How to Disable Physical Display Ports on NVIDIA GPUs Using the Display Mode Selector Tool

In this post, I’ll walk you through how to use the NVIDIA Display Mode Selector Tool to disable the physical display ports on NVIDIA GPUs. I’ll be demonstrating this on a VMware ESXi host, passing the GPU through to a virtual machine (VM) and using that VM to change the display mode of an NVIDIA RTX A5000 GPU.

What is the NVIDIA Display Mode Selector Tool?

The Display Mode Selector Tool allows you to configure the physical display output mode of NVIDIA GPUs—i.e. whether the physical display ports are enabled or disabled.

WARNING
Changing the configuration of the physical display ports, such as enabling or
disabling them, should only be performed if a particular use case absolutely requires it since this alters its fundamental behavior in the system. If modified incorrectly or used on an incompatible system, this can cause the GPU PCle board and system to be permanently unusable.

Why would you need to do this?

Some NVIDIA GPUs support both display-off and display-enabled modes. However, they must be in display-off mode to be used with NVIDIA vGPU software. If not set correctly, the GPU won’t be usable with vGPU.

From the latest vGPU documentation (v18.4), the default display modes for several GPUs are:

GPU	Mode as Supplied from the Factory
NVIDIA A40	Display-off
NVIDIA L40	Display-off
NVIDIA L40S	Display-off
NVIDIA L20	Display-off
NVIDIA L20 liquid cooled	Display-off
NVIDIA RTX 5000 Ada	Display enabled
NVIDIA RTX 6000 Ada	Display enabled
NVIDIA RTX A5000	Display enabled
NVIDIA RTX A5500	Display enabled
NVIDIA RTX A6000	Display enabled

Virtual GPU Software User Guide – NVIDIA Docs

TIP
A GPU that ships with display-off mode, like the NVIDIA A40, may have been changed to display-enabled mode. Always verify.

At the time of writing, only the GPUs listed above support the Display Mode Selector Tool. Other GPUs that support NVIDIA vGPU do not support this tool and do not require display mode switching.

Getting NVIDIA Display Mode Selector Tool

Download the NVIDIA Display Mode Selector Tool from NVIDIA’s website. You’ll need to sign up for a free NVIDIA Developer Account to access the download and accept the warnings.

The NVIDIA Display Mode Selector Tool is compatible with the following operating systems and hypervisors:
o Windows® version 10 64-bit RS5 or later
o Linux 64-bit
NVIDIA-Display-Mode-Selector-Tool-User-Guide_v08_Feb24.pdf

Server Configuration (BIOS)

If you’re using GPUs based on the NVIDIA Ampere architecture or newer, ensure the following BIOS options are enabled:

VT-D/IOMMU
SR-IOV
Memory Mapped I/O above 4 GB
Alternative Routing ID Interpretation (ARI)

Here’s how I configured the BIOS on my Dell PowerEdge R740:

Tip
These are all BIOS Settings found under Configuration.

Step 1 – Enable VT-D/IOMMU

Go to Processor Settings
Enable Virtualisation Technology

Step 2 – Enable SR-IOV

Under Integrated Devices, enable SR-IOV Global Enable
Click Apply

After clicking Apply, changes will show under Pending Value

Choose Apply and Reboot or At Next Reboot

Step 3 – Enable Memory Mapped I/O above 4GB

Under Integrated Devices, enable Memory Mapped I/O above 4 GB

TIP
Apply all changes together using Apply and Reboot to avoid multiple restarts.

Step 4 – Configure Power Policy for GPUs

Refer to your server’s documentation. For the Dell R740:

NOTE: When using systems with GPU, ensure that you install PSUs with 1100 W or higher, and set the PSU configuration to non-redundant mode.
Dell PowerEdge R740 Installation and Service Manual | Dell UK

VMware Configuration

We’ll now configure the GPU for passthrough, attach it to a VM, and set the necessary options to support large MMIO.

Step 1 – Enable GPU Passthrough

In the vSphere Client, select the host that contains the GPU
Browse to Configure -> Hardware -> PCI Devices -> All PCI Devices
Filter by Vendor Name: NVIDIA

Select the GPU and click Toggle Passthrough

The status should now read Enabled

Note
Enabling passthrough for the GPU may also enable passthrough for its audio device.

Step 2 – Create a VM and install an OS

Create a VM and install your OS (I used Ubuntu 24.04). If your GPU has large BAR memory, ensure the following:

64-bit guest OS is installed
64-bit MMIO is enabled for the VM
EFI boot must be enabled if BAR1 memory exceeds 256 MB
Secure Boot must be disabled

TIP
VMware’s blog post “Using GPUs with Virtual Machines on vSphere – Part 2: VMDirectPath I/O” is an excellent resource on MMIO.

The following table is an example on some GPUs that require large MMIO space and the amount of MMIO space that each GPU requires.

GPU	MMIO Space Required
NVIDIA A40	128 GB
NVIDIA RTX A5000	64 GB
NVIDIA RTX A5500	64 GB
NVIDIA RTX A6000	128 GB
Quadro RTX 6000 Passive	64 GB
Quadro RTX 8000 Passive	64 GB
Tesla V100 (all 16 GB variants)	32 GB
Tesla V100 (all 32 GB variants)	64 GB

To enable 64-bit MMIO and large MMIO space. Under VM Options -> Advanced -> Edit Configuration

Add the following parameters:

pciPassthru.use64bitMMIO=”TRUE”
pciPassthru.64bitMMIOSizeGB=<n>

Replace <n> with the required MMIO size for your GPU.

Step 3 – Attach the GPU to the VM

Shut down the VM
Edit the VM
Under Add New Device, add a PCI Device

Select the GPU (e.g. [RTX A5000])

After adding the GPU, note the memory reservation requirement noted at the top.
To use passthrough devices, the VM has to have a full memory reservation

You should see the GPU you selected listed under New PCI device and can click OK.

Disable the physical display ports

Now let’s use the Display Mode Selector Tool.

Step 1 – Download the Tool

Download the NVIDIA Display Mode Selector Tool from NVIDIA’s website. You’ll need to sign up for a free NVIDIA Developer Account to access the download and accept the warnings.

Copy the binary displaymodeselector to your VM. I used WinSCP to transfer it to my Ubuntu 24.04 VM.

Step 2 – Switch to Tool

sudo su -

Step 3 – Grant Execute Permissions

chmod +x /home/vxworld/displaymodeselector

Step 4 – Query the GPU Mode

Using the command below, we can check the current display mode

/home/vxworld/displaymodeselector --listgpumodes

Step 5- Set the GPU Display mode to physical_display_disabled

The commands below will set the display mode to disable for all GPUs passed to the VM. If you want to apply a change to a single GPU, read the documentation on the additional command line parameters.

WARNING
Changing the configuration of the physical display ports, such as enabling or
disabling them, should only be performed if a particular use case absolutely requires it since this alters its fundamental behaviour in the system. If modified incorrectly or used on an incompatible system, this can cause the GPU PCle board and system to be permanently unusable.

Proceed at your own risk. I accept no responsibility or liability for any damage caused.

If you are confident in what you are doing, enter the following command

/home/vxworld/displaymodeselector --gpumode physical_display_disabled

Press y if you are sure you want to contine

Press y if you want to update all GPUs

You should see, Successfully update GPU mode. If you get an error or after a reboot, you cannot access the GPU, check out the Troubleshooting Section below.

Step 6 – Query the GPU Mode

After rebooting, re-run the command:

Ensure the mode is now physical_display_disabled.

Troubleshooting

ERROR: access to physical memory denied

If you see the error ‘Operation not permitted’ – ‘ERROR: access to physical memory denied’

Make sure Secure Boot is disabled for the VM.

Go to VM Options -> Boot Options
Disable Enable UEFI secure boot

EEPROM protection function failed

If you get the error ‘ERROR: set EEPROM protection function failed’, It is likely large MMIO has not been set correctly.

Check that:

The correct MMIO size is set in VM configuration
(Refer to the Create a VM section above)
64-bit MMIO is enabled

Conclusion

Disabling the physical display ports on supported NVIDIA GPUs using the Display Mode Selector Tool is a critical step when preparing a GPU for use with NVIDIA vGPU software—particularly if the GPU defaults to display-enabled mode. While the process involves several prerequisite BIOS, VMware, and VM-level configurations, following each step carefully ensures compatibility and stability.

Keep in mind that this is a low-level change that can render your GPU or host system unusable if applied incorrectly. Always refer to NVIDIA’s official documentation and test changes in a controlled environment before deploying them in production.

By completing this process, your system should now be correctly configured to support vGPU workloads, unlocking powerful virtualised GPU capabilities for your virtual machines.

If you run into any issues or have questions, feel free to share your experience in the comments below.