Driver v.595 + RTX PRO 4500 Blackwell - crashes even when watching videos in the browser

Graphics / Linux Linux
1

Hello.

I’m experiencing intermittent crashes: a black

nvidia-bug-report.log.gz (719,4 КБ)

screen, the graphics card fan turns on at 100%, and there’s no response to any actions other than a hard reset.

Here are some messages from the logs:

Спойлер

kwin_wayland[2207]: Invalid framebuffer status: “GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT”

kernel: [drm:nv_drm_gem_alloc_nvkms_memory_ioctl [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000a00] Failed to allocate NVKMS memory for GEM object

kernel: nvidia-modeset: ERROR: GPU:0: Failed to query display engine channel state: 0x0000ca7e:6:0:0x0000000f

kernel: nvidia-modeset: ERROR: GPU:0: The requested configuration of display devices (GBT AORUS FI27Q (DP-0)) is not supported on this GPU.

kernel: [drm:nv_drm_gem_alloc_nvkms_memory_ioctl [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000a00] Failed to allocate NVKMS memory for GEM object

kwin_wayland[2207]: Pageflip timed out! This is a bug in the nvidia-drm kernel driverkwin_wayland[2207]: Please report this at https://forums.developer.nvidia.com/c/gpu-graphics/linuxkernel: BUG:

kernel NULL pointer dereference, address: 0000000000000030

kernel: #PF: supervisor read access in kernel mode

kernel: #PF: error_code(0x0000) - not-present page

And so on. Bug report attached.

P.S.: Previously, I had an AMD RX 7800 XT and there were no crashes. After installing the RTX PRO 4500 Blackwell, the first crash occurred on the second day. The graphics card was purchased on 3/31/2026.

2

Greetings.

My attempts to resolve the issue myself seem to have been successful.

What helped:

Спойлер

(Due to a more recent driver, tested on Windows 595.97 vs. 590.x on Linux)

  1. While troubleshooting the issue with the AI ​​bot, it asked for the EDID, which I provided. It said there was an error in the EDID, which it offered to fix.

(!) See the attached report. The monitor issue started around 2022, if I remember correctly. At the time, I had an RTX 2080 Super and a different RAM kit (2x32 GB XMP 3200 MHz from Kingston). There was an issue with the monitor resolution, which was resolved by changing the monitor firmware from F03 to F08. (Other users have also had resolution issues, and with other video cards, such as the 3080.) After a while, the monitor started crashing—black screens, graphics card fans at 100%, and only a hard reset helped. Unable to find a solution, I bought an RX 7800 XT, and the crashes stopped.

(!) Whether you can trust the AI ​​bot’s conclusions is up to you to decide; I’m no expert.

Finally, after I entered the corrected EDID through the registry, the crashes stopped… At least, that day I spent hours testing benchmarks and games that usually caused crashes, and there were no crashes.

  1. But after a while, the crashes returned. I continued my search. I tested the RAM. My RAM: G.Skill Ripjaws V [F4-3600C16Q-128GVK] - showed errors at 3600 MHz XMP in Memtest, but no errors in JDEC at 2667 MHz for 15 hours.

I tested with a different RAM set, and no errors either.

I returned the G.Skill Ripjaws V [F4-3600C16Q-128GVK] and… the errors disappeared… Same number of tests, same 3600 MHz XMP profile. The only change was that I changed the installation order: some memory was in slot 1, so I moved it to slot 2 or 3, etc. - I wasn’t paying attention.

No crashes in the last week. Same tests, same games.
The main thing is that my work processes, which run for hours, don’t cause the monitor to crash like they used to.

What was it? Most likely, a combination of issues: a faulty EDID, the NVIDIA driver (the RX 7800 XT didn’t crash the monitor with the same monitor, an uncorrected EDID, and the same memory configuration), and changing the RAM slots.

I don’t have time to run tests under Linux anymore; I bought a different monitor.

edid-crash-report-en.pdf (94,2 КБ)