Error in DXGI: Device was removed, reason provided.
Error in DXGI: Device was removed, reason provided.
I’ve conducted thorough research on Google, ChatGPT, and the available information about this issue. The number of potential causes is extensive, just as is the list of supposed solutions. Regardless of how similar the error is, it appears across various graphics cards, CPU motherboard combinations, and other components, with some fixes working while others don’t.
The system in question:
- Ryzen 5 3600
- Asrock B450M / AC
- Corsair Vengeance RGB PRO 64GB (4 16GB sticks) at 3200Mhz
- Team Group SSD 1 GB (older models, not NVME)
- GTX 1660 TI 6GB
- Off-brand PSU, 500W gold
- Stock AMD cooler
Games played:
- Stardew Valley – no issues
- Deep Rock Galactic – no issues
- Warframe – runs well initially, crashes after ~10 minutes (DXGI)
- Darktide – stable at low settings, 50-60 FPS, crashes after ~15 minutes (DXGI)
- Atomic Heart – similar to Darktide
- Cyberpunk – scales well for the hardware, excellent until ~15 minutes later
I understand it seems more frequent with higher-end, demanding titles. Yet these games are generally compatible across a range of cards and models. All my system checks confirmed it meets the minimum requirements at low settings.
Steps taken:
- Reinstalled GPU drivers via DDU, ensuring they’re up to date
- Installed several older driver versions (around 7)
- Performed registry adjustments
- Undervolted the GPU between -50 and -200
- Updated BIOS, reset to default, disabled XMP
- Lowered game settings
- Disabled anti-aliasing globally
- Turned off latency features
- Reinstalled games, verified integrity
- Switched to a new NVME SSD
- Replaced PSU with a 650W 80Gold unit
- Switched cooler to a Wraith RGB model (not ideal but significantly improved)
- Tested with GTX1080 – ran hot but never exceeded 80-82, handled other adjustments
- Tested with a brand new RTX 3060
- Reinstalled Windows
Despite these efforts, the same crash persists. The exact cause remains unclear. All that’s left is to replace the CPU, motherboard, and RAM. I’ve ordered a new motherboard (B550) and a new CPU (Ryzen 5 5600), but haven’t upgraded memory yet. If anyone wants details, I’ll share my findings – though I’m quite exhausted.
Team Group SSD 1 GB (the older ones, not NVME)
SATA based M.2 or 2.5" SSD?
Corsair Vengeance RGB PRO 64GB (4 16GB sticks) 3200Mhz
For clarity, if these are four sticks made up of two 2x16GB units, check their PCB revisions and let me know.
Off brand PSU, 500W gold
This seems likely to be the main cause of your problems. Non-branded, poorly built power supplies are more likely to fail than a proper PSU for running the whole system.
Take out and reinstall the GPU drivers using DDU, ensuring they are fully updated.
Users should run DDU in Safe Mode to eliminate all GPU drivers (Intel, AMD, Nvidia) before manually installing the latest driver—either the latest version or a compatible one—in an elevated command, such as right-clicking the installer and selecting Run as Administrator.
Updated BIOS, reset back to default settings, and disable XMP (currently running at 2666Mhz).
For reference, please mention the BIOS version of your motherboard at this stage.
Swapped the PSU for a new 650W 80Gold
650W and 80 Gold correspond to the advertised power output and 80+ efficiency ratings, respectively. Could you specify the make and model of your PSU?
Reinstalled Windows
Have you recreated your bootable USB installer to check for corruption? Are you installing the OS in offline mode and then updating drivers there?
I’ve purchased a new motherboard (B550 this time) and a new CPU (Ryzen 5 5600), but not the RAM yet.
Regarding memory, have you experimented with just two sticks of RAM to see if the problem persists?
Also, have you attempted transferring your existing GPU to a donor machine to determine if it’s causing the issue?
Thank you very much for your time; you have no idea what you're doing.
Considering a SATA-based M.2 or 2.5" SSD?
The 2.5" SSD has been replaced with a new NVME model from the same brand:
TEAMGROUP T-Force G50 1TB SLC Caching 3D TLC NAND NVMe InnoGrit PCIe Gen4x4 M.2 2280 Gaming SSD
Could you please share the make and model of your PSU?
Segotep 650W 80 Plus Gold Certified Non-Modular ATX Power Supply
It's recommended to launch DDU in Safe Mode to eliminate all GPU drivers—right-click the installer and select Run as Administrator.
With seven distinct drivers, both new and old, it’s worth noting the BIOS version of your motherboard at this moment.
Date: 4/11/2025
Would you like to recreate your bootable USB installer to check for corruption? Installing the OS offline, then applying all drivers via an elevated command while offline?
Yes, a bootable USB was built on another machine, installed Windows offline, and a separate USB was used to download all motherboard drivers individually from ASROCK.
Regarding memory, have you experimented with just two RAM sticks to see if the problem persists?
Not yet.
Finally, have you attempted transferring your existing GPU to a donor system to determine if it's causing the issue?
No, but I did test three different GPUs in the same setup, all resulting in the same error, consistent timing, and identical performance across games.
GTX 1660 TI (original)
GTX 1080
RTX 3060
At this stage, it’s hard to pinpoint the cause—possibly between the PCIe slot or the CPU. The variety of GPU/CPU combinations over the years has led to many similar problems without a clear solution.
EDIT:
I almost forgot this point:
If you have four sticks (two 2x16GB units), check their PCB revisions and let me know.
Where can I find this information?
Yes, it’s two kits of 16GB each. They were purchased at different times but share the same model numbers. I bought one first, then a few months later another, just clicking “purchase again” on Amazon.
You would find it on the side of your ram sticks;
https://www.corsair.com/us/en/explo...NE...gpGnfWntz5
Corsair is the sole provider that obtains RAM ICs from various suppliers.
I purchased one kit initially, then a couple of months later bought another, simply clicking "purchase again" on Amazon.
I think there might be two distinct IC'd ram kits involved. It’s better to use one kit instead of mixing different ones.
GTX 1660 TI (original)
GTX 1080
RTX 3060
The issue here is that each card has a higher TDP than the previous one.
Adding complexity, the RTX 3000 series was known for significant temporary load spikes.
I recommended checking a donor motherboard because sourcing parts for troubleshooting will only give you brand new components for older technology, which won’t fit a modern build. Sometimes a specific ASRock or Gigabyte board was actually the cause of the problem.