMs will claim the two GPUs on the device and the Kubernetes is configured successfully with no GPUs. 
+- **Create two GPU VMs followed by Kubernetes configuration on your device**: In this scenario, the two GPU VMs claim the two GPUs on the device and the Kubernetes is configured successfully with no GPUs. 
 
-- **Configure Kubernetes on your device followed by creation of a GPU VM**: In this scenario, the Kubernetes will claim both the GPUs on your device and the VM creation will fail as no GPU resources are available.
+- **Configure Kubernetes on your device followed by creation of a GPU VM**: In this scenario, the Kubernetes claims both the GPUs on your device and the VM creation will fail as no GPU resources are available.
 
 <!--Li indicated that this is fixed. If you have GPU VMs running on your device and Kubernetes is also configured, then anytime the VM is deallocated (when you stop or remove a VM using Stop-AzureRmVM or Remove-AzureRmVM), there is a risk that the Kubernetes cluster will claim all the GPUs available on the device. In such an instance, you will not be able to restart the GPU VMs deployed on your device or create GPU VMs. -->