ranova

Thank you for your work and that information, I guess I was using some legacy nvidia docker commands: services: llama-swap: image: ghcr.io/mostlygeek/llama-swap:cuda13 container_name: llama-swap-cuda restart: unless-stopped ports: - "8887:8080" volumes: - /mnt/user/AI/models:/models environment: - NVIDIA_VISIBLE_DEVICES=all - NVIDIA_DRIVER_CAPABILITIES=all - LD_LIBRARY_PATH=/custom-bin/bin:/usr/local/cuda/lib64:/usr/lib/x86_64-linux-gnu - GGML_CUDA_FORCE_FA3=1 - GGML_CUDA_GRAPH_OPT=1 deploy: resources: reservations: devices: - driver: nvidia count: all capabilities: [gpu] entrypoint: ["/app/llama-swap"] command: ["--config", "/app/config.yaml", "--listen", "0.0.0.0:8080"] Updated it to the following and its now working again: version: "3.8" services: llama-swap: image: ghcr.io/mostlygeek/llama-swap:cuda13 container_name: llama-swap-cuda restart: unless-stopped runtime: nvidia ports: - "8887:8080" volumes: - /mnt/samsungevo/AI/models:/models environment: - NVIDIA_VISIBLE_DEVICES=GPU-xxx-xxx-xxx-xxxx-xxxx,GPU-xxx-xxx-xxx-xxxx-xxxx - NVIDIA_DRIVER_CAPABILITIES=all - LD_LIBRARY_PATH=/custom-bin/bin:/custom-bin-turbo/bin:/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/local/cuda/lib64:/usr/lib/x86_64-linux-gnu - GGML_CUDA_FORCE_FA3=1 - GGML_CUDA_GRAPH_OPT=1 entrypoint: ["/app/llama-swap"] command: ["--config", "/app/config.yaml", "--listen", "0.0.0.0:8080"]

I updated to open source 595 driver and none of my docker containers are launching that requires the 3090. How do i revert back to the 590 open source driver since I also have a 5060ti in the system getting this error trying to startup: Failed starting container: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running prestart hook #0: exit status 1, stdout: , stderr: Auto-detected mode as 'legacy' nvidia-container-cli: ldcache error: process /sbin/ldconfig failed with error code: 1: unknown edit: had to restore the last known good unraid flash backup to get open sourced 590 driver and my GPUs working again. Really not sure if it was the 595 driver update or something got corrupted. I had updated to the 595 driver, didnt reboot yet because I was hitting the 3090 with a lot of LLM requests and it seemingly froze the 3090. Upon rebooting, 595 was active and both the 3090 and 5060ti were detected but I kept getting the ldconfig errors when trying to load docker containers that used the GPUs

Sorry, I meant when can we expect 590 drivers to work with blackwell/50 series cards? Are we waiting on nvidia or unraid to do updates? 590 brings the latest cuda features which is required for newer features such as MXFP4: https://huggingface.co/blog/RakshitAralimatti/learn-ai-with-me

ah thanks for that confirmation. Are we waiting on an updated unraid kernel or nvidia to release the open source versions?

I am having issues trying to get my 5060ti working with 590 or production drivers. It just refuses to detect the GPU unless I use 580 open source drivers. I am trying to use this card for AI and it seems I might need 590 with CUDA 13.1 to enable features like MXFP4 quants. Here are dmesg related to nvidia: [ 53.783367] nvidia: loading out-of-tree module taints kernel. [ 53.784393] nvidia: module license 'NVIDIA' taints kernel. [ 53.785380] Disabling lock debugging due to kernel taint [ 53.786380] nvidia: module license taints kernel. [ 53.789893] atlantic: Detect ATL2FW 1030025 [ 53.897722] atlantic: Detect ATL2FW 1030025 [ 53.898530] RAPL PMU: API unit is 2^-32 Joules, 3 fixed counters, 655360 ms ovfl timer [ 53.899526] RAPL PMU: hw unit of domain pp0-core 2^-14 Joules [ 53.900515] RAPL PMU: hw unit of domain package 2^-14 Joules [ 53.901483] RAPL PMU: hw unit of domain pp1-gpu 2^-14 Joules [ 53.904155] nvidia-nvlink: Nvlink Core is being initialized, major device number 238 [ 53.910461] nvidia 0000:02:00.0: enabling device (0000 -> 0003) [ 53.912629] nvidia 0000:02:00.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=none [ 53.926592] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 590.48.01 Mon Dec 8 11:22:45 UTC 2025 [ 54.333462] cryptd: max_cpu_qlen set to 1000 [ 54.333828] i915 0000:00:02.0: [drm] Found ALDERLAKE_P (device ID 46a8) display version 13.00 stepping D0 [ 54.336174] i915 0000:00:02.0: [drm] VT-d active for gfx access [ 54.336817] Console: switching to colour dummy device 80x25 [ 54.336845] i915 0000:00:02.0: vgaarb: deactivate vga console [ 54.336901] i915 0000:00:02.0: [drm] Using Transparent Hugepages [ 54.337326] i915 0000:00:02.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=io+mem [ 54.338442] nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms 590.48.01 Mon Dec 8 12:50:27 UTC 2025 [ 54.338811] AES CTR mode by8 optimization enabled [ 54.340218] [drm] [nvidia-drm] [GPU ID 0x00000200] Loading driver [ 54.340351] [drm] Initialized nvidia-drm 0.0.0 for 0000:02:00.0 on minor 1 [ 126.681089] NVRM: The NVIDIA GPU 0000:02:00.0 (PCI ID: 10de:2d04) NVRM: installed in this system requires use of the NVIDIA open kernel modules. [ 126.681112] NVRM: GPU 0000:02:00.0: RmInitAdapter failed! (0x22:0x56:1001) [ 126.682133] NVRM: GPU 0000:02:00.0: rm_init_adapter failed, device minor number 0 [ 126.686160] nvidia_uvm: module uses symbols nvUvmInterfaceDisableAccessCntr from proprietary module nvidia, inheriting taint. [ 128.677050] NVRM: The NVIDIA GPU 0000:02:00.0 (PCI ID: 10de:2d04) NVRM: installed in this system requires use of the NVIDIA open kernel modules. [ 128.677071] NVRM: GPU 0000:02:00.0: RmInitAdapter failed! (0x22:0x56:1001) [ 128.678182] NVRM: GPU 0000:02:00.0: rm_init_adapter failed, device minor number 0 [ 132.298419] NVRM: The NVIDIA GPU 0000:02:00.0 (PCI ID: 10de:2d04) NVRM: installed in this system requires use of the NVIDIA open kernel modules. [ 132.298441] NVRM: GPU 0000:02:00.0: RmInitAdapter failed! (0x22:0x56:1001) [ 132.299501] NVRM: GPU 0000:02:00.0: rm_init_adapter failed, device minor number 0 [ 132.333543] NVRM: The NVIDIA GPU 0000:02:00.0 (PCI ID: 10de:2d04) NVRM: installed in this system requires use of the NVIDIA open kernel modules. [ 132.333563] NVRM: GPU 0000:02:00.0: RmInitAdapter failed! (0x22:0x56:1001) [ 132.334915] NVRM: GPU 0000:02:00.0: rm_init_adapter failed, device minor number 0 Here is my diagnostics. nvidia-smi wont detect anything under production or new feature branches ugreen-diagnostics-20260121-1607.zip

Ahhh thanks, that did it, installed bc via nerdtools and was able to set aspm for my arc GPU

Trying to get that ASPM script to work for 04:00.0 VGA compatible controller: Intel Corporation DG2 [Arc A310] (rev 05) (prog-if 00 [VGA controller]) LnkCap: Port #0, Speed 2.5GT/s, Width x1, ASPM L0s L1, Exit Latency L0s <64ns, L1 <1us LnkCtl: ASPM Disabled; RCB 64 bytes, LnkDisable- CommClk- lspci -t output: -[0000:00]-+-00.0 +-06.0-[01]----00.0 +-06.2-[02-05]----00.0-[03-05]--+-01.0-[04]----00.0 | \-04.0-[05]----00.0 +-07.0-[06-2f]-- +-07.2-[30-59]-- +-0d.0 +-0d.2 +-0d.3 +-14.0 +-14.2 +-15.0 +-15.1 +-16.0 +-17.0 +-1c.0-[5a]----00.0 +-1c.2-[5b]----00.0 +-1c.4-[5c]----00.0 +-1c.6-[5d]----00.0 +-1d.0-[5e]----00.0 +-1f.0 +-1f.3 +-1f.4 \-1f.5 I tried the following in the aspm script: ROOT_COMPLEX="00:06.2" ENDPOINT="04:00.0" I get this error loop: /tmp/user.scripts/tmpScripts/aspm arc/script: line 172: printf: 0x: invalid hex number /tmp/user.scripts/tmpScripts/aspm arc/script: line 179: bc: command not found setpci: Unknown register "". Any ideas?

Thanks for the UGREEN LED Driver plugin. Was wondering if you could add options to it to: turn off LED if a hard drive is spun down in Unraid modify brightness level Change what the first LED does. Right now it just blinks bright blue, not sure what that means and would rather be static

Not sure if this can be done, but is there a way to add support for dGPU fan control? I have an Intel Arc a310 and while idle, the GPU fan spins up and down every few seconds and its quite annoying!

On a different topic, when I try to install "ITE IT87 Driver", I get this error message: plugin: installing: it87-driver.plg Executing hook script: pre_plugin_checks plugin: downloading: it87-driver.plg ... done Executing hook script: pre_plugin_checks -----------------------Downloading IT87 Drivers!------------------------- ---------This could take some time, please don't close this window!---------- -----------------------Can't download IT87 Drivers----------------------- plugin: run failed: '/bin/bash' returned 1 Executing hook script: post_plugin_checks

ahh, Im using the custom kernel from thor, which added the Ugreen LED driver, thought I could use the two together

@ich777 saw that you released the ugreen-led driver. How do you get this running on the Ugreen Nas lines? Thanks!