GPU partitioning (GPU-P) is not supported on a Windows 10 or Windows 11 client host, even though Hyper-V itself is available on Pro and Enterprise editions. Microsoft documents GPU-P for Windows Server 2025 or later running on server-class hardware with a supported partition-capable GPU, current vendor drivers, SR-IOV, and VT-d/IOMMU enabled. A physical GPU is divided into partitions that can be assigned to virtual machines (VMs), increasing VM density compared with dedicating an entire GPU through Discrete Device Assignment (DDA).
The practical path is therefore: deploy Windows Server 2025, verify a supported GPU and platform firmware, install matching host and guest drivers, confirm that Windows reports the device as partitionable, choose one of the GPU’s valid partition counts, and assign a partition to a Generation 2 VM. The exact partition profile, licensing, migration behavior, and workload performance remain dependent on the GPU vendor and server configuration.
As an Amazon Associate I earn from qualifying purchases.
What GPU-P does—and what it does not do
GPU-P lets multiple VMs share one physical GPU. Each VM receives a hardware-defined portion of the GPU’s resources rather than owning the whole device. This can support more GPU-enabled VMs per host, but it does not guarantee that every VM receives the same performance as a bare-metal workload. The GPU model, partition profile, driver, license, and workload all matter.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
DDA uses a different resource model: one or more complete physical GPUs are assigned to one VM. A physical GPU cannot be used for GPU-P and DDA at the same time. Choose GPU-P when sharing and VM density are priorities; choose DDA when a workload needs a dedicated device and fits DDA’s compatibility and operational model.
#1 Best Overall
- AI Performance: 767 AI TOPS
- OC mode: 2632 MHz (OC mode)/ 2602 MHz (Default mode)
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Axial-tech fan design features a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- A 2.5-slot design maximizes compatibility and cooling efficiency for superior performance in small chassis
Host requirements: the first compatibility gate
Use Windows Server 2025 or later
Microsoft’s current GPU-P guidance requires Windows Server 2025 or a later Windows Server release as the host. Windows 10 and Windows 11 Pro or Enterprise can run the Hyper-V role, but that fact only establishes client Hyper-V availability; it does not establish GPU-P support. Microsoft’s GPU-P troubleshooting guidance specifically excludes Windows 10/11 Pro client hosts and desktop-class hardware.
Use server-class hardware and the right firmware features
The server must expose the virtualization capabilities required by the GPU and Hyper-V platform. In BIOS or UEFI, enable CPU virtualization, SR-IOV, and the platform’s IOMMU implementation (Intel VT-d or AMD-Vi). Reboot after changing firmware settings. For clustered deployments, processors must support IOMMU DMA bit tracking. Microsoft gives AMD EPYC 7003 and later (Milan) and 4th Gen Intel Xeon Scalable processors (Sapphire Rapids) as examples for live migration. AMD EPYC 7002 (Rome) supports GPU partitioning but not live migration with GPU partitioning.
GPU-P is not a way to turn a desktop graphics card into a supported server accelerator. Confirm the complete server, slot, power, firmware, GPU, and driver combination with the server OEM and GPU vendor before buying hardware.
Recommended Free Tools
Supported GPUs and software dependencies
Microsoft’s GPU-P overview lists these supported models:
Rank #2
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5070 Ti
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
- NVIDIA A2
- NVIDIA A10
- NVIDIA A16
- NVIDIA A40
- NVIDIA L2
- NVIDIA L4
- NVIDIA L40
- NVIDIA L40S
- NVIDIA RTX Pro 6000 Blackwell Server Edition
- AMD Radeon PRO V710
This is a starting list, not a guarantee that every card works in every server or with every driver, license, partition profile, or application. The vendor’s supported driver branch and licensing terms can determine whether the device appears as available for GPU-P. Install the vendor-provided GPU-P driver on every host. Install the corresponding guest driver inside each VM as well.
For NVIDIA GPU-P live migration, Microsoft specifies the driver included in NVIDIA vGPU Software v18.x or later and a license that supports GPU-P. Other vendors can have different driver and license requirements, so validate those details before deployment rather than assuming a generic display driver is sufficient.
Guest operating-system requirements
Host support and guest support are separate checks. Microsoft’s GPU-P troubleshooting guidance lists Generation 2 VMs running one of the following guest systems:
- Windows 10 or later
- Windows 10 Enterprise multi-session or later
- Windows Server 2019 or later
- Ubuntu 18.04, 20.04, or 22.04 LTS
Create the VM as Generation 2, install the current vendor GPU driver in the guest, and verify that the intended application sees the virtual GPU. A supported guest image cannot compensate for an unsupported host, GPU, or firmware configuration.
Rank #3
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5060
- Integrated with 8GB GDDR7 128bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
Prepare the host and VM: a practical workflow
- Confirm the platform. Check that the host is Windows Server 2025 or later, the machine is server-class, and the planned GPU appears on the current vendor and Microsoft support lists. Do not proceed with a Windows client host for GPU-P.
- Install Hyper-V. Add the Hyper-V role using Server Manager or PowerShell, then restart when prompted. GPU-P requires the Hyper-V role, but installing that role alone does not make an unsupported client system eligible.
- Configure BIOS/UEFI. Enable CPU virtualization, SR-IOV, and VT-d/IOMMU. On a cluster, verify the settings on every node. A firmware change normally requires a restart before Windows can enumerate the GPU as partitionable.
- Install the physical GPU and host driver. Use the GPU vendor’s server driver intended for GPU-P. In a cluster, install the same GPU make, model, capacity, firmware family, and compatible driver branch on every server. Microsoft does not support heterogeneous GPU configurations for GPU-P clustering.
- Create a Generation 2 VM. Install a listed Windows or Ubuntu guest release, apply updates, and install the vendor’s guest GPU driver. Save application data before changing the VM’s GPU assignment.
- Check partitionability. Run PowerShell as Administrator and enter
Get-VMHostPartitionableGpu | FL Name,ValidPartitionCounts. The command should return the GPU and the partition counts that the hardware exposes. If it returns nothing, investigate firmware, driver, licensing, hardware, and host-version issues before attempting assignment. - Check Windows Admin Center, if used. The Windows Admin Center GPUs extension must be version 2.8.0 or later for the documented assignment workflow. Continue only when the GPU status is Partitioned. A status of Ready for DDA assignment describes DDA readiness, not GPU-P readiness.
- Choose a valid partition count. Select one of the counts returned by the GPU. The OEM defines the permitted profiles; you cannot invent an arbitrary number of partitions or change the resource ratio independently of the hardware profile.
- Assign the partition. Use the Windows Admin Center GPU workflow or the documented Hyper-V PowerShell commands, always from an elevated session. The current assignment guidance supports one GPU partition per VM and requires the VM and partition to be on the same host. Assignment can require the VM to be shut down and restarted, so schedule downtime.
- Validate the result. In the guest, confirm that the vendor device and driver load without warnings, then test the actual application. Check host and guest event logs, licensing status, and resource behavior before placing production workloads on the VM.
Microsoft’s assignment documentation describes automatic partition selection: the administrator assigns a valid partition profile, but does not select a particular physical slice for a particular VM. Plan capacity using the profiles exposed by the GPU rather than expecting manual placement controls.
Partition limits and capacity planning
Hardware-defined profiles
Every partition-capable GPU exposes a predefined set of valid partition counts. Those profiles determine how many VMs can share the device and how much of the GPU each receives. A card that exposes two profiles may support a different density and resource split from a card of another model, even when both are on Microsoft’s supported list.
One partition per VM in the current assignment workflow
The Windows Server assignment and troubleshooting guidance currently states that only one GPU partition can be assigned to a VM. Microsoft’s broader overview explains that, where multiple partitions are assigned, each can appear as another GPU; however, the explicit one-partition assignment limit is the practical constraint for the documented Windows Server workflow. Treat any change to that behavior as version- and platform-dependent, and verify current Microsoft documentation before designing around multiple partitions in one VM.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchCluster homogeneity
For failover or migration, install the same GPU make, model, and size on each server and keep the supported partition configuration aligned. A VM that migrates to a node with a different profile may not have an equivalent device available. Validate every node before enabling automated failover.
Rank #4
- Powered by the NVIDIA Blackwell architecture and DLSS 4. System Requirements: Minimum 850W PSU with 16-pin 12V-2x6 (12VHPWR) connector required. Verify before purchasing.
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability. Compatibility: 348mm (13.7") length, 3.6 slots, 4.3 lbs. Confirm case clearance and slot spacing. GPU bracket included.
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.6-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
Live migration: supported, but not free
GPU-P live migration is available with Windows Server 2025 when the hardware, processor, driver, licensing, and cluster requirements are met. During migration, GPU-P falls back to TCP/IP with compression. That mechanism can consume additional host CPU and make migration take longer than a CPU-only VM migration.
Use matching GPU configurations across nodes, confirm IOMMU DMA bit tracking, and install the required vendor software on every server. For NVIDIA deployments, Microsoft specifies NVIDIA vGPU Software v18.x or later with a license that supports GPU-P. Test planned maintenance and failover with the real workload; a VM that boots after migration is not necessarily delivering equivalent application performance.
GPU-P versus DDA
| Decision axis | GPU-P | DDA |
|---|---|---|
| Resource model | One physical GPU is divided into partitions shared by multiple VMs. | One or more complete physical GPUs are dedicated to one VM. |
| Potential VM density | Higher, because several VMs can share a device. | Lower, because each assigned GPU serves one VM. |
| Best fit | Workloads that benefit from shared acceleration and higher consolidation. | Workloads that need a dedicated GPU and the DDA compatibility or performance profile. |
| GPU APIs | Vendor capabilities such as DirectX 12, OpenGL, and CUDA may be available, subject to vendor support. | The same vendor APIs may be available, subject to the vendor’s DDA support. |
| Physical-device exclusivity | Cannot share the same physical GPU with DDA. | Cannot use the same physical GPU for GPU-P simultaneously. |
These are architectural distinctions, not performance promises. Benchmark the application on the selected GPU profile and confirm support with the vendor.
Free tools Windows power users keep installed
One-click scans. No signup required.
Troubleshooting common failures
The GPU does not appear as partitionable
- Verify the host is Windows Server 2025 or later, not Windows 10/11 Pro.
- Check that the machine is server-class and that the exact GPU model is supported.
- Confirm SR-IOV and VT-d/IOMMU are enabled in firmware and rebooted.
- Replace an incorrect, generic, or outdated host driver with the vendor’s GPU-P driver.
- Check vendor licensing; a licensing problem can prevent the device from being offered for GPU-P.
- Run
Get-VMHostPartitionableGpu | FL Name,ValidPartitionCountsfrom an elevated PowerShell session and inspect the returned counts.
Windows Admin Center shows the wrong status
Update the Windows Admin Center GPUs extension to version 2.8.0 or later. The workflow requires the GPU status to read Partitioned. Ready for DDA assignment indicates a different assignment model and is not sufficient evidence that GPU-P can be configured.
The VM cannot receive the partition
- Confirm the VM is Generation 2 and uses a supported guest operating system.
- Shut down and restart the VM if the assignment operation requires it.
- Ensure the VM and the partition are on the same host.
- Use a partition count returned by the GPU; unsupported values will not work.
- Check that another assignment method has not claimed the physical GPU.
The guest sees a device but the application fails
Install the vendor GPU driver inside the guest, not only on the host. Then check application API requirements, guest event logs, vendor licensing, and the selected partition profile. A visible device confirms enumeration, not that a particular CUDA, OpenGL, DirectX, or compute workload is supported.
Deployment checklist
- Windows Server 2025 or later host
- Server-class system with a supported GPU model and validated OEM configuration
- Hyper-V role installed
- CPU virtualization, SR-IOV, and VT-d/IOMMU enabled
- Vendor GPU-P driver and licensing installed on the host
- Generation 2 guest using a listed Windows or Ubuntu release
- Matching vendor GPU driver installed in the guest
Get-VMHostPartitionableGpureturns valid partition counts- Windows Admin Center GPUs extension 2.8.0 or later, if using that interface
- One supported partition profile selected for the VM
- Homogeneous GPU hardware and profiles across cluster nodes
- Live-migration processor, driver, licensing, and network requirements tested before production use
Bottom line for Windows client users
You can enable Hyper-V on Windows 10/11 Pro or Enterprise, but that does not enable GPU-P. For Microsoft’s documented GPU-P implementation, move the host role to Windows Server 2025 or later and use supported server hardware, GPU, firmware, drivers, and licensing. If the requirement is a dedicated GPU for one VM rather than shared capacity, evaluate DDA instead; do not mix the two assignment methods on one physical GPU.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minute
