Black Artifacts in RDS Sessions with NVIDIA RTX A4000 (DDA) on Windows Server 2025

Card & Driver

  • Card: NVIDIA RTX A4000

  • Driver: 573.92

Issue Description

We have an issue with an RTX card on a Windows Server 2025 Standard Host and VM.

Problem

  • RDS session runs in Hyper-V on Windows Server

  • Hardware passthrough enabled so the card is used in the session (DDA)

  • At random moments the display generates black areas in square parts of windows

  • The application does not freeze; it is purely a graphical issue

If you minimize everything in the session using Windows + M and open it again, everything is fine again.

The issue occurs with at least all +/- 15 users and different machines in the (RDS) session and feels completely random.

Troubleshooting Already Performed

  • Fan settings changed from ambient to performance, reducing temperature from ~60°C to ~35°C

  • Tested different driver versions, including version 8

  • RDP settings reset to default with automatic graphics quality

  • GPOs adjusted on the local TS (Hyper-V VM)

System Specifications

  • Windows Server 2025 fully updated

  • Resources checked:

    • CPU: ~5%

    • Graphics: ~30%

GPO Settings on the Terminal Server

To enable GPU usage in RDS, the following local GPO settings are applied:

Path:
Computer Configuration\Administrative Templates\Windows Components\Remote Desktop Services\Remote Desktop Session Host\Remote Session Environment\

  • Use hardware graphics adapters for all Remote Desktop Services sessions → Enabled

  • Use WDDM graphics display driver for Remote Desktop Connections → Enabled

  • Configure H.264/AVC hardware encoding → Enabled

  • Prioritize H.264/AVC 444 graphics mode → Disabled

Question

Do any of you have seen this issue before and maybe have seen a fix for this problem?

Thanks in advance!

After a lot of searching I finally found someone else with a similar problem.

I reported it earlier over at the Geforce forums but no response there. https://www.nvidia.com/en-us/geforce/forums/geforce-graphics-cards/5/586495/black-artifacts-while-using-solidworks-with-remote/

We have recently assembled from scratch a new computer from off-the-shelf components to use as a workstation/server.
The goal is having two designers using various CAD software, one using it locally, the other using it through Remote Desktop. Windows server 2025 was cleanly installed fresh from an official Microsoft ISO, currently on driver version 596.36.

We installed an older RTX 4070 as graphics card we had laying around with the hopes of reusing it, and started experiencing some black artifacts during use of Solidworks.
The artifacts show up occasionally for both the local user (working locally directly from a monitor) and also the remote one (not necessarily at the same time), so its unlikely the remote connection is the source of problem.

Applications don’t crash or close, artifacts happen across the whole screen, across Windows desktop, Windows Explorer, SolidWorks viewport, toolbars, etc, and are typically constrained to specific graphics elements (not randomly shaped across the screen). It often starts by blacking out the whole screen, then progressively UI elements like icons in toolbars, menu text, tabs, and various graphic elements recover from black to proper rendering when refreshed, like on mouse hover with the cursor.

Here is a screenshot and video of the issue:

Notice how the toolbars turned black, and menus and icons gradually refresh when hovered by the cursor. Is this similar to what you experienced?

This clearly indicates some sort hardware acceleration or graphics card issue, and immediately thought the card was either defective or not compatible for not being Solidworks certified. Since we had an older NVidia Quadro P4000 that is Solidworks certified laying around I tested it in the same computer. After proper drivers installed, soon the same black artifacts were happening again.

Maybe this card is too old? We bough a brand new RTX Pro 4000 Blackwell in the hopes of fixing the problem once and for all. Again another fresh OS install, new certified drivers and voilà: Black artifacts happening all the same.
Three defective cards? No way in hell! Something else has to be wrong with the computer. Motherboard, PSU, Overheating?

In the meantime since we moved on to Solidworks 2026 and could not go back because of file version compatibility, we were forced to also upgrade the older workstation to Solidworks 2026 (and consequently the required Windows11/Server 2025). In this older workstation no artifacts occur with the exact same setup, and same Windows ISO. It has been working flawlessly for several weeks now.

In yet another third older workstation with an aging NVidia Quadro P4000, again same exact setup, fresh install of Microsoft Windows 2025, Solidworks 2026, proper clean drivers, and a soon after boom! Black artifacts as well. On another computer now?

So why did the original workstation never have issues and all others do? What is common among them? Nvidia cards! The old workstation has an AMD Radeon Pro WX 5100 and all others we tested have NVidia cards.
Is this a compatibility issue with server editions of Windows specifically? Probably not, we tested regular Windows 11 as well and black artifacts occur all the same with simultaneous users.
My only guess is this is some sort of bug or incompatibility between Windows 11/Server 2025 and remote desktop/simultaneous users.

The graphics card is technically effectively driving the equivalent to three full HD monitors at 1920x1080 (two local and one remote), but that seems well withing the capabilities of even the older Quadro card.

We’ve had a similar setup using Windows 10 like this for years without any issues whatsoever with the AMD Radeon Pro WX 5100. We have since found out the issue also occurs outside SolidWorks (in AlphaCAM for example), albeit less often.

We spent quite some money on a professional line graphics card to try and avoid it only to find it doesn’t, and because it is professional product, no returns are accepted, so now we are stuck with a ineffective solution.

I have exhausted all my knowledge about this without success. This is probably an “exotic” setup, but only now have I found any post with what appears to be a similar issue.

Have you ever found a solution? Can this be reported anywhere? I’m not even sure NVidia is aware of the issue, and there doesn’t seem to be any way to report it.

Hi Duarte,

Thank you for your detailed response. To be honest, I wasn’t actually expecting any replies after 6 months so thank you!

Sadly we’ve yet to fix this issue. I actually reinstalled the latest driver last week (with DDU) from the NVIDIA site and sadly still experience the same issue.

The issues you show are exactly the same as what we have! The only difference is the program where the issue starts. But once it starts it also spreads to multiple apps like Explorer, etc.

I was wondering, do you also have the machine virtualized and push the card through via DDA? Or is it a physical machine that you RDP into?

And have you also seen this error in the logs of the affected machine? Last week I found out that this error has been popping up a lot:


"The description for Event ID 153 from source nvlddmkm cannot be found. Either the component that raises this event is not installed on your local computer or the installation is corrupted. You can install or repair the component on the local computer.

If the event originated on another computer, the display information had to be saved with the event.

The following information is included with the event:

\Device\Video5
Error occurred on GPUID: 66100000

The message resource is present but the message was not found in the message table."


However I still don’t know what causes it or if it’s related sadly.

Greetings,

Daniel

Thanks for the response Daniel

No, its a very simple setup running on bare metal, one user sits locally at the actual computer, the other connects via remote desktop from across the desk, no virtualization is used nor DDA.

I have also tried in the meantime to disable Hardware Accelerated GPU Scheduling but it seems to have had no effect, though I didn’t make sure it is a global option, or if it has to be turned off by user.

I had several looks in the NVida control panel but no other options seem relevant to the issue at hand.

Not at the office at the moment, but I look into those error logs and report back as soon as I get a chance.

Thanks

Indeed I have, a few different variations. Lots and lots of them.

The description for Event ID 153 from source nvlddmkm cannot be found. Either the component that raises this event is not installed on your local computer or the installation is corrupted. You can install or repair the component on the local computer.

If the event originated on another computer, the display information had to be saved with the event.

The following information was included with the event:

\Device\Video13
Error occurred on GPUID: 100

The message resource is present but the message was not found in the message table

The description for Event ID 13 from source nvlddmkm cannot be found. Either the component that raises this event is not installed on your local computer or the installation is corrupted. You can install or repair the component on the local computer.

If the event originated on another computer, the display information had to be saved with the event.

The following information was included with the event:

\Device\Video13
Graphics SM Warp Exception on (GPC 1, TPC 0, SM 0): MMU NACK Errors

The message resource is present but the message was not found in the message table

The description for Event ID 13 from source nvlddmkm cannot be found. Either the component that raises this event is not installed on your local computer or the installation is corrupted. You can install or repair the component on the local computer.

If the event originated on another computer, the display information had to be saved with the event.

The following information was included with the event:

\Device\Video13
Graphics Exception: ESR 0x515730=0xb010020 0x515734=0x4 0x515728=0x1c81fb60 0x51572c=0x1174

The message resource is present but the message was not found in the message table

But I’m not even sure what they mean. Searching the web I see lots of people complaining about them usually associated with crashes in video games. Not sure if this is related to the black artifacts.

I have in the meantime reported this to the driver feedback form, and surprisingly got response today asking a few more questions. That was unexpected. Hopefully they are at least aware of the issue now. Not sure if they’ll ever provide a solution, but if they do, I’ll be sure to share here.

Hi Duarte,

Great to hear! Up until now I never really received a reply to the ticket I submitted, other than being asked to create a forum post. So this is definitely good news!

Regarding the errors, it’s actually good to hear that you’re seeing them as well (and quite frequently too). Maybe that will help us pinpoint the root cause.

For example, the Windows Server 2025 host with the RTX A4000 is the only server generating these errors. I also have two other servers for different customers running Windows Server 2022 and 2019 and neither of them has generated a single one of these events over the last 14 days. The Server 2025 machine however generates quite a lot of them, even though the configuration is identical apart from the operating system.

On the days when the company is closed (Saturday and Sunday) no errors are logged at all.

I’m not entirely sure but it almost seems like a Windows 11 kernel related issue to me. After all, Windows 10 based platforms (including Server 2019 and Server 2022) don’t appear to show the same behavior.

I’m very curious to hear what NVIDIA driver support comes back with 😁

That has also crossed my mind. We’ve had a similar setup running for years over Windows 10 and never had issues, but since it was on an AMD card we can’t really rule it our.

Both Windows Server 2025 and currently Windows 11 currently suffer from the artifacts.

In the meantime if we ever get just one single user on the computer at a time, I’ll try to see if the errors are also logged, or if they only happen with simultaneous users.

The Driver Feedback respondent also asked about what GPO policy I had configured on Windows OS, such as :

  • Use hardware graphics adapters for all Remote Desktop Services sessions → Enabled
  • Use WDDM graphics display driver for Remote Desktop Connections → Enabled
  • Configure H.264/AVC hardware encoding → Enabled
  • Prioritize H.264/AVC 444 graphics mode → Disabled

On our device they are all Not configured. Do you have any in yours?

I have since disabled Hardware Accelerated GPU Scheduling, but no improvement.

Hi Duarte,

We experienced the same issue when only the Use hardware graphics adapters for all Remote Desktop Services sessions GPO was enabled.

Then we also changed the following settings to:

  1. Use WDDM graphics display driver for Remote Desktop Connections → Enabled
  2. Configure H.264/AVC hardware encoding → Enabled
  3. Prioritize H.264/AVC 444 graphics mode → Disabled

Unfortunately changing these additional settings did not resolve the issue for us either.

@daan17 Have you checked your servers in the last few days, has it still been happening?

We got a Windows update yesterday morning, and since then all day yesterday, and all day today we saw no more black screens, even under heavy load.

I’m even scared to say it out loud, but dare I ask, was actually fixed?

Hi Duarte,

Thank you! This sounds amazing.

As of last Friday, it was still showing Error 153 in the Event Log.

Today it isn’t but maybe I should wait a few more hours.

Do you know which KB fixed the issue?

We use patch management software, so updates are delayed for 7 days before installation. But we can manually approve patches 😁

Well not quite sure. The only update I have here registered for that time window is 2026-06 Security Update (KB5094126) (26200.8655) on 2026 07 02, the actual release states 2026 06 29

In the Update Release Notes I see nothing that seems related to the issue, but I’m no expert.

Looking at the event log, I no longer see those Event ID 153 or Event ID 13 errors starting 2026 07 01 though, it is possible the update was actually installed on this day, and we only started noticing the improvements the following day after a reboot.

We are currently on Windows 11 though, so the actual update numbers may differ for Windows Server 2025.

Good morning Duarte,

Thank you for the update!

I think you’re right. I don’t see this KB on Server 2025 either and all available updates have been installed on our end.

Unfortunately, the issue still persists:

I guess I’ll have to wait until the patch is released for Server 2025 😁

Do you still have a production Server 2025 instance in use or are you no longer running one?

I’d expect the update channel for Windows Server editions to be separate from regular desktop versions, but that is just a hunch. Possibly also slower, with some delay buffer to afford more testing for more critical infrastructure.

At the moment both computers are on Windows 11, no server editions running any longer. Maybe give it a few more days, see if it finally reaches you in the meantime.

If you desperately need any testing I do have a spare computer on hold I could possibly experiment on.

Do let me know if you get it fixed in the meantime.