Closed einarpetersen closed 1 year ago
A user said: According to that, your GPU falls under the requirements for InvokeAI - check them out here: https://invoke-ai.github.io/InvokeAI/#hardware-requirements
— Today at 6:37 PM Works perfectly just using the CPU - Why would adding GPU make it fail - But yeah I see the recommendations.... hmmm Or I should perhaps say used to work 🙂
— Today at 6:48 PM I will see if I can downgrade to CPU again and if it will work in the new UI Einar Petersen (Comics Creator)
— Today at 7:22 PM did not help to do the : pip install InvokeAI --use-pep517 --extra-index-url https://download.pytorch.org/whl/cpu install.... still getting File "/home/ep/InvokeAI/.venv/lib/python3.9/site-packages/torch/nn/modules/module.py", line 987, in convert return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking) torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 74.00 MiB (GPU 0; 1.95 GiB total capacity; 1.24 GiB already allocated; 6.06 MiB free; 1.28 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
So somehow InvokeAI still tries to use the CUDA device despite it being installed as CPU only...
export CUDA_VISIBLE_DEVICES=-1
(set in your shell environment as persistently as you like)pip uninstall torch
and then pip install torch --force-reinstall --extra-index-url https://download.pytorch.org/whl/cpu
. This should ensure you are using the CPU version of torch
.pip
caches, and try to reinstall again to make sure you're not getting a GPU torch
from somewhere.But if (1) works, that might be all that's needed.
Ended up doing a nuclear - I can generate images again, however masking/in-painting no longer works....
As in no mask appears no matter brush color picked layer chosen - Simply non existent - eraser has also disappeared
no mask appears no matter brush color picked layer chosen
This seems to be a UI issue; couldn't reproduce it on my end. Could you please try the Reset Web UI
button under Settings and see if it helps?
There has been no activity in this issue for 14 days. If this issue is still being experienced, please reply with an updated confirmation that the issue is still being experienced with the latest release.
Did a pip upgrade and then inpaint etc. started working again
There has been no activity in this issue for 14 days. If this issue is still being experienced, please reply with an updated confirmation that the issue is still being experienced with the latest release.
Is there an existing issue for this?
OS
Linux
GPU
cuda
VRAM
nvidia-smi --query-gpu=memory.total --format=csv,noheader 1999 MiB
What happened?
So I recently got my Nvidia up and running in Debian 11, however now because InvokeAI is trying to use CUDA I'm getting the following error:
File "/home/ep/InvokeAI/invokeai/lib/python3.9/site-packages/torch/nn/modules/module.py", line 987, in convert return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking) torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 74.00 MiB (GPU 0; 1.95 GiB total capacity; 1.21 GiB already allocated; 21.38 MiB free; 1.26 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
Here is an entire dump from starting --web
A matching Triton is not available, some optimizations will not be enabled. Error caught was: No module named 'triton'
** trying to reload previous model
** This can be caused by a missing or corrupted models file, and can sometimes be fixed by (re)installing the models. Do you want to run invokeai-configure script to select and/or reinstall models? [y] n
Opening the Web UI - I see GREEN Connected Dreamlike-diffusion-1.0
Trying to generate produces the following error
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 1.95 GiB total capacity; 24.36 MiB already allocated; 512.00 KiB free; 26.00 MiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
I can't seem to load any models, ger ERROR in the UI when I do - I know the models are installed, have downloaded multiple times and it was working prior in 2.3.0+a0 - Now after "upgrade" new UI with negative prompt says 2.3.0 weird
Has anyone gotten further - I can see a ton of people asking about very similar issues
Operating System: Debian GNU/Linux 11 KDE Plasma Version: 5.20.5 KDE Frameworks Version: 5.78.0 Qt Version: 5.15.2 Kernel Version: 5.10.0-21-amd64 OS Type: 64-bit Processors: 8 × Intel® Core™ i7-4790K CPU @ 4.00GHz Memory: 15.6 GiB of RAM Graphics Processor: NVIDIA GeForce GTX 750 Ti/PCIe/SSE2
System free -h total used free shared buff/cache available Mem: 15Gi 2.1Gi 9.8Gi 39Mi 3.7Gi 13Gi Swap: 975Mi 20Mi 955Mi
GPU:
nvidia-smi --query-gpu=memory.total --format=csv,noheader 1999 MiB
Thu Feb 16 18:10:10 2023
+-----------------------------------------------------------------------------+ | NVIDIA-SMI 470.103.01 Driver Version: 470.103.01 CUDA Version: 11.4 | |-------------------------------+----------------------+----------------------+ | GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | | | | MIG M. | |===============================+======================+======================| | 0 NVIDIA GeForce ... On | 00000000:01:00.0 On | N/A | | 22% 30C P0 2W / 38W | 343MiB / 1999MiB | 0% Default | | | | N/A | +-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+ | Processes: | | GPU GI CI PID Type Process name GPU Memory | | ID ID Usage | |=============================================================================| | 0 N/A N/A 843 G /usr/lib/xorg/Xorg 129MiB | | 0 N/A N/A 1392 G /usr/bin/kwin_x11 29MiB | | 0 N/A N/A 1460 G /usr/bin/plasmashell 31MiB | | 0 N/A N/A 1564 G ...b/firefox-esr/firefox-esr 119MiB | | 0 N/A N/A 3189 G /usr/bin/systemsettings5 25MiB | +-----------------------------------------------------------------------------+
Screenshots
No response
Additional context
No response
Contact Details
No response