On one of my PCs, I encounter frequent intermittent errors, both when starting Unreal Engine 5.5.4 and when making Blueprint edits or running my game in editor. I also had these errors on earlier versions of Unreal before upgrading (5.5.3 and 5.5.2). The errors are most often MeshDescription / Oodle decompression or EXCEPTION_ACCESS_VIOLATION errors. I have attached a few of them to this post. My project code (code that I have written) is 100% blueprint. Interestingly, I have moved my project to another PC (similar PC, but AMD processor, Nvidia 4080 and less RAM) and I do not encounter these crashes. Outside of Unreal Engine, PC does not crash (except Baldur’s Gate 3, but I believe that is an Unreal Engine based game).
Here are the specs of the crashing PC:
CPU: INTEL® Core™ i9-14900K
GPU: NVIDIA® GeForce RTXT™ 4080 Super
DRAM: 96GB VENGEANCE DDR5-6000
STORAGE: 4TB M.2 NVMe SSD
MOTHERBOARD: MSI B760M
OS: WINDOWS® 11 PRO
Here are the things I have done to attempt to fix these crashes (none of which have rectified the crashes) :
- Updated to latest NIVIDIA Drivers (I’ve done clean full reinstalls and tried both the studio and game drivers and reset the global settings to defaults)
- Full reinstall of both Epic games launcher and Unreal 5.5.4 (I also ran verify on the 5.5.4 install)
- Deleted the DerivedDataCache, Saved, and Intermediate directories in both my project directory and in the Unreal Engine AppData directory (\Users<my-username>\AppData\Local\UnrealEngine\5.5)
- Switched from DirectX12 to DirectX11 (in this case, some of the crashes were different, but still had a lot of crashes)
- Reduced Engine Scalability settings from “High” to “Low” or “Medium”
- Shut off anything running in the background (Discord, NVidia overlay, etc)
- Created a completely blank project and still encountered crashes (although a bit less frequent).
- Tried shutting Oodle compression off in project settings (still get crashes including Oodle crashes).
I am also aware that some intel i9-14900k chips have had voltage issues. Here is what I tried in case that was the problem:
- Updated latest bios including updating to latest Intel Microcode (version 0x12B) - I did get fewer crashes with this, but did not fix Oodle error and EXCEPTION_ACCESS_VIOLATION errors. I had a lot more RHI/GPUish looking crashes before.
- Lowered Intel CPU Lite Load from 12 to 9 (this was my attempt at “under-volting” - this did nothing but half my performance). Unfortunately, my bios restricts me from making many of the other under-volting/underclocking recommendations (for example, I am unable to tune my active performance core ratios from 57x to 53x as many have suggested).
- Verified RAM XMP is enabled
- Ran Cinebench performance testing (No crashes)
- Ran performance benchmark on Intel XTU - (after microcode update, CPU was no longer thermal throttled)
- Ran PC Manufacturer Diagnostic tests (Corsair diagnostics - CPU, GPU, memory, storage - all passed)
Before I reach out to my PC manufacturer and/or Intel - am I missing anything that I should try? I did turn nanite on an off on a bunch of things including skeletal meshes and landscapes. I’m not sure if that might have created some sort of artifacts that cause things to hiccup? Are there any other temp files anywhere I should try to clear out?
