I tried getting Stable Diffusion running using this guide, but when I try running webui-user.bat, it's giving me this:
Python 3.10.6 [tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49] [MSC v.1932 64 bit [AMD64]] Commit hash: Traceback [most recent call last]: File "D:\stable-diffusion-webui-master\launch.py", line 293, in
prepare_enviroment[]
run_python["import torch; assert torch.cuda.is_available[], 'Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check'"]
return run[f'"{python}" -c "{code}"', desc, errdesc]
raise RuntimeError[message]
Looking into this error, I found this, which near the end there is a post saying how to disable the check, but that only leads me to get an error upon trying to generate an image saying ""LayerNormKernelImpl" not implemented for 'Half'", and looking into this, it seems to be related to CUDA, so it's likely related. Looking into CUDA, I found it's an NVIDIA thing, but I do have an NVIDIA GPU [According to task manager: NVIDIA GeForce GTX 1050 Ti]. Or rather, I have two GPUs, one Intel and one NVIDIA [maybe it's trying the wrong GPU?]. I've never worked with GPUs or any of this stuff, so I have no idea what's going on beyond what I've stated.
Any help on how to get this working?
Edit: I figured this post would just fall into obscurity, but since I'm still getting the occasional response, I feel I should point out that as per my conversation with CommunicationCalm166 in the comments, I have fixed the issue: I just needed to update my Nvidia drivers. That said, others facing similar issues may still find value in the other solutions posted.
@jcha7965 Greetings - I humbly and possibly ignorantly suggest fully listing your physical available resources, your desired VM resources, and your current VM resources.
@RSchuddinh @RPUL @sudheee02 @KazuXee
Thank you for the thread, all. It helped me save my job and get my VM back online quickly, although not as fully desired.
I fully believe it is my lack of understanding and knowledge to blame for the struggles I'm having getting to the desired server and VM configuration.
Physical Server Resources available:
Dell R740XD
2x Intel[R] Xeon[R] Gold 6148 CPU @ 2.40GHz
24x 32GB 2400MT/s - 768GB ECC RAM
Dell BOSS card, 2x micron 480ish nvme
PERC H730 Adapter, 4x HUC101212CSS600 HGST 1.2TB
H730 controlling "backplane" [which is actually front of server, love that we can't say "frontplane" lol]
BOSS in PCIe slot. I read it was supposed to default to RAID 1, but it did not, or at least, I can't find where it for sure tells me that is operating correctly in RAID 1, which I desire to verify
Predeployment phase
Currently only 1 VM
VM Status:
256GB of nvme
32 GB of RAM
12 cores Windows Server '16
I desire VM to get to:
OS only on nvme, and BOSS verified and running in RAID1
Be able to understand this enough that some smaller programs, but needed fast operations, will also run on nvme
other programs will be setup to run on the disk drives in raid 10, OR i might find compatible sas ssd
The Raid10 is not setup yet, i believe it is currently running in jbod mode as i have 4TB of storage available to assign to the VM
I desire ability to assign up to 128GB of RAM to the VM, likely only 32-64GB will be required, but until that is known for sure and I will monitor for usage, I need to allow certain programs within the VM to "stretch their legs" a little bit as we're testing some settings and hosts.
I desire ability to get 10G networking to the server for use by all VMs. Can this be done securely? I want to understand this all well enough that a financial program can run in its own VM and be completely secure into the network from the other VMs...is that possible or is a 2 or 4 port 10g nic required?
Thank you so much, I will update here and below with information as I find it elsewhere or you helpful people reply.