Summary
Thus far I have noticed this issue on three games:
Eversiege: Untold Ages
Incursion: Red River
Windrose
There are two issues:
Texture Flickering occurs on certain models under unknown conditions. When GPU load exceeds 100% this will result in a memory allocation error and a crash. From what I can tell this is exclusive to 5070-TIs and is also exclusive to the Unreal Engine.
Lumen+Global Illumination cause an odd interaction where theres green-red-yellow strobing lights from light sources. I had to set Windrose settings to low.
My initial conclusion was this was a driver or hardware issue. After thorough testing, and raising it with an NVIDIA rep, I now am quite certain its at least not a hardware issue. I’ve also submitted a ticket with Epic but I’ve heard nothing back as well and its been weeks so now this post to see if we can get some traction.
In this post I’ll also be posting a PII redacted version of the chat logs with the NVIDA reps when we were trying to debug and resolve this.
Customer (- -) via channel 'CSS Web’04/13/2026 04:53 PM
Full Specs:
MSI Gaming RTX 5070 Ti 16G Ventus 3X OC Graphics Card (16GB GDDR7, 256-bit, Extreme Performance: 2497 MHz, DisplayPort x 3 2.1a, HDMI 2.1b, NVIDIA Blackwell Architecture)
AMD Ryzen 7 7800X3D 8-Core Processor
B850 MSI MOBO Gaming Plus Wifi
CORSAIR Vengeance RGB DDR5 RAM 32GB
Windows 11 x64
Details:
I believe there is a driver issue with the current 595.97, 5070-ti and UE5. I first noticed the issue with Ever Siege when I would get massive flickering, screen tearing, and almost immediate crashing once getting into the game. As these issues seemed to be novel to me, I attempted fixing them (more on that below) especially since other 5000 series card users reported no issues, though notably (and anecdotally) no 5070 ti users that I found. I returned the game, but later got a different UE5 game, Incursion Red River, where after being in the game for about ten seconds I would experience screen flickering before crashing. This only occurs with UE5 games, as I’ve run BF6, done a maxed out Benchmark on Wukong and Cyberpunk 2077 to ensure there were no issues.
As to ensure there were no issues on my end I:
Troubleshooted all of my hardware by ensuring my GPU was not voltage throttled
Memtest86 by RAM
Increased my memory page file
Flashed my BIOS
Used DDU in safe mode to remove my GPU Drivers and then the NVIDIA APP to reinstall them in normal windows mode
Ensure Windows was up to date
Ensured Windows Update image health
Bear in mind this is on a fresh install of windows (as of Februrary this year) and these components were recently purchased (as of December last year)
Crash entry from Unreal:
Code: UE-Test_C
LoginId:b4ea824e480e27dceaf7bc86ded8c074
Unhandled Exception: EXCEPTION_ACCESS_VIOLATION reading address 0x00000207acedda28
Test_C_Win64_Shipping
Test_C_Win64_Shipping
Test_C_Win64_Shipping
Test_C_Win64_Shipping
Test_C_Win64_Shipping
Test_C_Win64_Shipping
Test_C_Win64_Shipping
Test_C_Win64_Shipping
Test_C_Win64_Shipping
Test_C_Win64_Shipping
Test_C_Win64_Shipping
Test_C_Win64_Shipping
Test_C_Win64_Shipping
kernel32
ntdll
Customer (- -) via channel 'CSS Web’04/14/2026 05:51 AM
Just realized I didn’t add steps to reproduce as well as additional information that’s important.
First:
Texture flickering leading to a cascading crash seems to be linked to GPU usage being around 100%. Reducing settings to minimum or reducing resolution, thus reducing GPU usage, will still induce texture flickering but will allow the game to run for a time before crashing. Ever Siege had incredible amounts of texture flickering whereas Incursion Red River has far less.
This was all done on a 2560x1440p 180hz monitor. I did also test on 1 920x1080p monitor and noticed similar results, however, on Ever Siege and the main proponent seems to be GPU usage. Additionally, running the game Windowed Borderless dramatically increased stability.
Also I’ve attached a GPU-Z Sensor Log of a benchmark reading of Cyberpunk 2077 running at the settings below I did last night in case you need that.
Results of Benchmark: Average FPS 115.49 Min FPS 32.64 Max FPS 138.20 Time 64.20 Number of Frames 7414 Specs: Windowed Borderless 2560x1440p DLSS Super Resolution Transformer Model NVIDIA DLAA DLAA Sharpness 1 DLSS Ray Reconstruction DLSS Frame Generation DLSS Multi Frame Generation 4x Texture Quality High Path Tracing On Crowd Density High FOV 90 Contact Shadows On Improved Facial Lighting Geometry On Anisotropy 16 Local Shadow Mesh Quality High Local Shadow Quality High Cascaded Shadow Quality High Cascaded Shadows Range High Cascaded Shadows Resolution High Distant Shadows Resolution High Volumetric Fog Resolution Ultra Volumetric Cloud Quality Ultra Max Dynamic Decals Ultra Screen Space Reflections Quality Psycho Subsurface Scattering Quality High Ambient Occlusion High Color Precision High Mirror Quality High LOD High Full computer Specs: AMD Ryzen 7 7800X3D 8-Core Processor B850 MSI MOBO Gaming Plus Wifi CORSAIR Vengeance RGB DDR5 RAM 32GB NVidia Geforce 5070-ti Windows 11 x64
Rule Response04/13/2026 04:53 PM
NVIDIA REP via channel 'Email’04/14/2026 08:13 AM
Hello,
Thank you for contacting NVIDIA Customer Care.
My name is Adnan and I will be assisting you today.
I understand you are experiencing flickering, screen tearing, and crashes in Unreal Engine 5 games with driver version 595.97 on your RTX 5070 Ti, while other games are working normally.
Based on your description, this may be related to driver compatibility with UE5 titles or specific rendering features used by those games.
This behavior is likely due to a driver-level compatibility issue with UE5 rendering pipelines (such as DX12, shader compilation, or frame generation), especially since non-UE5 titles are stable on your system.
Please try the below troubleshooting steps and confirm.
Install an earlier stable driver version (for example, roll back to a previous 59x.xx version) and check if the issue persists.
In the game settings, try switching from DX12 to DX11 (if available) and disable features like DLSS/Frame Generation temporarily.
To assist further, please help us with the following details:
Exact driver version currently installed (confirm 595.97?)
Does the issue occur in all UE5 games or only specific titles?
Are DLSS or Frame Generation enabled in these games?
Does disabling DLSS/Frame Generation change the behavior?
Are you using G-SYNC or V-SYNC?
Does the issue occur in windowed mode as well as fullscreen?
Are there any overlays enabled (NVIDIA overlay, Discord, MSI Afterburner, etc.)?
What is your PSU wattage and are you using the default GPU power cables?
Have you tried a clean driver installation again with “Perform clean install”?
Does the issue occur immediately on launch or only after gameplay begins?
Are there any temperature spikes or GPU usage abnormalities before the crash?
Please share your observations, and we will assist you further.
Best regards,
NVIDIA REP
NVIDIA Customer Care
Customer (- -) via channel 'CSS Web’04/14/2026 10:22 AM
Thank you for your prompt response,
I followed your troubleshooting steps. This had occured on driver xxx already, so I updated to the current 595.97 and tried again to no avail. Following your advice, I did a clean removal using DDU to driver 595.79 instead to test that one and experienced identical results. I also tested 581.57 on a whim and experienced identical results.
In game settings cannot use DX11 to render. Additionally, DLSS/Frame Generations and Scaling do not seem to have an impact this as I’ve toggled them on and off in Incursion Red River trying every conceivable setting, although Ever Siege does not have fine tuned settings for this. Fortunately, Incursion Red River does and it allows me to more accurately observe the effects and precisely push GPU usage up to 100% and induce the crash (at the time, see below).
Since my last addendum (which should answer your inquiries regarding whether I’m experiencing undervolting, additionally, I will be addressing all your questions below) I don’t know what I did but I’m having an excessively difficult time inducing the crashing but I’m still getting massive texture flickering. Additionally, Post Processing with Unreal (specifically lighting/illumination) is also not interacting properly on higher settings with both Ever Siege and Incursion Red River having identical results of strobing green/red/yellow lights. Also, post processing is causing massive runaway usage of VRAM and GPU capacity.
As for your questions:
Exact driver version currently installed (confirm 595.97?)
Yes.
Does the issue occur in all UE5 games or only specific titles?
Only UE5 titles, though not The Finals, though that was the only other UE5 title I checked.
Are DLSS or Frame Generation enabled in these games?
I tested Incursion Red River with DLSS with DLSS, it does not have the option for Frame Generation, both on and off and noticed no discernable difference in texture flickering.
Does disabling DLSS/Frame Generation change the behavior?
No.
Are you using G-SYNC or V-SYNC?
No. I do not have a G-Sync capable monitor and I do not use V-Sync as a rule, though I did try toggling it on and noticed no discernible difference.
Does the issue occur in windowed mode as well as fullscreen?
Yes.
Are there any overlays enabled (NVIDIA overlay, Discord, MSI Afterburner, etc.)?
No.
What is your PSU wattage and are you using the default GPU power cables?
MONTECH Century II - 850W High-End ATX Gaming Power Supply - 80 Plus Gold & Cybenetics Platinum
I was using the cable that came with the PSU as it was a 450 watt cable, but as I was concerned about undervolting (even after volting benchmark testing) I did another volting test with the split cable that came with the gpu to two PCIE cables that was attached with the addendum. Needless to say I had nothing to worry about with undervolting as it was within expected ranges.
Have you tried a clean driver installation again with “Perform clean install”?
One of the first things I did, and also you’ll note I went above and beyond by fully removing the driver with DDU and installing it cleanly.
Does the issue occur immediately on launch or only after gameplay begins?
As soon as gameplay begins
Are there any temperature spikes or GPU usage abnormalities before the crash?
No temperature or power abnormalities.
Please let me know if you require any further documentation or follow up information.
NVIDIA REP via channel 'Email’04/14/2026 10:55 AM
Hello,
Thank you for the detailed update. I appreciate the thorough troubleshooting you’ve already performed.
Upon reviewing your logs and observations, I can confirm the following:
- The GPU-Z log shows the GPU operating within expected parameters under load
- GPU temperature peaks around ~69–70°C, which is well within safe limits
- Power draw, voltage, and clocks appear stable with no signs of throttling or power instability
- GPU usage reaches ~99% during load without abnormal drops prior to transitions
Based on this validation:
- There are no indications of thermal, power, or hardware instability from the GPU side
- The issue is reproducible specifically in Unreal Engine 5 titles
- The behavior (texture flickering, lighting artifacts, VRAM spikes, and eventual crash) points toward a rendering/driver interaction rather than a hardware fault
Given everything you have tested so far, this strongly suggests:
- A compatibility issue between the current driver branch and UE5 rendering pipelines (DX12 / shader compilation / post-processing effects)
- Potential instability triggered when GPU utilization reaches sustained 100% in UE5 workloads
Please try the following additional steps to further isolate:
In NVIDIA Control Panel:
- Set “Power management mode” to Prefer maximum performance
- Set “Low Latency Mode” to Off
- Set “Shader Cache Size” to Unlimited
- Limit GPU utilization slightly:
- Cap FPS (e.g., 120 or lower) using in-game limiter or NVIDIA Control Panel
- This helps avoid sustained 100% load conditions where the issue is triggered
- Disable specific UE5-heavy features (if available):
- Lumen (Global Illumination / Reflections)
- Nanite (if toggle exists)
- Post-processing quality → set to Medium/Low
- Clear shader cache:
- Delete contents of:
C:\Users%username%\AppData\Local\NVIDIA\DXCache
C:\Users%username%\AppData\Local\NVIDIA\GLCache
Important note:
Since your system is stable in non-UE5 workloads, passes stress tests, and shows no hardware anomalies, this does not currently indicate a defective GPU. The pattern aligns more with a software/driver compatibility issue.
Please share your results after trying the above steps, and I will assist you further.
Best regards,
NVIDIA Customer Care
Customer (- -) via channel 'CSS Web’04/14/2026 11:58 AM
Alright I have more details for you.
I set everything as you advised and still noticed persistent texture flickering throughout my tests with the additional results:
(my numbers are not related to yours)
Strobing green/red/yellow lights only occur when Post Processing/Illumination is on High (or higher) and Lumen is on. I can track this directly in Incursion Red River as it has a direct toggle for it, whereas for Ever Siege it does not but given the results are identical it is more than likely it uses Lumen. To Reproduce set illumination above medium (Higher with Lumen toggled on. Note: Either post processing can be set above medium OR Lumen can be toggled on but not both at the same time without issue.
Post processing above medium seems to be abnormally GPU intensive. Sitting in the hideout of Incursion Red River (which is an isolated space with just the player and no AI) at medium I would get 50% GPU utilization while standing still with all other display settings set to bare minimum with DLSS disabled Lumen Disabled and using TAA (120 FPS using your prescribed settings FOV 90). At high it averages about 68% while standing still. At Epic it would average about 76% while standing still.
With GPU limited crashing did not occur on Ever Siege nor Incursion Red River
To note: Nanite toggle does not exist for either game.
As it bears repeating: irrespective of the illumination fixing, or GPU utilization minimization, texture flickering persisted throughout.
Let me know if you need any more information!
Thanks!
NVIDIA REP via channel 'Email’04/15/2026 02:44 AM
Helo,
Thank you for your response.
Apologies that the issue persist.
Based on the understanding of the issue, I request you to try the following steps.
Forced PCIe Link Speed (Gen 4 vs Gen 5): The RTX 5070 Ti is a PCIe Gen 5 card. The B850 motherboard and Ryzen 7800X3D support Gen 5. However, signal integrity issues on the motherboard traces or a marginal riser cable (if used) can cause correctable PCIe errors that manifest only under the specific high-bandwidth, high-state-switching load of UE5’s Nanite/Lumen streaming. These errors do not crash Furmark or Cyberpunk 2077, but they corrupt the texture streamer cache in UE5, leading to the “strobing” and flickering the user describes.
Enter BIOS (press Del during boot).
Navigate to Settings > Advanced > PCIe/PCI Subsystem Settings.
Locate the primary GPU slot configuration (usually PCI_E1 Gen Switch).
Change from [Auto] or [Gen5] to [Gen4].
Save and Exit.
Test: If the flickering reduces significantly, it indicates a signal integrity issue on the Gen 5 bus.
- Disable Hardware Accelerated GPU Scheduling (HAGS) & Windowed Optimizations: The user noted “Windowed Borderless dramatically increased stability.” This is a massive diagnostic clue. In Windows 11, Fullscreen Optimizations force the game into a special flip model (Independent Flip) that interacts with HAGS and MPO (Multi-Plane Overlay). A bug in the current driver or Windows Update may be causing the GPU scheduler to incorrectly assign priorities to the UE5 render thread in true Fullscreen mode, leading to the Access Violation when the GPU hits 100% utilization.
Open Windows Settings > System > Display > Graphics.
Click “Change default graphics settings”
Toggle “Hardware-accelerated GPU scheduling” to OFF.
Reboot.
Additionally, locate the game executable (e.g., Ever Siege.exe). Right-click > Properties > Compatibility.
Check “Disable fullscreen optimizations”. (This forces the older, sometimes more stable Exclusive Fullscreen mode bypassing the new flip model).
Please take your time with the steps and let me know the results at your earliest convenience.
If there are any questions or concerns, feel free to contact us.
Regards,
NVIDIA REP
NVIDIA Customer Care
Customer (- -) via channel 'CSS Web’04/15/2026 06:30 AM
Thank you for the apology, I understand bugs are unexpected and they take time to work through. I appreciate the effort involved on your end.
I tried your steps. I tested both the Forced PCIe link by itself, the Disabled Hardware Accelerated GPU Accelerated Scheduling by itself, and with both of those steps together just to be sure to be extra thorough with the testing. The result was no change in flickering.
I also set the applications in question to disable full screen optimization and run as administrator for all three tests.
Thanks!
NVIDIA REP via channel 'Email’04/15/2026 08:35 AM
Hello,
Thank you for the update.
I appreciate the troubleshooting steps you’ve completed so far. Based on the information provided, the issue appears to be related to Unreal Engine 5 (UE5) rather than the NVIDIA driver or GPU.
In this case, I recommend reporting the issue to the developer team for further investigation using the following link:
[Epic Developer Community]
Since all relevant troubleshooting has been completed on our end and no GPU or driver-related issues were identified, the developer team will be best suited to assist further.
Please let us know if you need any additional assistance.
Best regards,
NVIDIA Customer Care
Customer (- -) via channel 'CSS Web’04/15/2026 08:39 AM
Alright, then could you give me a write up that I could then send over to them as a support for this report as I believe you’ll be better able to articulate the precise nature of this bug in question as well as able to help them hone in on its exact nature?
NVIDIA REP 'Email’04/15/2026 11:10 AM
Hello,
Thank you for the update.
Please inform them that you have already contacted NVIDIA Customer Care and that all troubleshooting steps have been completed. Based on our analysis, the issue appears to be related to Unreal Engine 5 (UE5) rather than the NVIDIA driver or GPU.
Regards,
NVIDIA Customer Care
Please select what you are reporting on:
Creative
What Type of Bug are you experiencing?
Other
Steps to Reproduce
Play an Unreal game using latest rendering technology and a 5070-TI will induce texture flickering.
With Lumen and Illumination features set above Low (whatever that means for these three games as I could not get a clear answer from these devs) will induce green/red/yellow strobe lights on light sources.
Expected Result
No texture flickering, no strobing lights
Observed Result
Texture Flickering occurs on certain models under unknown conditions. When GPU load exceeds 100% this will result in a memory allocation error and a crash. From what I can tell this is exclusive to 5070-TIs and is also exclusive to the Unreal Engine.
Lumen+Global Illumination cause an odd interaction where theres green-red-yellow strobing lights from light sources.
Platform(s)
PC