discrete device assignment windows server 2016

This browser is no longer supported.

Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.

Plan for Deploying Devices using Discrete Device Assignment

Applies to: Windows Server 2022, Microsoft Hyper-V Server 2016, Windows Server 2016, Microsoft Hyper-V Server 2019, Windows Server 2019

Discrete Device Assignment allows physical PCIe hardware to be directly accessible from within a virtual machine. This guide will discuss the type of devices that can use Discrete Device Assignment, host system requirements, limitations imposed on the virtual machines as well as security implications of Discrete Device Assignment.

For Discrete Device Assignment's initial release, we have focused on two device classes to be formally supported by Microsoft: Graphics Adapters and NVMe Storage devices. Other devices are likely to work and hardware vendors are able to offer statements of support for those devices. For these other devices, please reach out to those hardware vendors for support.

To learn about other methods of GPU virtualization, see Plan for GPU acceleration in Windows Server . If you are ready to try out Discrete Device Assignment, you can jump over to Deploying Graphics Devices Using Discrete Device Assignment or Deploying Storage Devices using Discrete Device Assignment to get started.

Supported Virtual Machines and Guest Operating Systems

Discrete Device Assignment is supported for Generation 1 or 2 VMs. Additionally, the guests supported include Windows 10, Windows Server 2019, Windows Server 2016, Windows Server 2012r2 with KB 3133690 applied, and various distributions of the Linux OS.

System Requirements

In addition to the System Requirements for Windows Server and the System Requirements for Hyper-V , Discrete Device Assignment requires server class hardware that is capable of granting the operating system control over configuring the PCIe fabric (Native PCI Express Control). In addition, the PCIe Root Complex has to support "Access Control Services" (ACS), which enables Hyper-V to force all PCIe traffic through the I/O MMU.

These capabilities usually aren't exposed directly in the BIOS of the server and are often hidden behind other settings. For example, the same capabilities are required for SR-IOV support and in the BIOS you may need to set "Enable SR-IOV." Please reach out to your system vendor if you are unable to identify the correct setting in your BIOS.

To help ensure hardware the hardware is capable of Discrete Device Assignment, our engineers have put together a Machine Profile Script that you can run on an Hyper-V enabled host to test if your server is correctly setup and what devices are capable of Discrete Device Assignment.

Device Requirements

Not every PCIe device can be used with Discrete Device Assignment. For example, older devices that leverage legacy (INTx) PCI Interrupts are not supported. Jake Oshin's blog posts go into more detail - however, for the consumer, running the Machine Profile Script will display which devices are capable of being used for Discrete Device Assignment.

Device manufactures can reach out to their Microsoft representative for more details.

Device Driver

As Discrete Device Assignment passes the entire PCIe device into the Guest VM, a host driver is not required to be installed prior to the device being mounted within the VM. The only requirement on the host is that the device's PCIe Location Path can be determined. The device's driver can optionally be installed if this helps in identifying the device. For example, a GPU without its device driver installed on the host may appear as a Microsoft Basic Render Device. If the device driver is installed, its manufacturer and model will likely be displayed.

Once the device is mounted inside the guest, the Manufacturer's device driver can now be installed like normal inside the guest virtual machine.

Virtual Machine Limitations

Due to the nature of how Discrete Device Assignment is implemented, some features of a virtual machine are restricted while a device is attached. The following features are not available:

Discrete Device Assignment passes the entire device into the VM. This means all capabilities of that device are accessible from the guest operating system. Some capabilities, like firmware updating, may adversely impact the stability of the system. As such, numerous warnings are presented to the admin when dismounting the device from the host. We highly recommend that Discrete Device Assignment is only used where the tenants of the VMs are trusted.

If the admin desires to use a device with an untrusted tenant, we have provided device manufactures with the ability to create a Device Mitigation driver that can be installed on the host. Please contact the device manufacturer for details on whether they provide a Device Mitigation Driver.

If you would like to bypass the security checks for a device that does not have a Device Mitigation Driver, you will have to pass the -Force parameter to the Dismount-VMHostAssignableDevice cmdlet. Understand that by doing so, you have changed the security profile of that system and this is only recommended during prototyping or trusted environments.

PCIe Location Path

The PCIe Location path is required to dismount and mount the device from the Host. An example location path looks like the following: "PCIROOT(20)#PCI(0300)#PCI(0000)#PCI(0800)#PCI(0000)" . The Machine Profile Script will also return the Location Path of the PCIe device.

Getting the Location Path by Using Device Manager

Some devices, especially GPUs, require additional MMIO space to be allocated to the VM for the memory of that device to be accessible. By default, each VM starts off with 128MB of low MMIO space and 512MB of high MMIO space allocated to it. However, a device might require more MMIO space, or multiple devices may be passed through such that the combined requirements exceed these values. Changing MMIO Space is straight forward and can be performed in PowerShell using the following commands:

The easiest way to determine how much MMIO space to allocate is to use the Machine Profile Script . To download and run the machine profile script, run the following commands in a PowerShell console:

For devices that are able to be assigned, the script will display the MMIO requirements of a given device like the example below:

The low MMIO space is used only by 32-bit operating systems and devices that use 32-bit addresses. In most circumstances, setting the high MMIO space of a VM will be enough since 32-bit configurations aren't very common.

When assigning MMIO space to a VM, the user needs to be sure set the MMIO space to the sum of the requested MMIO space for all desired assigned devices plus an additional buffer if there are other virtual devices that require a few MB of MMIO space. Use the default MMIO values described above as the buffer for low and high MMIO (128 MB and 512 MB, respectively).

If a user were to assign a single K520 GPU as in the example above, they must set the MMIO space of the VM to the value outputted by the machine profile script plus a buffer--176 MB + 512 MB. If a user were to assign three K520 GPUs, they must set the MMIO space to three times 176 MB plus a buffer, or 528 MB + 512 MB.

For a more in-depth look at MMIO space, see Discrete Device Assignment - GPUs on the TechCommunity blog.

Machine Profile Script

In order to simplify identifying if the server is configured correctly and what devices are available to be passed through using Discrete Device Assignment, one of our engineers put together the following PowerShell script: SurveyDDA.ps1.

Before using the script, please ensure you have the Hyper-V role installed and you run the script from a PowerShell command window that has Administrator privileges.

If the system is incorrectly configured to support Discrete Device Assignment, the tool will display an error message as to what is wrong. If the tool finds the system configured correctly, it will enumerate all the devices it can find on the PCIe Bus.

For each device it finds, the tool will display whether it is able to be used with Discrete Device Assignment. If a device is identified as being compatible with Discrete Device Assignment, the script will provide a reason. When a device is successfully identified as being compatible, the device's Location Path will be displayed. Additionally, if that device requires MMIO space , it will be displayed as well.

Submit and view feedback for

Additional resources

discrete device assignment windows server 2016

Passing through devices to Hyper-V VMs by using discrete device assignment

Doctor Scripto

July 14th, 2016 2 0

Summary : Learn how to attach a device from your Hyper-V host to your VM by using a new feature of Windows Server 2016.

Today we have a guest blogger, Rudolf Vesely, who has blogged here on previous occasions. Here is a link to his previous posts if you would like to read them.

Here is what Rudolf says about himself.

I am an Azure Solutions Architect at Rackspace in London. I believe that public cloud is the future for most organizations and this is why I specialize in public clouds (Amazon Web Services and Azure) and hybrid cloud strategy. My unique role at Rackspace is to guide our customers in their cloud strategy and ensure that their strategy is aligned with their business.

I started my career as a developer and have continued programming and scripting as a hobby. This is the reason why I am a big proponent of DevOps. I believe PowerShell is a fantastic tool that facilitates the process of joining the Dev and Ops worlds.

Contact information:

Introduction

Many new features in Windows Server 2016 (in any Technical Preview) will draw your attention, and it’s very easy to miss some of them. This is true especially when people don’t speak or blog about them too much like they do for Windows PowerShell 5.0, Storage Spaces Direct, or Storage Replica, as examples.

One feature that drew my attention is a new feature in Hyper-V called discrete device assignment. It can be very simply described as a device pass-through feature, the likes of which has existed on other hypervisors for many years.

Microsoft started with device pass-through on Hyper-V with disk pass-through (attaching a physical disk without using VHD / VHDX), but true pass-through came with single root I/O virtualization (SR-IOV) on Windows Server 2012. I recommend that you read John Howard’s excellent blog post series that describes SR-IOV and hardware and system requirements .

On Windows Server 2016, we finally get the ability to directly work with devices on the host and attach them to a child partition (guest VM) without being limited to only networking and storage. This feature was probably built for passing-through graphics processing units (GPUs) in Azure for N-series VMs (GPU-enabled VMs), but we can use it for anything else. Please keep in mind that this is at your own risk. Microsoft will not support your setups, and you may also have very hard security issues. Any device pass-through on any hypervisor opens the possibility to take down the host (for example, by triggering an error on the PCI Express bus) or worse, taking control of your host.

The last thing you need to consider is whether you have hardware to test on it. You need a modern client computer or server that has Intel Virtualization Technology for Directed I/O (VT-d) or AMD Virtualization (AMD-V). I use an industrial mini PC, which will be my new router and wireless access point (all virtualized), but you should be fine with a modern laptop. So, if you’re still reading, activate Intel VT-d in firmware of your testing computer, install the latest Technical Preview of Windows Server 2016, and start with PnpDevice.

PnpDevice module

On Windows Server 2016, thanks to the PnpDevice module, we finally have the possibility to work with hardware without directly touching Windows Management Instrumentation (WMI). The PnpDevice module will be very important for Nano Server and non-GUI servers, so I recommend that you try it.

Get-Command -Module PnpDevice

CommandType Name Version Source

----------- ---- ------- ------

Function Disable-PnpDevice 1.0.0.0 PnpDevice Function Enable-PnpDevice 1.0.0.0 PnpDevice Function Get-PnpDevice 1.0.0.0 PnpDevice Function Get-PnpDeviceProperty 1.0.0.0 PnpDevice

Let’s take a look what is attached to your system by using:

Get-PnpDevice –PresentOnly | Sort-Object –Property Class

Screenshot of devices attached to the system.

A lot of devices were returned. Now it’s a good idea to choose what should be passed. I do not have multiple GPUs to pass it, but my goal is to virtualize my router and access point, and I have two wireless adapters in mini PCI-e slots. I found that the easiest way to find them is by using vendor ID:

(Get-PnpDevice -PresentOnly).Where{ $_.InstanceId -like '*VEN_168C*' } # 168C == Qualcomm Atheros

Screenshot of result by using the vendor ID.

If you installed drivers on your host for devices that you want to pass through (I did not), you can filter according to Class Net like this (example from my Surface Book):

I will assume that you have installed the VM and chosen a device to attach to it. I installed Windows 10 version 1511 because this is the easiest method to test that dual dynamic acceleration (DDA) works. Later I will try replacing Windows with virtualized pfSense (FreeBSD appliance) and DD-WRT (Linux appliance). Especially in FreeBSD, I might have problems with drivers, and I am sure that I will have no issues with drivers in Windows 10.

Let’s get VM object and device object of my Wi-Fi adapter:

$vmName = 'VMDDA1' $instanceId = '*VEN_168C&DEV_002B*'

$vm = Get-VM -Name $vmName $dev = (Get-PnpDevice -PresentOnly).Where{ $_.InstanceId -like $instanceId }

Our first action should be to disable the device. You can check Device Manager by typing devmgmt.msc in the PowerShell console to see that the correct device was disabled.

# Make sure that your console or integrated scripting environment (ISE) runs in elevated mode Disable-PnpDevice -InstanceId $dev.InstanceId -Confirm:$false

Screenshot that shows disabled device in Device Manager.

Now we need to dismount the device from the host by using the Dismount-VmHostAssignableDevice cmdlet. To specify a location of the device, we need to get a specific property that is not presented in the device object by using Get-PnpDeviceProperty .

locationPath = (Get-PnpDeviceProperty -KeyName DEVPKEY_Device_LocationPaths -InstanceId $dev.InstanceId).Data[0] Dismount-VmHostAssignableDevice -LocationPath $locationPath -Force –Verbose

Now if you refresh the device object, you can see that something changed: Device is described as “Dismounted”:

(Get-PnpDevice -PresentOnly).Where{ $_.InstanceId -like $instanceId }

You can check available assignable devices by using Get-VMHostAssignableDevice :

Screenshot that shows available assignable devices.

The last step is to attach an assignable device to the VM by using Add-VMAssignableDevice like this:

Add-VMAssignableDevice -VM $vm -LocationPath $locationPath –Verbose

You may get error like this one:

This is because you haven’t modified the VM configuration after you created it. There are two main requirements for passing through. The first is that the VM has to have Automatic Stop Action set to Turn off the virtual machine . This will probably be the only VM on your host that has this configuration because we usually want Shut down or Save . The second requirement is memory configuration. Dynamic memory is allowed, but minimum and startup memory have to be equal. Let’s fix it, and finally attach our device.

Set-VM -VM $vm -DynamicMemory -MemoryMinimumBytes 1024MB -MemoryMaximumBytes 4096MB -MemoryStartupBytes 1024MB -AutomaticStopAction TurnOff Add-VMAssignableDevice -VM $vm -LocationPath $locationPath –Verbose

Screenshot that shows AutomaticStopAction is set to TurnOff and memory is configured.

If you spend some time playing with DDA, you can finish, for example, with multiple Wi-Fi adapters and one physical like I did (all functional):

Screenshot that shows multiple Wi-Fi adapters and one physical adapter.

Restore configuration

That was fun! Now it’s time to return devices to the host.

# Remove all devices from a single VM Remove-VMAssignableDevice -VMName VMDDA0 -Verbose

# Return all to host Get-VMHostAssignableDevice | Mount-VmHostAssignableDevice -Verbose

# Enable it in devmgmt.msc (Get-PnpDevice -PresentOnly).Where{ $_.InstanceId -match 'VEN_168C&DEV_002B' } | Enable-PnpDevice -Confirm:$false -Verbose

$vmName = 'VMDDA0' $instanceId = '*VEN_168C&DEV_002B*' $ErrorActionPreference = 'Stop' $vm = Get-VM -Name $vmName $dev = (Get-PnpDevice -PresentOnly).Where{ $_.InstanceId -like $instanceId } if (@($dev).Count -eq 1) {

Disable-PnpDevice -InstanceId $dev.InstanceId -Confirm:$false $locationPath = (Get-PnpDeviceProperty -KeyName DEVPKEY_Device_LocationPaths -InstanceId $dev.InstanceId).Data[0] Dismount-VmHostAssignableDevice -LocationPath $locationPath -Force -Verbose Set-VM -VM $vm -DynamicMemory -MemoryMinimumBytes 1024MB -MemoryMaximumBytes 4096MB -MemoryStartupBytes 1024MB -AutomaticStopAction TurnOff

# If you want to play with GPUs: # Set-VM -VM $vm -StaticMemory -MemoryStartupBytes 4096MB -AutomaticStopAction TurnOff # Set-VM -VM $vm -GuestControlledCacheTypes $true -LowMemoryMappedIoSpace 2048MB -HighMemoryMappedIoSpace 4096MB -Verbose

Add-VMAssignableDevice -VM $vm -LocationPath $locationPath -Verbose

$dev | Sort-Object -Property Class | Format-Table -AutoSize Write-Error -Message ('Number of devices: {0}' -f @($dev).Count)

Thank you for reading, and see you in the next article.

Thank you, Rudolf, for sharing your time and knowledge.

I invite you to follow me on Twitter and Facebook . If you have any questions, send email to me at [email protected] , or post your questions on the Official Scripting Guys Forum . Also check out my Microsoft Operations Management Suite Blog . See you tomorrow. Until then, peace.

Ed Wilson , Microsoft Scripting Guy

Doctor Scripto Scripter, PowerShell, vbScript, BAT, CMD

Comments are closed. Login to edit/delete your existing comments

In the mount sub-section, $locationPath variable is missing $ sign.

I had some problems and it took me hours to find a solution. So I want to share it. Dismounting the device from the host worked once, but I forgot to remove my VM from a cluster… So I had to undo my actions and start again and ended in “ The required virtualization driver (pcip.sys) failed to load. ” Anyway here ist the solution: You have to delete the device from the registry yourself. 1. “ psexec -s -i regedit.exe ” 2. You find your device as “Dismounted” under “ HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Enum\PCIP ” e.g. “PCI Express Graphics Processing Unit – Dismounted” 3. Delete the whole key with subkeys under PCIP. 4. Reboot and you can unmount it again from the host 🙂

NEWSLETTER SIGN-UP

Using Discrete Device Assignment in Windows Server 2016

John Savill | Apr 02, 2016

Q. What is discrete device assignment in 2016 Hyper-V?

A. Windows Server 2012 introduced support for SR-IOV which enables virtual functions (can be thought of as virtual network adapters on the physical network adapter card) from compatible network adapters that connect to a system via PCI Express to be mapped directly to Hyper-V VMs, completely bypassing the Hyper-V virtual switch thus providing very low latency connectivity. Windows Server 2016 introduces Discrete Device Assignment (DDA) that enables PCI Express connected devices to be connected directly through to virtual machines. This can be useful in certain scenarios such as types of storage (especially NVMe devices) and even graphical adapters that may not typically work with applications via RemoteFX because of unrecognized drivers. When using DDA that hardware is passed through directly to the VM and therefore the hardware's IHV drivers are used instead of generic virtual device drivers. To use DDA the system must support Access Control Services that enables the secure pass-through of a PCI Express device. Additionally the host must support SLAT (a base requirement for Hyper-V 2016) and interrupt remapping (Intel VT-d2 or AMD I/O MMU).

Note that when using DDA the devices cannot also be used in the host or shared between VMs. The devices are directly passed through to a VM. This also means live migration of the VM would not be possible since the VM is directly bound to hardware in a specific host. This means this is different that SR-IOV which does not block live migration of the VM since behind the scenes a NIC team is used to switch between the SR-IOV NIC and a vmNIC during migration.

The host must be running Windows Server 2016 however the guest OS can be Windows Server 2016, Windows Server 2012 R2, Windows 10 (1511 and above) and certain Linux distributions.

This is the technology that will be leveraged to enable the N series VMs in Azure which have GPU capabilities.

More information about text formats

IT systems management and monitoring

freshidea - Fotolia

How to deploy graphics devices using Hyper-V DDA

Admins can map a vm to a gpu with hyper-v, but they must first complete a few steps, such as prepare the vm and dismount the pcie device from the host partition..

Brien Posey

Windows Server 2016 now includes the ability to perform Discrete Device Assignment in its feature set, a feature which enables IT administrators to map VMs directly to a Peripheral Component Interconnect Express device and gives those VMs exclusive access to the device.

There are three main steps in performing a Discrete Device Assignment of a GPU: Preparing the VM, dismounting the Peripheral Component Interconnect Express, or PCIe, device from the host partition and making the device assignment within the VM. Admins must perform all three of these steps within PowerShell.

Unfortunately, this process can have some ambiguity. Some hardware vendors have additional requirements that go beyond those of Hyper-V. As a result, admins should get in touch with their GPU vendor and determine if the GPU works with Hyper-V Discrete Device Assignment (DDA) and if it must meet any special requirements before deployment.

Prepare the VM

Generally, the only step admins must complete to prepare a VM for a Hyper-V DDA is to set the VM's automatic stop action by selecting Turn off the virtual machine in the Hyper-V Manager platform. You can use a PowerShell command for performing the same action:

Screenshot of Automatic Stop Action options

The GPU vendor might require admins to perform some additional preparatory steps on the VM. These steps can include enabling a CPU's write combining feature, configuring the 32-bit memory-mapped I/O space or configuring MMIO space greater than 32 bits. You can use additional PowerShell commands to perform these steps:

Dismount the PCIe device from the host partition

Admins must take two steps to dismount a GPU's PCIe device from the host partition. First, admins must ask the GPU vendor if the GPU uses a security mitigation driver. If the GPU does, they must install it before removing the GPU from the host partition. The dismount command varies depending on whether a security mitigation driver exists.

Next, admins must find the GPU's location path by opening the Device Manager in the parent partition, right clicking on the GPU and selecting the Properties command from the shortcut menu. When the GPU properties sheet appears, select the Details tab and then select Location paths from the Property drop-down menu. Make note of the path that references PCI Root.

Screenshot of Microsoft Basic Display Adapter

If admins use a security mitigation driver, they can use this command to dismount the PCIe device:

Otherwise, they should use this command:

Assign the GPU to the VM

The last step in performing a Hyper-V DDA is to map the GPU to the VM. From the host partition, admins must enter this command:

Afterward, admins can boot the VM and install the GPU vendor's video drivers into the VM.

Dig Deeper on IT systems management and monitoring

Assign GPUs to virtual machines with VMware vGPU mode

VMware vSphere Bitfusion accelerates ML and AI workloads

How to create snapshots for Azure VMs and managed disks

Running GPU passthrough for a virtual desktop with Hyper-V

The new API is faster and cheaper than the previous ChatGPT interface, and users can opt out of submitting their data to it, ...

Though the end of Heroku's free-tier services has left many developers and software teams in the cold, a flock of cloud platform ...

Teams that can write clear and detailed defect reports will increase software quality and reduce the time needed to fix bugs. ...

REST may be a somewhat non-negotiable standard in web API development, but has it fostered overreliance? Learn why other design ...

Developers can use Microsoft Azure Logic Apps to build, deploy and connect scalable cloud-based workflows. Learn how it measures ...

Those unable to make the jump to microservices still need a way to improve architectural reliability. Here are five ways software...

Azure management groups, subscriptions, resource groups and resources are not mutually exclusive. Businesses can -- and often do ...

Amazon CodeGuru reviews code and suggests improvements to users looking to make their code more efficient as well as optimize ...

Establishing sound multi-cloud governance practices can mitigate challenges and enforce security. Review best practices and tools...

Many organizations struggle to manage their vast collection of AWS accounts, but Control Tower can help. The service automates ...

There are several important variables within the Amazon EKS pricing model. Dig into the numbers to ensure you deploy the service ...

AWS users face a choice when deploying Kubernetes: run it themselves on EC2 or let Amazon do the heavy lifting with EKS. See ...

Latency and lag time plague web applications that run JavaScript in the browser. Here are 11 reasons why WebAssembly has the ...

Has there ever been a better time to be a Java programmer? From new Spring releases to active JUGs, the Java platform is ...

Software developers can find good remote programming jobs, but some job offers are too good to be true. Follow these tips to spot...

Linux admins can use Cockpit to view Linux logs, monitor server performance and manage users. Use the tool to help admins manage ...

Hyperscale data centers can hold thousands of servers and process much more data than an enterprise facility. However, they can ...

Organizations that build 5G data centers may need to upgrade their infrastructure. These 5G providers offer products like virtual...

Working Hard In IT

My view on it from the trenches.

Setting up Discrete Device Assignment with a GPU

Introduction.

Let’s take a look at setting up Discrete Device Assignment with a GPU. Windows Server 2016 introduces Discrete Device Assignment (DDA). This allows a PCI Express connected device, that supports this, to be connected directly through to a virtual machine.

The idea behind this is to gain extra performance. In our case we’ll use one of the four display adapters in our NVIDIA GROD K1 to assign to a VM via DDA. The 3 other can remain for use with RemoteFX. Perhaps we could even leverage DDA for GPUs that do not support RemoteFX to be used directly by a VM, we’ll see.

As we directly assign the hardware to VM we need to install the drivers for that hardware inside of that VM just like you need to do with real hardware.

I refer you to the starting blog of a series on DDA in Windows 2016:

Discrete Device Assignment — Guests and Linux

Here you can get a wealth of extra information. My experimentations with this feature relied heavily on these blogs and MSFT provide GitHub script to query a host for DDA capable devices. That was very educational in regards to finding out the PowerShell we needed to get DDA to work! Please see A 1st look at Discrete Device Assignment in Hyper-V to see the output of this script and how we identified that our NVIDIA GRID K1 card was a DDA capable candidate.

Requirements

There are some conditions the host system needs to meet to even be able to use DDA. The host needs to support Access Control services which enables pass through of PCI Express devices in a secure manner. The host also need to support SLAT and Intel VT-d2 or AMD I/O MMU. This is dependent on UEFI, which is not a big issue. All my W2K12R2 cluster nodes & member servers run UEFI already anyway. All in all, these requirements are covered by modern hardware. The hardware you buy today for Windows Server 2012 R2 meets those requirements when you buy decent enterprise grade hardware such as the DELL PowerEdge R730 series. That’s the model I had available to test with. Nothing about these requirements is shocking or unusual.

A PCI express device that is used for DDA cannot be used by the host in any way. You’ll see we actually dismount it form the host. It also cannot be shared amongst VMs. It’s used exclusively by the VM it’s assigned to. As you can imagine this is not a scenario for live migration and VM mobility. This is a major difference between DDA and SR-IOV or virtual fibre channel where live migration is supported in very creative, different ways. Now I’m not saying Microsoft will never be able to combine DDA with live migration, but to the best of my knowledge it’s not available today.

The host requirements are also listed here: https://technet.microsoft.com/en-us/library/mt608570.aspx

The processor must have either Intel’s Extended Page Table (EPT) or AMD’s Nested Page Table (NPT).

The chipset must have:

The firmware tables must expose the I/O MMU to the Windows hypervisor. Note that this feature might be turned off in the UEFI or BIOS. For instructions, see the hardware documentation or contact your hardware manufacturer.

You get this technology both on premises with Windows Server 2016 as and with virtual machines running Windows Server 2016; Windows 10 (1511 or higher) and Linux distros that support it. It’s also an offering on high end Azure VMs (IAAS). It supports both Generation 1 and generation 2 virtual machines. All be it that generation 2 is X64 bit only, this might be important for certain client VMs. We’ve dumped 32 bit Operating systems over decade ago so to me this is a non-issue.

For this article I used a DELL PowerEdge R730, a NVIIA GRID K1 GPU. Windows Server 2016 TPv4 with CU of March 2016 and Windows 10 Insider Build 14295.

Microsoft supports 2 devices at the moment:

NVMe (Non-Volatile Memory express) SSD controllers

Other devices might work but you’re dependent on the hardware vendor for support. Maybe that’s OK for you, maybe it’s not.

Below I describe the steps to get DDA working. There’s also a rough video out on my Vimeo channel: Discrete Device Assignment with a GPU in Windows 2016 TPv4 .

Preparing a Hyper-V host with a GPU for Discrete Device Assignment

First of all, you need a Windows Server 2016 Host running Hyper-V. It needs to meet the hardware specifications discussed above, boot form EUFI with VT-d enabled and you need a PCI Express GPU to work with that can be used for discrete device assignment.

It pays to get the most recent GPU driver installed and for our NVIDIA GRID K1 which was 362.13 at the time of writing.

On the host when your installation of the GPU and drivers is OK you’ll see 4 NIVIDIA GRID K1 Display Adapters in device manager.

We create a generation 2 VM for this demo. In case you recuperate a VM that already has a RemoteFX adapter in use, remove it. You want a VM that only has a Microsoft Hyper-V Video Adapter.

In Hyper-V manager I also exclude the NVDIA GRID K1 GPU I’ll configure for DDA from being used by RemoteFX. In this show case that we’ll use the first one.

OK, we’re all set to start with our DDA setup for an NVIDIA GRID K1 GPU!

Assign the PCI Express GPU to the VM

Prepping the gpu and host.

As stated above to have a GPU assigned to a VM we must make sure that the host no longer has use of it. We do this by dismounting the display adapter which renders it unavailable to the host. Once that is don we can assign that device to a VM.

Let’s walk through this. Tip: run PoSh or the ISE as an administrator.

We run Get-VMHostAssignableDevice. This return nothing as no devices yet have been made available for DDA.

I now want to find my display adapters

#Grab all the GPUs in the Hyper-V Host

$MyDisplays = Get-PnpDevice | Where-Object {$_.Class -eq “Display”}

$MyDisplays | ft -AutoSize

This returns

As you can see it list all adapters. Let’s limit this to the NVIDIA ones alone.

#We can get all NVIDIA cards in the host by querying for the nvlddmkm

#service which is a NVIDIA kernel mode driver

$MyNVIDIA = Get-PnpDevice | Where-Object {$_.Class -eq “Display”} |

Where-Object {$_.Service -eq “nvlddmkm”}

$MyNVIDIA | ft -AutoSize

If you have multiple type of NVIDIA cared you might also want to filter those out based on the friendly name. In our case with only one GPU this doesn’t filter anything. What we really want to do is excluded any display adapter that has already been dismounted. For that we use the -PresentOnly parameter.

#We actually only need the NVIDIA GRID K1 cards, let’s filter some #more,there might be other NVDIA GPUs.We might already have dismounted #some of those GPU before. For this exercise we want to work with the #ones that are mountedt he paramater -PresentOnly will do just that.

$MyNVidiaGRIDK1 = Get-PnpDevice -PresentOnly| Where-Object {$_.Class -eq “Display”} |

Where-Object {$_.Service -eq “nvlddmkm”} |

Where-Object {$_.FriendlyName -eq “NVIDIA Grid K1”}

$MyNVidiaGRIDK1 | ft -AutoSize

Extra info: When you have already used one of the display adapters for DDA (Status “UnKnown”). Like in the screenshot below.

We can filter out any already unmounted device by using the -PresentOnly parameter. As we could have more NVIDIA adaptors in the host, potentially different models, we’ll filter that out with the FriendlyName so we only get the NVIDIA GRID K1 display adapters.

In the example above you see 3 display adapters as 1 of the 4 on the GPU is already dismounted. The “Unkown” one isn’t returned anymore.

Anyway, when we run

We get an array with the display adapters relevant to us. I’ll use the first (which I excluded form use with RemoteFX). In a zero based array this means I disable that display adapter as follows:

Disable-PnpDevice -InstanceId $MyNVidiaGRIDK1[0].InstanceId -Confirm:$false

When you now run

Again you’ll see

The disabled adapter has error as a status. This is the one we will dismount so that the host no longer has access to it. The array is zero based we grab the data about that display adapter.

#Grab the data (multi string value) for the display adapater

$DataOfGPUToDDismount = Get-PnpDeviceProperty DEVPKEY_Device_LocationPaths -InstanceId $MyNVidiaGRIDK1[0].InstanceId

$DataOfGPUToDDismount | ft -AutoSize

We grab the location path out of that data (it’s the first value, zero based, in the multi string value).

#Grab the location path out of the data (it’s the first value, zero based)

#How do I know: read the MSFT blogs and read the script by MSFT I mentioned earlier.

$locationpath = ($DataOfGPUToDDismount).data[0]

$locationpath | ft -AutoSize

This locationpath is what we need to dismount the display adapter.

#Use this location path to dismount the display adapter

Dismount-VmHostAssignableDevice -locationpath $locationpath -force

Once you dismount a display adapter it becomes available for DDA. When we now run

As you can see the dismounted display adapter is no longer present in display adapters when filtering with -presentonly. It’s also gone in device manager.

Yes, it’s gone in device manager. There’s only 3 NVIDIA GRID K1 adaptors left. Do note that the device is unmounted and as such unavailable to the host but it is still functional and can be assigned to a VM.That device is still fully functional. The remaining NVIDIA GRID K1 adapters can still be used with RemoteFX for VMs.

It’s not “lost” however. When we adapt our query to find the system devices that have dismounted I the Friendly name we can still get to it (needed to restore the GPU to the host when needed). This means that -PresentOnly for system has a different outcome depending on the class. It’s no longer available in the display class, but it is in the system class.

And we can also see it in System devices node in Device Manager where is labeled as “PCI Express Graphics Processing Unit – Dismounted”.

We now run Get-VMHostAssignableDevice again see that our dismounted adapter has become available to be assigned via DDA.

This means we are ready to assign the display adapter exclusively to our Windows 10 VM.

Assigning a GPU to a VM via DDA

You need to shut down the VM

Change the automatic stop action for the VM to “turn off”

This is mandatory our you can’t assign hardware via DDA. It will throw an error if you forget this.

I also set my VM configuration as described in https://blogs.technet.microsoft.com/virtualization/2015/11/23/discrete-device-assignment-gpus/

I give it up to 4GB of memory as that’s what this NVIDIA model seems to support. According to the blog the GPUs work better (or only work) if you set -GuestControlledCacheTypes to true.

“GPUs tend to work a lot faster if the processor can run in a mode where bits in video memory can be held in the processor’s cache for a while before they are written to memory, waiting for other writes to the same memory. This is called “write-combining.” In general, this isn’t enabled in Hyper-V VMs. If you want your GPU to work, you’ll probably need to enable it”

#Let’s set the memory resources on our generation 2 VM for the GPU

Set-VM RFX-WIN10ENT -GuestControlledCacheTypes $True -LowMemoryMappedIoSpace 2000MB -HighMemoryMappedIoSpace 4000MB

You can query these values with Get-VM RFX-WIN10ENT | fl *

We now assign the display adapter to the VM using that same $locationpath

Add-VMAssignableDevice -LocationPath $locationpath -VMName RFX-WIN10ENT

Boot the VM, login and go to device manager.

We now need to install the device driver for our NVIDIA GRID K1 GPU, basically the one we used on the host.

Once that’s done we can see our NVIDIA GRID K1 in the guest VM. Cool!

You’ll need a restart of the VM in relation to the hardware change. And the result after all that hard work is very nice graphical experience compared to RemoteFX

What you don’t believe it’s using an NVIDIA GPU inside of a VM? Open up perfmon in the guest VM and add counters, you’ll find the NVIDIA GPU one and see you have a GRID K1 in there.

Start some GP intensive process and see those counters rise.

Remove a GPU from the VM & return it to the host.

When you no longer need a GPU for DDA to a VM you can reverse the process to remove it from the VM and return it to the host.

Shut down the VM guest OS that’s currently using the NVIDIA GPU graphics adapter.

In an elevated PowerShell prompt or ISE we grab the locationpath for the dismounted display adapter as follows

$DisMountedDevice = Get-PnpDevice -PresentOnly |

Where-Object {$_.Class -eq “System” -AND $_.FriendlyName -like “PCI Express Graphics Processing Unit – Dismounted”}

$DisMountedDevice | ft -AutoSize

We only have one GPU that’s dismounted so that’s easy. When there are more display adapters unmounted this can be a bit more confusing. Some documentation might be in order to make sure you use the correct one.

We then grab the locationpath for this device, which is at location 0 as is an array with one entry (zero based). So in this case we could even leave out the index.

$LocationPathOfDismountedDA = ($DisMountedDevice[0] | Get-PnpDeviceProperty DEVPKEY_Device_LocationPaths).data[0]

$LocationPathOfDismountedDA

Using that locationpath we remove the DDA GPU from the VM

#Remove the display adapter from the VM.

Remove-VMAssignableDevice -LocationPath $LocationPathOfDismountedDA -VMName RFX-WIN10ENT

We now mount the display adapter on the host using that same locationpath

#Mount the display adapter again.

Mount-VmHostAssignableDevice -locationpath $LocationPathOfDismountedDA

We grab the display adapter that’s now back as disabled under device manager of in an “error” status in the display class of the pnpdevices.

#It will now show up in our query for -presentonly NVIDIA GRIDK1 display adapters

#It status will be “Error” (not “Unknown”)

We grab that first entry to enable the display adapter (or do it in device manager)

#Enable the display adapater

Enable-PnpDevice -InstanceId $MyNVidiaGRIDK1[0].InstanceId -Confirm:$false

The GPU is now back and available to the host. When your run you Get-VMHostAssignableDevice it won’t return this display adapter anymore.

We’ve enabled the display adapter and it’s ready for use by the host or RemoteFX again. Finally, we set the memory resources & configuration for the VM back to its defaults before I start it again (PS: these defaults are what the values are on standard VM without ever having any DDA GPU installed. That’s where I got ‘m)

#Let’s set the memory resources on our VM for the GPU to the defaults

Set-VM RFX-WIN10ENT -GuestControlledCacheTypes $False -LowMemoryMappedIoSpace 256MB -HighMemoryMappedIoSpace 512MB

Now tell me all this wasn’t pure fun!

Nope, you cannot do that. It acts as hardware, both in the posivitive way (stellar peformance for certain use cases) and in the negative way (you lose some capabilities you’ve become used to with virtualization). Now do note that this is TPv4 or a v1 implementation. We’ll see where this lands in the future. DDA is only for selecte use cases & needs whee the benefits outweigh the drawback and as it breaks through the virtualization layer as such it is also only for trusted admin scenarios.

Haha, yes, understand. But suppose you add a NMVe this way and then reboot the host while heavy IO is going on… “power Off” -> Really ??? 🙂 Even it’s real HW, you don’t need to turn off or cut power to a real HW system either… Same goes for SRIOV actually, so just sounds like it’s still in beta-testingstage for that matter… Put differently: DD is totally useless if Power Off will be your only choice @RTM…

I would not call that totally useless 🙂 A desktop is not totally useless because it can’t save state when you have a brown out. And you also manage a desktop, for planned events you shut it down. The use case determines what’s most immportant.

Shutdown wasn’t an option. Byebye CAU in VDI environment… Are would you go shutdown each VM manually ? I guess it will get better by the time it RTMs. I reckon MS understands that aswell…

Depends on use case. Ideally it comes without any restrictions. Keep the feedback coming. MSFT reads this blog every now and then 🙂 and don’t forget about uservoice https://windowsserver.uservoice.com/forums/295047-general-feedback !

So do you think some of the newer graphics cards that will “support” this type of DDA will be able to expose segments of their hardware? let’s say, an ATI FirePro s7150. It has the capability to serve ~16 users, but today, only one VM can use the entire card.

It’s early days yet and I do hope more information both from MSFT and card vendors will become available in the next months.

Pingback: The Ops Team #018 – “Bruised Banana” | The Ops Team | | php Technologies

Pingback: The Ops Team #018 - "Bruised Banana"

Pingback: Discrete Device Assignment with Storage Controllers - Working Hard In IT

I’m super close on this. I have the GPU assigned (a K620), but when I install the drivers and reboot, windows is ‘corrupt’. It won’t boot, and it’s not repairable. I can revert to snapshots pre-driver, but that’s about it. I’ve tried this in a Win 8 VM and a Win 10 VM. Both generation 2.

I have not seen that. Some issues with Fast Ring Windows 10 releases such as driver issues / errors but not corruption.

I think my issues is due to my video card. I’m testing this with a K620. I’m unclear if the K620 supports Access Control Services. I’m curious, your instructions use the -force flag on the Dismount-VmHostAssignableDevice command. Was the -force required with your GRID card as well? That card would absolutely support Access Control Services, I’m wondering if the -force was included for the card you were using, or for the PCI Express architecture. Thanks again for the article, I’m scrambling to find a card that supports Access Control Services to test further. I’m using the 620 because it does not require 6-pin power (My other Quadro cards do).

Hi, I’m still trying to get the details from MSFT/NVIDIA but without the force it doens’t work but throws an error. You can always try that. It’s very unclear what’s exactly supported and what not and I’ve heard (nor read) contradicting statements by the vendors involved. Early days yet.

The error is: The operation failed. The manufacturer of this device has not supplied any directives for securing this device while exposing it to a virtual machine. The device should only be exposed to trusted virtual machines. This device is not supported when passed through to a virtual machine.

Hi, I’m wondering if you have any information or experience with using DDA combined with windows server remoteApps technology. I have set up a generation 2 Windows 10 VM with a NVIDIA Grid K2 assigned to it. Remote desktop sessions work perfectly, however my remoteApp sessions occasionally freeze with a dwm.exe appcrash. I’m wondering if this could be something caused by the discrete device assignment? Are remoteApps stable with DDA?

I’m also used a PowerEdge R730 and a Tesla K80, Everything goes fine following your guide by the letter, until installing the driver on the VM, where I get a Code 12 error “This device cannot find enough free resources that it can use. (Code 12)” in Device Manager (Problem Status : {Conflicting Address Range} The specified address range conflicts with the address space.)

Any ideas what might be causing this, the driver is the latest version, and installed on the host without problems.

Hi, i have kinda same problem, same error msg but on IBM x3650 M3 with a gtx970 (thing its gtx960 works well..) u fixed it in any way? thx in advice =))

Same here with the R730 and a Tesla K80. Just finished up the above including installing Cuda and I get the same Code 12 error. Anyone figure out how to clear this error up?

i have the same problem with a HP DL 380p Gen.8:

I had the Problem in the HOST System too, there i had to enable the “PCI Express 64-BIT BAR Support” in the Bios. Then die Card works in the HOST System.

But not in the VM.

Nice read. I’ve been looking around for articles about using pass through with non-quadro cards, but haven’t been able to find much. Yours is actually the best I’ve read. By that I mean two nvidia retail geforce series cards, one for the host one for a pass through to a VM. From what I’ve read I don’t see anything to stop it working, so long as the guest card is 700 series or above, since the earlier cards don’t have support. Is that a fair assesment?

Hi. I have an error when Dismount-VmHostAssignableDevice: “The current configuration does not allow for OS control of the PCI Express bus. Please check your BIOS or UEFI settings.” What check in BIOS? Or maybe try Uninstall in Device manager?

Hello, did you found solution to this problem? I have same problem on my HP Z600 running Windows Server 2016.

I assigned a K2 GPU to a VM but now I am not able to boot the VM anymore…

I will get an error that a device is assigned to the VM and therefore it cannot be booted.

Cluster of 3x Dell R720 on Windows Server 2016, VM is limited to a single Node which has the K2 card (the other two node don’t have graphics cards yet).

Sure you didn’t assing it to 2 VMs by mistake? If both are shut down you can do it …

It looks like it just won’t work when the VM is marked as high available. When I remove this flag and store it on a local hdd of a cluster node it works.

Do you know if HP m710x with Intel P580 support DDA?

No I don’t. I’ve only use the the NVidia GRID/Tesla series so far. Ask your HP rep?

Tried to add a MVIDIA TESLA M10 (GRID) card (4xGPU) to 4 different VMs. It worked flawlessly but after that I could not get back all the GPUs when I tried to remove it from the VM. After Remove-VMAssignableDevice the GPU disappeared from the VM Device manager but I could not mount it back at the host. When listing it shows the “System PCI Express Graphics Processing Unit – Dismounted” line with the “Unknown” status. GPU disappeared from the VM but cannot be mounted and enabled as of your instructions. GPU disappeared? What could possibly caused this?

I have not run into that issue yet. Sorry.

This is great work and amazing. I have tried with NVIDIA Quadro K2200 and able to use OpenGL for apps I need.

One thing I noticed, Desktop Display is attached to Microsoft Hyper V Video and dxdiag shows it as primary video adapter. I am struggling to find a way if HYper V Video could be disabled and VM can be forced to use NVIDIA instead for all Display processing as primary video adapter. Thoughts?

Well, my personal take on this it’s not removable and it will function as it does on high end systems with an on board GPU and a specialty GPU. It used the high power one only when needed to save energy – always good, very much so on laptops. But that’s maybe not a primary concern. If your app is not being served by the GPU you want it to be serviced by you can try and dive into the settings in the control panel / software of the card, NVIDIA allows for this. Look if this helps you achieve what you need.

My vm is far from stable with a gpu through dda. (Msi r9 285 Gaming 2Gb). Yes it does work, performance is great, sometimes the vm locks up and gives a gpu driver issue. I dont get errors that get logged, just reboots or black/blue screens. Sometimes the vm crashes and comes online during the connection time of a remote connection. (Uptime reset).

I dont know if it is a problem with Ryzen. (1600x) 16Gb gigabyte ab350 gaming 3.

Launching Hwinfo64 within the vm those complete lockup the host and the vms. Outside the vm no problems.

Still great guide, the only one I could find.

Vendors / MSFT need to state support – working versus supported and this is new territory.

I disable ULPS, to prevent the gpu from idleing this morning. Vm did keep online for over 4 hours. But still at somepoint it goes doen. Here alle the error codes of the bluescreens -> http://imgur.com/a/vNWuf It seems like a driver issue to me.

when reading “Remove a GPU from the VM & return it to the host.” there is a typo.

Where-Object {$_.Class -eq “System” -AND $_.FriendlyName -like “PCI Express Graphics Processing Unit – Dismounted”}

the –, should be –

I got stuck when trying to return the gpu back to the main os, this helped

I see your website formats small -‘s as big ones

hmm now it doesnt, anyway, the –, should be – (guess i made a typo myself first)

ok something weird is going on..

Pingback: MCSA Windows Server 2016 Study Guide – Technology Random Blog

We are running the same setup with Dell 730 and Grid K1, all the setup as you listed works fine and the VM works with the DDA but after a day or so the grid inside the VM reports “Windows has stopped this device because it has reported problems. (Code 43)”

I have read that NVidia only support Grid k2 and above with DDA, so I am wondering if that’s the reason the driver is crashing?

We are running 22.21.13.7012 driver version

Have you seen this occur in your setup

It’s a lab setup only nowadays. The K1 is getting a bit old and there are no production installs I work with using that today. Some drivers do have know issues. Perhaps try R367(370.16) the latest update of the branch that still support K1/K2 with Windows Server 2016.

Thanks for your quick reply,

Yes it is an older card, we actually purchased this card some time ago for use with a Windows 2012 R2 RDS session host not knowing that it wouldn’t work with remotefx through session host.

We are now hoping to make use of it in server 2016 remotefx but I don’t think this works with a session host either, so are now testing out DDA.

We installed version 370.12 which seems to install driver version 22.21.13.7012 listed in Device manager.

I will test this newer version and let you know the results

Thanks again.

Did a quick check:

RemoteFX & DDA with 22.21.13.7012 works and after upgrading to 23.21.13.7016 it still does. Didn’t run it for loner naturally. I have seen error 43 with RemoteFX VM a few times, but that’s immediate and I have never found a good solution but replacing the VM with a clean one. Good luck.

Hello, you can read more on how to clean BIOS. Whether or not to include SR-IOV and what else will be needed.

Need help setting up the BIOS Motherboard “SuperMicro X10DRG-Q”, GPU nVIDIA K80

I assigned the video card TESLA k80, it is defined in the virtual machine, but when I look dxdiag I see an error

I have attached a Grid K1 card via DDA to a Windows 10 VM and it shows up fine and installs drivers OK in the VM but the VM still seems to be using the Microsoft Hyper-V adapter for video and rendering (Tested with Heaven Benchmark). GPU Z does not pick up any adapter. When I try to access the Nvidia Control panel I get a message saying “You are not currently using a display attached to an Nvidia GPU” Host is Windows Server 2016 Standard with all updates.

If anyone has any ideas that would help a lot, thanks.

Hi Everybody,

Someone can help me out here? 🙂

I have a “problem” with the VM console after implementing DDA. When installed the drivers on the HyperV host and configured DDA on the host and assigned a GPU to the VM that part works fine. After installing the drivers on the VM to install the GPU the drivers gets installed just fine. But after installing and a reboot of the VM I cannot manage the VM through the hyper-V console and the screen goes black. RDP on the VM works fine. What am I doing wrong here?

My setup is:

Server 2016 Datacenter Hyper-V HP Proliant DL380. Nividia Tesla M10 128 GB Profile: Virtualisation Optimisation.

I have tested version 4.7 NVIDIA GRID (on host and VM) and 6.2 NVIDIA Virtual GPU Software (on host and VM.

Kind regards

Does the GRID K1 need a nvidia vGPU license? I’m considering purchasing a K1 on ebay but am concern once I install in my server that the functionality of the card will be limited w/o a license. Is their licensing “honor” based? My intent is to use this card in a DDA configuration. If the functionality is limited I will likely need to return. Please advise. Thanks!

Nah, that’s an old card pre-licensing era afaik.

Thanks! Looks like I have this installed per the steps in this thread – a BIG THANK YOU! My guest VM (Win 10) sees the K1 assigned to it but is not detected on the 3D apps I’ve tried. Any thoughts on this?

I was reading some other posts on nvidia’s gridforums and the nvidia reps said to stick with the R367 release of drivers (369.49); which I’ve done on both the host and guest VM (I also tried the latest 370.x drivers). Anyway, I launch the CUDA-Z utility from the guest and no CUDA devices found. Cinebench sees the K1 but a OpenGL benchmark test results in 11fps (probably b/c it’s OpenGL and not CUDA). I also launch Blender 2.79b and 2.8 and it does not see any CUDA devices. Any thoughts on what I’m missing here?

I’m afraid no CUDA support is there with DDA.

Thanks for the reply. I did get CUDA to work by simply spinning up a new guest… must of been something corrupt with my initial VM. I also use the latest R367 drivers w/ no issue (in case anyone else is wondering).

Good to know. Depending on what you read CUDA works with passthrough or is for shared GPU. The 1st is correct it seems. Thx.

Great post, thank you for the info. My situation is similar but slightly different. I’m running a Dell PE T30 (it was a black Friday spur of the moment buy last year) that I’m running windows 10 w/Hyper-V enabled. There are two guests, another Windows 10 which is what I use routinely for my day-to-day life, and a Windows Server 2016 Essentials. This all used to be running on a PE2950 fully loaded running Hyper-V 2012 R2 core…

When moving to the T30 (more as an experiment) I was blow away at home much the little GPU on the T30 improved my windows 10 remote desktop experience. My only issue, not enough horse power. It only has two cores and I’m running BlueIris video software, file/print service and something called PlayOn for TV recording. This overwhelmed the CPU.

So this year I picked up a T130 with Windows 2016 standard with four cores and 8 threads. But, the T130 does not have a GPU, so, I purchased a video card and put it in. Fired it up, and, no GPU for the Hyper-V guests. I had to add the Remote desktop role to the 2016 server to then let Hyper-V use it, and then, yup, I needed an additional license at an additional fee, which I don’t really want to pay if I don’t have to… So my question:

– Is there an EASY way around this so I can use WS2016S as the host and the GPU for both guests but not have to buy a license? I say easy because the DDA sounds like it would meet this needs (for one guest?), but, also seems like more work than I’d prefer to embark on..

– Do I just use windows 10 as my Host and live the limitations, which sounds like the only thing I care about is the Virtualizing GPUs using RemoteFX. But I’m also confused on this since windows 10 on the T30 is using the GPU to make my remote experience better. So I know I’m missing some concept here…

Thanks for the help – Ed.

I cannot dismount my Grid K1 as per your instructions My setup is as follows

Motherboard: Supermicro X10DAi (SR-IOV enabled) Processor: Intel Xeon E2650-V3 Memory: 128GB DDR4 GPU: Nvidia Grid K1

When I try to dismount the card from the host I get the following: Dismount-VmHostAssignableDevice : The operation failed. The current configuration does not allow for OS control of the PCI Express bus. Please check your BIOS or UEFI settings. At line:1 char:1 + Dismount-VmHostAssignableDevice -force -locationpath $locationpath + ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + CategoryInfo : InvalidArgument: (:) [Dismount-VMHostAssignableDevice], Virtualizat ionException + FullyQualifiedErrorId : InvalidParameter,Microsoft.HyperV.PowerShell.Commands.DismountVMHos tAssignableDevice

Either a BIOS (EUFI) configuration issues that can be corrected or a BIOS (EUFI) that does not fully support OS control of the bus. Check the BIOS settings, but normally the vendors should specify what gear is compatible with DDA. If is is not an BIOS upgrade might introduce it. But this is a game of dotting the i’s and crossing the t’s before getting the hardware. Opportunistic lab builds with assembled gears might work but no guarantees given.

OK I now got 2 Nvidia Grid K1 cards installed in a HPDL380p server, 6 GPUs are showing healthy, 2 are showing error code 43, I have tried every variation of driver, BIOS, firmware, and am at my wits end, I know these cards are not faulty

Hi, thanks for the post. I did all the steps of the publication and I have an error when starting the VM windows10 Generation 2: I get the following message:

[Content] ‘VMWEBSCAN’ failed to start.

Virtual Pci Express Port (Instance ID 2614F85C-75E4-498F-BD71-721996234446): Failed to Power on with Error ‘A hypervisor feature is not available to the user.’.

[Expanded Information] ‘VMWEBSCAN’ failed to start. (Virtual machine ID B63B2531-4B4D-4863-8E3C-D8A36DC3E7AD)

‘VMWEBSCAN’ Virtual Pci Express Port (Instance ID 2614F85C-75E4-498F-BD71-721996234446): Failed to Power on with Error ‘A hypervisor feature is not available to the user.’ (0xC035001E). (Virtual machine ID B63B2531-4B4D-4863-8E3C-D8A36DC3E7AD)

I am using a PowerEdge R710 gen2 and Nvidia QUADRO P2000 that is supposed to support DDA.

Well. make sure you have the latest BIOS. But it is old hardware and support for DDA is very specific with models, hardware, Windows versions etc. The range of supported hardware is small. Verify everything like moel of CPU, chipset, R-IOV, VT-d/AMD-Vi, MSI/MSI-X, 64 PCI BAR, IRQ remapping. I would note even try with Windows Server 2019. That is only for the Tesla models, not even GRID is supported. Due to the age of the server and required BIOS support I’m afraid this might never work and even if it does can break at any time. Trial and error. You might get lucky but it will never be supported and it might break at every update.

Pingback: How-to install a Ubuntu 18.04 LTS guest/virtual machine (VM) running on Windows Server 2019 Hyper-V using Discrete Device Assignment (DDA) attaching a NVIDIA Tesla V100 32Gb on Dell Poweredge R740 – Blog

Any idea if this’ll work on IGPU such as intel’s UHD? Can’t find anything about it on the net

Can you add multiple gpu’s to the vm ?

As long as the OS can handle it sure, multiple GPUs, multiple NVMEs …

Need your advise. We are planning to create Hyper -V cluster based on two HP DL380 servers. Both servers will have Nvidia GPU card inside. The question is if it’s possible to create Hyper-v cluster based on those 2 nodes with most VMs with high availability and one VM on each node without it but with DDA to GPU? So, if I understand from this thread and comments correctly, I have to store VMs on shared data storage as usual, but for VMs with DDA I have to store them on local drive of the node. And I have to unflag HA for VMs with DDA. That’s all. Am I right?

Thanks in advance

You can also put them on shared storage but they cannot live migrate. the auto-stop action has to be set to shutdown. Whether you can use local storage depends on the storage array. On S2D having storage, other than the OS, outside of the virtual disks from the storage pool is not recommended. MSFT wants to improve this for DDA but when or if that will be available in vNext is unknown. Having DDA VM’s on shared storage also causes some extra work and planning if you want them to work on another node. Also see https://www.starwindsoftware.com/blog/windows-2016-makes-a-100-in-box-high-performance-vdi-solution-a-realistic-option “Now do note that the DDA devices on other hosts and as such also on other S2D clusters have other IDs and the VMs will need some intervention (removal, cleanup & adding of the GPU) to make them work. This can be prepared and automated, ready to go when needed. When you leverage NVME disks inside a VM the data on there will not be replicated this way. You’ll need to deal with that in another way if needed. Such as a replica of the NVMe in a guest and NVMe disk on a physical node in the stand by S2D cluster. This would need some careful planning and testing. It is possible to store data on a NVMe disk and later assign that disk to a VM. You can even do storage Replication between virtual machines, one does have to make sure the speed & bandwidth to do so is there. What is feasible and will materialize in real live remains to be seen as what I’m discussing here are already niche scenarios. But the beauty of testing & labs is that one can experiments a bit. Homo ludens, playing is how we learn and understand.”

Many thanks for you reply. Very useful. And what about GPU virtualization (GPU-PV)? Just as an idea to install Windows 10 VM and use GPU card on it. We’ll install CAD system on this VM and users will have access to it via RDP. Will it work fine?

Hyper-V only has RemoteFX which is disabled by default as it has some security risks being older technology. Then there DDA. GPU-PV is not available and while MSFT has plans/ is working on improvements I know no roadmap or timeline details for this.

Pingback: How to Customize Hyper-V VMs using PowerShell

Hi, I try to use DDA on my Dell T30 with an i7-6700k build in. Unfortunately I get the error when I try to dismount my desired device. Andy idea? Is the system not able to use DDA?

Dismount-VMHostAssignableDevice : The operation failed. The current configuration does not allow for OS control of the PCI Express bus. Please check your BIOS or UEFI settings. At line:1 char:1 + Dismount-VMHostAssignableDevice -LocationPath “PCIROOT(0)#PCI(1400)#U … + ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + CategoryInfo : InvalidArgument: (:) [Dismount-VMHostAssignableDevice], VirtualizationException + FullyQualifiedErrorId : InvalidParameter,Microsoft.HyperV.PowerShell.Commands.DismountVMHostAssignableDevice

Kind regards Jens

I have only used DDA with certified OEM solutions. It is impossible for me to find out what combinations of MobBo/BIOS/GPU cards will work and are supported.

Dell T30 with an i7-6700k and enabled all virtual things in BIOS.

When I try to dismount the card from the host I get the following: Dismount-VmHostAssignableDevice : The operation failed. The current configuration does not allow for OS control of the PCI

Did someone get this running with an Dell T30?

Privacy Preference Center

Consent management.

Cookie Settings

The WorkingHardInIT blog is a non commercial blog where technical information is shared with the global community. We only store the minimal data need for the shortest amount of time to be able to run the website and let you interact with it. We never share and/or sell any personal or general information about this website to anyone.

All Cookies

Cookies are used minimally where needed, which you can turn off at any time by modifying your internet browser’s settings.

Lenovo Press

Solutions & Software

Introduction to Windows Server 2016 Hyper-V Discrete Device Assignment

Planning / implementation.

David Tanaka

Form Number

This paper describes the steps on how to enable Discrete Device Assignment (also known as PCI Passthrough) available as part of the Hyper-V role in Microsoft Windows Server 2016.

Discrete Device Assignment is a performance enhancement that allows a specific physical PCI device to be directly controlled by a guest VM running on the Hyper-V instance. Specifically, this new feature aims to deliver a certain type of PCI device class, such as Graphics Processing Units (GPU) or Non-Volatile Memory express (NVMe) devices, to a Windows Server 2016 virtual machine, where the VM will be given full and direct access to the physical PCIe device.

In this paper we describe how to enable and use this feature on Lenovo servers using Windows Server 2016 Technical Preview 4 (TP4). We provide the step-by-step instructions on how to make an NVIDIA GPU available to a Hyper-V virtual machine.

This paper is aimed at IT specialists and IT managers wanting to understand more about the new features of Windows Server 2016 and is part of a series of technical papers on the new operating system.

Introduction Installing the GPU and creating a VM Enabling the device inside the VM Restoring the device to the host system Summary

Change History

Changes in the April 18 update:

Corrections to step 3 and step 4 in "Restoring the device to the host system" on page 12.

Configure and Buy

Full change history.

IMAGES

Setting up Discrete Device Assignment with a GPU
Discrete Device Assignment -- Description and background
Using Discrete Device Assignment in Windows Server 2016
Pass-through wired or wireless (Wi-Fi) NIC to VM using Hyper-V Discrete Device Assignment (DDA
Remote Desktop Services 2019 generally available with Windows Server 2019
Windows Server 2016: Hyper-V Discrete Device Assignment

VIDEO

MTH202 ASSIGNMENT 2
We Aced The Assignment
070-410 Part 2: Installing and Configuring Windows Server 2012 DataCenter
Data Deduplication In window Server Part 1
Digital Census Device Connectivity
NPTEL DISCRETE MATHEMATICS WEEK 1 ASSIGNMENT 1 ANSWERS

COMMENTS

Deploy graphics devices using Discrete Device Assignment
Starting with Windows Server 2016, you can use Discrete Device Assignment, or DDA, to pass an entire PCIe Device into a VM. This will allow high performance access to devices like NVMe storage or Graphics Cards from within a VM while being able to leverage the devices native drivers.
Plan for deploying devices using Discrete Device Assignment
Discrete Device Assignment allows physical PCIe hardware to be directly accessible from within a virtual machine. This guide will discuss the type of devices that can use Discrete Device Assignment, host system requirements, limitations imposed on the virtual machines as well as security implications of Discrete Device Assignment.
Discrete Device Assignment -- Description and background
With Windows Server 2016, we're introducing a new feature, called Discrete Device Assignment, in Hyper-V. Users can now take some of the PCI Express devices in their systems and pass them through directly to a guest VM. This is actually much of the same technology that we've used for SR-IOV networking in the past.
Passing through devices to Hyper-V VMs by using discrete ...
You can check available assignable devices by using Get-VMHostAssignableDevice: The last step is to attach an assignable device to the VM by using Add-VMAssignableDevice like this: Add-VMAssignableDevice -VM $vm -LocationPath $locationPath – Verbose You may get error like this one:
Using Discrete Device Assignment in Windows Server 2016 - IT Pro
Windows Server 2016 introduces Discrete Device Assignment (DDA) that enables PCI Express connected devices to be connected directly through to virtual machines.
How to deploy graphics devices using Hyper-V DDA | TechTarget
Windows Server 2016 now includes the ability to perform Discrete Device Assignment in its feature set, a feature which enables IT administrators to map VMs directly to a Peripheral Component Interconnect Express device and gives those VMs exclusive access to the device.
Setting up Discrete Device Assignment with a GPU - Working ...
Windows Server 2016 introduces Discrete Device Assignment (DDA). This allows a PCI Express connected device, that supports this, to be connected directly through to a virtual machine. The idea behind this is to gain extra performance. In our case we’ll use one of the four display adapters in our NVIDIA GROD K1 to assign to a VM via DDA.
Introduction to Windows Server 2016 Hyper-V Discrete Device ...
This paper describes the steps on how to enable Discrete Device Assignment (also known as PCI Passthrough) available as part of the Hyper-V role in Microsoft Windows Server 2016. Discrete Device Assignment is a performance enhancement that allows a specific physical PCI device to be directly controlled by a guest VM running on the Hyper-V instance.
Discrete Device Assignment -- GPUs - Microsoft Community Hub
In Hyper-V within Server 2016 TP4, there are two types of VMs, Generation 1 and Generation 2. Generation 1 is intended to run older 32-bit operating systems and 64-bit operating systems which depend on the VM having a structure very like a PC. Generation 2 is intended for 64-bit operating systems which don’t depend on a PC architecture.