What is the issue you’re having?
Voici le rapport rempli avec toutes les infos qu’on a collectées :
What is the issue you’re having?
When connecting to a server with pure mode enabled (sv_pureLevel), RedM must restart its client process to switch build/apply pure mode (launch args show -switchcl:). During this restart, the game consistently crashes or gets stuck on the “no-ui” (“:(”) screen. Two distinct crash types have been observed during this exact transition:
RAGE error: ERR_GFX_STATE (crash_hash: redm.exe+25B22EF)
Failed to allocate memory for Vulkan. VkResult: VK_ERROR_OUT_OF_DEVICE_MEMORY (crash_hash: glue.dll+937C2)
Logs show EADDRINUSE (libuv) socket conflicts occurring at the same time as the switch, suggesting the old process (referenced by -switchcl:) may not be fully releasing GPU/VRAM and network port resources before the new process attempts to reallocate them. First-time launches (fresh cache, no pure mode) work perfectly fine — the crash is specific to the build-switch/pure-mode-restart step.
What have you tried already to fix the issue?
Deleted RedM cache folder
Switched UpdateChannel: beta → canary (both tested, same crash on transition)
Reinstalled NVIDIA driver via DDU + clean install
Tested both Vulkan and DX12 graphics API
Disabled NVIDIA overlay
Verified/killed lingering RedM/RDR2 processes before each test (RedM.exe, RedM_ROSService, RedM_b1491_GTAProcess.exe, RedM_ChromeBrowser)
Full system restarts between attempts
Confirmed only Studio driver installed (previously had both Game Ready + Studio installed simultaneously — removed the conflict)
What server did you get this issue on?
178.208.177.134:30120 (pure mode enabled server)
Windows version
Windows 11 24H2/25H2
System specifications
GPU: NVIDIA GeForce RTX 5060 Ti (PCI ID 10de:2d04)
NVIDIA Driver: v610.62.0.0 (Studio driver)
Vulkan API version: 1.4.341
RDR2 build: 1.0.1491.50 (b1491)
Antivirus software
defender windows
CitizenFX crash zip file (‘Save information’ on a crash)
CfxCrashDump_2026_07_13_20_06_50.zip (2.05 MB)
Log files
CitizenFX_log_2026-07-13T200613.log (56.6 KB)
Additional comments
This appears to be an RTX 50-series-specific timing/resource-handoff issue in the pure mode client restart mechanism (-switchcl), likely related to the recently merged RedM pure mode support (still canary-only as of writing). Both crash types occur ~7-8 seconds after the new process reaches the Render init stage, always during the switch — never on a normal (non-pure) connection. Happy to provide additional dumps or test proposed fixes if a dev wants to follow up.
