issue in ASA Game and Conan Exiles Enhanced

MoatazD · July 31, 2026, 8:52pm

Summary

GPU Crash Investigation Report

Prepared for submission to: NVIDIA Support, Funcom Support, Studio Wildcard Support
Report Date: July 31, 2026

1. System Specifications

Component	Details
CPU	Intel Core i5-14600K
GPU	NVIDIA GeForce RTX 4060, 8GB VRAM (Adapter: AD107-A, Device ID 2882)
RAM	32GB (tested via MemTest86, multiple passes, JEDEC defaults, no XMP — PASSED)
Motherboard	ASUS Prime B760-Plus D4
PSU	600W (budget brand, ~1 year old)
Storage	NVMe SSD
OS	Windows 11 25H2 (Build 10.0.26200.8973)
GPU Drivers Tested	610.88, 596.36 (Funcom-recommended version), and 3 other recent versions — identical crash signature on all 5

2. Issue Summary

Two distinct problems were identified and are documented separately below.

Issue A — Full System Kernel Crashes (BSOD) — RESOLVED

Issue B — Game-Specific GPU Hang Crashes (Conan Exiles Enhanced & ARK: Survival Ascended) — UNRESOLVED

3. ISSUE A: Full System Kernel Crashes (RESOLVED)

Symptoms

Random full-system BSOD/reboot, unrelated to any specific application.

Crash Events

Date/Time	Bugcheck	Details
2026-07-28 23:56:39	0x3b SYSTEM_SERVICE_EXCEPTION	Access violation in `nt!ExWaitForRundownProtectionRelease`, faulting process `svchost.exe` — kernel-level pointer corruption
2026-07-31 01:40:32	0x1a MEMORY_MANAGEMENT	MEMORY_CORRUPTION_ONE_BIT — corrupt PTE detected, faulting process `BEService.exe`

Root Cause Identified

Live monitoring via HWiNFO64 during gameplay showed:

CPU Core VID reaching 1.499V (excessive for i5-14600K under partial load)
“Core Power Limit Exceeded” = Yes and “Package/Ring Power Limit Exceeded” = Yes

The motherboard’s default power management (“Auto” performance preferences) allowed the CPU to exceed Intel’s official power limits (125W/181W) despite no manual overclocking, consistent with the widely-documented Intel 13th/14th Gen (Raptor Lake) voltage/power instability issue.

Fix Applied

BIOS → Ai Tweaker → Performance Preferences → Intel Default Settings (applied)
Confirmed in BIOS: Long Duration Package Power Limit (PL1) = 125W, Short Duration Package Power Limit (PL2) = 181W
Intel(R) Adaptive Boost Technology → Disabled
CPU Core/Cache Voltage → Offset Mode, Sign [-], value 0.050V (mild undervolt)

Result

No further full-system kernel crashes/BSOD since this fix was applied.

Note

PSU is a budget 600W unit (~1 year old) and remains a secondary suspect for transient voltage sag under rapid GPU power draw spikes, but was deprioritized once the CPU power-limit fix resolved the kernel-level crashes.

4. ISSUE B: Game-Specific GPU Hang Crashes (UNRESOLVED)

Symptoms

Conan Exiles Enhanced: Crashes almost immediately upon entering any world (singleplayer or multiplayer, reproducible even on a private server without BattlEye active).
ARK: Survival Ascended: Crashes are intermittent — occurs a few minutes into gameplay during heavy scene/streaming moments. When triggered, the entire GPU output freezes system-wide for several seconds (confirmed: unrelated GPU-accelerated processes such as video playback and screen-recording software freeze simultaneously), sometimes recovering, sometimes escalating into a full crash.

Aftermath GPU Crash Dump Data (from 3 separate Conan Exiles crash events, different sessions, different NVIDIA driver versions)

Field	Value (identical across all 3 events)
Device State	Hung
Engine Reset	True
Adapter Reset	False
Faulting Shader	`compute_02` (Compute shader)
Shader Hash	5874795813757559664
Fingerprint Hash	3436632223279915891 (identical in all instances)
Occurs during	`Nanite::Streaming` / `GPUSceneUpdate` / `NiagaraGpuComputeDispatch` (AsyncCompute queue)
Engine Version	UE 5.6.1-370197+++exiles+release

NVIDIA Driver-Level Xid Error Correlation

Cross-referencing Windows System Event Log (nvlddmkm provider) against exact crash timestamps:

Game	Local Crash Time	Xid Code	Meaning
ARK: Survival Ascended	2026-07-31 18:34:12	Xid 13 (×6)	Graphics Engine Exception — official NVIDIA documentation states this is an execution-level exception (illegal operation performed by a shader — e.g., out-of-bounds memory access, illegal instruction). NVIDIA’s own guidance: “may be a hardware issue — run field diagnostics to confirm; if not hardware, issue is in application code.”
Conan Exiles Enhanced	2026-07-31 18:28:15	Xid 153 (×3)	Recurs dozens of times throughout normal operation unrelated to crashes — likely a routine/benign power-state event, not diagnostic on its own

Critical Update: Nanite Disabled Test (ARK)

r.Nanite=0 was set in ARK’s Engine.ini (fully disabling Nanite). The identical crash still occurred (Crash ID: CAA64375490D68330C69A0A9B3A89342, same signature: GPU Crash dump Triggered, ArkAscended.exe stack). This rules out Nanite specifically as the sole trigger — the underlying GPU hang is not exclusively tied to Nanite rendering.

Comparative/Control Tests

Test	Result
Fortnite (UE5, Nanite enabled by default, Easy Anti-Cheat, no BattlEye) — 1+ hour continuous play	Zero crashes — rules out a generic UE5+Nanite+GPU/driver incompatibility
Friend on same server, same mods, GTX 1660 Ti (Turing architecture)	Zero crashes in either game
Private server without BattlEye active (Conan)	Crash still occurs — rules out BattlEye as sole root cause (though Funcom’s own community bug tracker lists BattlEye as a contributing/aggravating factor for a similar known issue: “Frequent crashes (singleplayer and multiplayer) - apparently improved by shutting down battleye, but not solved - since June 2026”)

Full List of Troubleshooting Steps Performed (all ruled out as sole root cause)

Tested 5 different NVIDIA driver versions (610.88, 596.36 [Funcom-recommended], and 3 others) — identical crash signature and Fingerprint Hash on all
r.RDG.AsyncCompute=0 — no effect
r.Nanite.AsyncRasterization=0 — no effect
r.Nanite.ProgrammableRaster=0 — no effect (confirmed persisted in Engine.ini post-crash)
r.Nanite.Streaming.Async=0 — no effect
r.Nanite=0 (Nanite fully disabled) — Conan: crash resolved but breaks foliage/tree rendering (Nanite-only assets, no fallback mesh); ARK: crash still occurred
GPU physically reseated in PCIe slot — PCI Express Error Counters stopped climbing (previously continuously incrementing), PCIe Link Speed stabilized near 16.0 GT/s, but crash signature unchanged
TDR delay increased from default 2s to 10s/20s — no change in crash timing whatsoever, confirming a hard GPU hang rather than a timing/performance-threshold issue
Hardware-Accelerated GPU Scheduling toggled both ways — no effect
Discord overlay disabled — no effect
DirectX/NVIDIA shader caches fully cleared — no effect
Verified Engine.ini modifications were actually being read by the game engine (file modification timestamps confirmed changes on each launch)
CPU voltage/power-limit issue (Issue A above) corrected — unrelated, did not affect Issue B
Full clean OS reinstall + DDU (Display Driver Uninstaller) clean driver installs performed multiple times throughout testing

Engine/Game Details at Time of Crash (from crash context)

ARK crash occurred on map Astraeos_WP, 62,479 NPCs loaded (62,409 stasised), 20 mods active, 1 player connected, ~51 FPS at time of crash
GPU: NVIDIA GeForce RTX 4060, Driver 596.36 confirmed in crash context
CPU: Intel i5-14600K, 32GB RAM

5. Requested Action

NVIDIA Support: Please investigate Xid 13 (Graphics Engine Exception) occurring specifically and reproducibly under Unreal Engine 5.5/5.6 titles (Conan Exiles Enhanced, ARK: Survival Ascended) on RTX 4060 (AD107). Per your own documentation, Xid 13 recurring with a specific workload strongly suggests an application-level shader bug, but given the identical crash reproduces across 5 different driver versions, a driver/hardware compatibility investigation with AD107 + these specific UE5 game builds is requested.
Funcom Support / Studio Wildcard Support: Please review the attached Aftermath GPU crash dumps and shader hash (compute_02, hash 5874795813757559664, Fingerprint Hash 3436632223279915891 for Conan). The crash reproduces consistently and is NOT resolved by disabling Nanite in ARK, ruling out Nanite as the sole cause. Requesting confirmation whether this matches a known internal issue.

6. Attachments to Include When Submitting

Aftermath GPU crash dumps (.nv-gpudmp files) — located in each game’s Saved/Crashes/UECC-* folder
UEMinidump.dmp files from the same crash folders
Windows kernel minidumps from Issue A (for reference/context only): C:\WINDOWS\Minidump\*.dmp
This report

Report compiled from live diagnostic sessions including HWiNFO64 monitoring, WinDbg/cdb minidump analysis, NVIDIA Aftermath crash dump analysis, and Windows Event Log correlation.

What Type of Bug are you experiencing?

Other

Steps to Reproduce

123

Expected Result

123

Observed Result

123

Affects Versions

5.6

Platform(s)

Windows