RuntimeError: No CUDA GPUs are available w/2 RTX4060ti 16gb

v0.4.6 Win 10
[2025-01-09 02:31:10.470] [error] Adding extra search path checkpoints C:\Users\mikee\Documents\ComfyUI\models\checkpoints\

[2025-01-09 02:31:10.472] [error] Adding extra search path classifiers C:\Users\mikee\Documents\ComfyUI\models\classifiers
Adding extra search path clip C:\Users\mikee\Documents\ComfyUI\models\clip
Adding extra search path clip_vision C:\Users\mikee\Documents\ComfyUI\models\clip_vision
Adding extra search path configs C:\Users\mikee\Documents\ComfyUI\models\configs
Adding extra search path controlnet C:\Users\mikee\Documents\ComfyUI\models\controlnet
Adding extra search path diffusers C:\Users\mikee\Documents\ComfyUI\models\diffusers
Adding extra search path diffusion_models C:\Users\mikee\Documents\ComfyUI\models\diffusion_models
Adding extra search path embeddings C:\Users\mikee\Documents\ComfyUI\models\embeddings
Adding extra search path gligen C:\Users\mikee\Documents\ComfyUI\models\gligen
Adding extra search path hypernetworks C:\Users\mikee\Documents\ComfyUI\models\hypernetworks
Adding extra search path loras C:\Users\mikee\Documents\ComfyUI\models\loras
Adding extra search path photomaker C:\Users\mikee\Documents\ComfyUI\models\photomaker
Adding extra search path style_models C:\Users\mikee\Documents\ComfyUI\models\style_models
Adding extra search path unet C:\Users\mikee\Documents\ComfyUI\models\unet
Adding extra search path upscale_models C:\Users\mikee\Documents\ComfyUI\models\upscale_models
Adding extra search path vae C:\Users\mikee\Documents\ComfyUI\models\vae
Adding extra search path vae_approx C:\Users\mikee\Documents\ComfyUI\models\vae_approx
Adding extra search path animatediff_models C:\Users\mikee\Documents\ComfyUI\models\animatediff_models
Adding extra search path animatediff_motion_lora C:\Users\mikee\Documents\ComfyUI\models\animatediff_motion_lora
Adding extra search path animatediff_video_formats C:\Users\mikee\Documents\ComfyUI\models\animatediff_video_formats
Adding extra search path ipadapter C:\Users\mikee\Documents\ComfyUI\models\ipadapter
Adding extra search path liveportrait C:\Users\mikee\Documents\ComfyUI\models\liveportrait
Adding extra search path insightface C:\Users\mikee\Documents\ComfyUI\models\insightface
Adding extra search path layerstyle C:\Users\mikee\Documents\ComfyUI\models\layerstyle
Adding extra search path LLM C:\Users\mikee\Documents\ComfyUI\models\LLM
Adding extra search path Joy_caption C:\Users\mikee\Documents\ComfyUI\models\Joy_caption
Adding extra search path sams C:\Users\mikee\Documents\ComfyUI\models\sams
Adding extra search path blip C:\Users\mikee\Documents\ComfyUI\models\blip
Adding extra search path CogVideo C:\Users\mikee\Documents\ComfyUI\models\CogVideo
Adding extra search path xlabs C:\Users\mikee\Documents\ComfyUI\models\xlabs
Adding extra search path instantid C:\Users\mikee\Documents\ComfyUI\models\instantid
Adding extra search path custom_nodes C:\Users\mikee\Documents\ComfyUI\custom_nodes/
Adding extra search path download_model_base C:\Users\mikee\Documents\ComfyUI\models
Setting output directory to: C:\Users\mikee\Documents\ComfyUI\output
Setting input directory to: C:\Users\mikee\Documents\ComfyUI\input
Setting user directory to: C:\Users\mikee\Documents\ComfyUI\user

[2025-01-09 02:31:11.495] [info] [START] Security scan
[DONE] Security scan

ComfyUI-Manager: installing dependencies done.

** ComfyUI startup time:
[2025-01-09 02:31:11.496] [info] 2025-01-09 02:31:11.495437

[2025-01-09 02:31:11.534] [info] ** Platform: Windows
[2025-01-09 02:31:11.535] [info]
** Python version: 3.12.8 (main, Dec 19 2024, 14:41:01) [MSC v.1942 64 bit (AMD64)]
** Python executable: C:\Users\mikee\Documents\ComfyUI.venv\Scripts\python.exe
** ComfyUI Path: C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI
** Log path: C:\Users\mikee\Documents\ComfyUI\comfyui.log

[2025-01-09 02:31:12.510] [error]
Prestartup times for custom nodes:

[2025-01-09 02:31:12.511] [error] 2.0 seconds: C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI\custom_nodes\ComfyUI-Manager

[2025-01-09 02:31:12.545] [error] Set cuda device to: 2

[2025-01-09 02:31:14.567] [error] Traceback (most recent call last):

[2025-01-09 02:31:14.568] [error] File “C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI\main.py”, line 136, in
import execution
File “C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI\execution.py”, line 13, in
import nodes
File “C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI\nodes.py”, line 22, in
import comfy.diffusers_load
File “C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI\comfy\diffusers_load.py”, line 3, in

[2025-01-09 02:31:14.569] [error] import comfy.sd
File “C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI\comfy\sd.py”, line 6, in
from comfy import model_management
File “C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI\comfy\model_management.py”, line 145, in
total_vram = get_total_memory(get_torch_device()) / (1024 * 1024)

[2025-01-09 02:31:14.570] [error] ^^^^^^^^^^^^^^^^
[2025-01-09 02:31:14.571] [error] ^^
File “C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI\comfy\model_management.py”, line 114, in get_torch_device
return torch.device(torch.cuda.current_device())
^^^
[2025-01-09 02:31:14.572] [error] ^^^^^^^^^^^^^^^^^^^^^^^^
File “C:\Users\mikee\Documents\ComfyUI.venv\Lib\site-packages\torch\cuda_init_.py”, line 940, in current_device
lazy_init()
[2025-01-09 02:31:14.573] [error]
File "C:\Users\mikee\Documents\ComfyUI.venv\Lib\site-packages\torch\cuda_init
.py", line 319, in _lazy_init
torch._C._cuda_init()
RuntimeError: No CUDA GPUs are available

This appears to be a regular CUDA error, so we need to check on a few things.

  1. 2x RTX4060ti - can you please detail how they’re configured?
  2. Are drivers up to date?
  3. OS updates done?
  4. If there isn’t an obvious fix from that, we’ll need to do isolation testing. The fastest way to get through that is to find out what is / is not working - can you please test some other apps, e.g.:
    a) Any other 3d / design software you have installed
    b) A modern game or two

The tests need to occur after you received the ComfyUI error, but really only need to confirm that the apps start and suffer no obvious performance issues.

Oh wait a minute - did you use command line args or the server settings panel to set the cuda device to 2?

Please remove that argument, or set it to 0 or 1 (unless you add a third 4060 :slight_smile:).

Thank you for you response. I was able to get in once I did a complete reinstall. I set it to 1 and it started up normally. But for some reason its still not seeing the 2nd card. I’ve downloaded the MultiGPU extension but when I add the UnetLoader it only sees CPU and Cuda:0. I there anything else I need to DL? Here is the log there are still errors but it will let me generate flux schnell images but I get OOM errors when trying to use Flux Dev.

[2025-01-10 02:19:37.288] [info] [START] Security scan
[DONE] Security scan

ComfyUI-Manager: installing dependencies done.

[2025-01-10 02:19:37.289] [info] ** ComfyUI startup time: 2025-01-10 02:19:37.288713

[2025-01-10 02:19:37.318] [info] ** Platform: Windows

[2025-01-10 02:19:37.319] [info] ** Python version: 3.12.8 (main, Dec 6 2024, 18:10:20) [MSC v.1942 64 bit (AMD64)]
** Python executable: C:\Users\mikee\Documents\ComfyUI.venv\Scripts\python.exe
** ComfyUI Path: C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI
** Log path: C:\Users\mikee\Documents\ComfyUI\comfyui.log

[2025-01-10 02:19:38.254] [error]
Prestartup times for custom nodes:

[2025-01-10 02:19:38.255] [error] 2.0 seconds: C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI\custom_nodes\ComfyUI-Manager

[2025-01-10 02:19:38.283] [error] Set cuda device to: 1

[2025-01-10 02:19:40.228] [error] Total VRAM 16380 MB, total RAM 65478 MB

[2025-01-10 02:19:40.230] [error] pytorch version: 2.5.1+cu121
Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA GeForce RTX 4060 Ti : cudaMallocAsync

[2025-01-10 02:19:41.407] [error] Using pytorch attention

[2025-01-10 02:19:42.851] [error] [Prompt Server] web root: C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI\web_custom_versions\desktop_app

[2025-01-10 02:19:43.301] [error] MultiGPU: Initialization started

[2025-01-10 02:19:43.303] [error] MultiGPU: Initial device cuda:0
MultiGPU: Starting core node registration
MultiGPU: Registered core node UNETLoader
MultiGPU: Registered core node VAELoader
MultiGPU: Registered core node CLIPLoader
MultiGPU: Registered core node DualCLIPLoader
MultiGPU: Registered core node TripleCLIPLoader
MultiGPU: Registered core node CheckpointLoaderSimple
MultiGPU: Registered core node ControlNetLoader
MultiGPU: Checking for module at custom_nodes\ComfyUI-LTXVideo
MultiGPU: Module ComfyUI-LTXVideo not found - skipping
MultiGPU: Checking for module at custom_nodes\ComfyUI-Florence2
MultiGPU: Module ComfyUI-Florence2 not found - skipping
MultiGPU: Checking for module at custom_nodes\ComfyUI_bitsandbytes_NF4
MultiGPU: Module ComfyUI_bitsandbytes_NF4 not found - skipping
MultiGPU: Checking for module at custom_nodes\x-flux-comfyui
MultiGPU: Module x-flux-comfyui not found - skipping

[2025-01-10 02:19:43.303] [error] MultiGPU: Checking for module at custom_nodes\ComfyUI-MMAudio
MultiGPU: Module ComfyUI-MMAudio not found - skipping
MultiGPU: Checking for module at custom_nodes\ComfyUI-GGUF
MultiGPU: Module ComfyUI-GGUF not found - skipping
MultiGPU: Checking for module at custom_nodes\PuLID_ComfyUI
MultiGPU: Module PuLID_ComfyUI not found - skipping
MultiGPU: Registration complete. Final mappings: UNETLoaderMultiGPU, VAELoaderMultiGPU, CLIPLoaderMultiGPU, DualCLIPLoaderMultiGPU, TripleCLIPLoaderMultiGPU, CheckpointLoaderSimpleMultiGPU, ControlNetLoaderMultiGPU

[2025-01-10 02:19:43.307] [info] ### Loading: ComfyUI-Manager (V2.55.3)

[2025-01-10 02:19:43.309] [info] ### ComfyUI Revision: UNKNOWN (The currently installed ComfyUI is not a Git repository)

[2025-01-10 02:19:43.314] [error]
Import times for custom nodes:
0.0 seconds: C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI\custom_nodes\websocket_image_save.py

[2025-01-10 02:19:43.315] [error] 0.0 seconds: C:\Users\mikee\Documents\ComfyUI\custom_nodes/ComfyUI-MultiGPU
0.0 seconds: C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI\custom_nodes\ComfyUI-Manager

[2025-01-10 02:19:43.322] [error] Starting server

[2025-01-10 02:19:43.323] [error] To see the GUI go to: h**p://127.0.0.1:8000

[2025-01-10 02:19:43.395] [info] [ComfyUI-Manager] default cache updated: h**ps://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/alter-list.json

[2025-01-10 02:19:43.409] [info] [ComfyUI-Manager] default cache updated: h**ps://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/model-list.json

[2025-01-10 02:19:43.424] [info] [ComfyUI-Manager] default cache updated: h**ps://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/github-stats.json

[2025-01-10 02:19:43.468] [info] [ComfyUI-Manager] default cache updated: h**ps:// raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/extension-node-map.json

[2025-01-10 02:19:43.496] [info] [ComfyUI-Manager] default cache updated: h**ps:// raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json

[2025-01-10 02:19:45.001] [info] FETCH DATA from: C:\Users\mikee\AppData\Local\Programs@comfyorgcomfyui-electron\resources\ComfyUI\custom_nodes\ComfyUI-Manager\extension-node-map.json
[2025-01-10 02:19:45.004] [info] [DONE]

thanks in advance

(i had to add ** to the urls to post)

Oh, BTW

  1. No special configuration just plugged em in and they worked.
  2. Drivers up to date
  3. OS up to date
  4. Text Gen WebUI uses both cards for training llms and LMStudio uses both cards for inference.
    No modern games :frowning: but GTA5 and 7D2D run normally

I have no experience with that extension, I’m afraid. I’d recommend reaching out to the author on their GH repo, as I doubt they’re looking here for issues.