Open mathiasfoster opened 2 years ago
Same here. Running python3.8 under Linux Mint 20.04
EDIT: It's not getting passed self.unet = self.core.compile_model(self._unet, device)
at https://github.com/bes-dev/stable_diffusion.openvino/blob/master/stable_diffusion.py#L40. device == 'CPU'
and self._unet
==
if that helps.
@mathiasfoster it looks like the fish terminal automatically terminates hanged (???) process. But I can't to reproduce this error on my side, could you please provide more information about it?
How much RAM do you have free? I had this same issue because I only had 6G free and the EOM killer kicked in.
Getting same error. Seems like issue with RAM.
if you are in linux, you can have a swap file the bigger you want. https://wiki.archlinux.org/title/swap#Swap_file
I was on WSL2 machine and the problem was because the RAM was limited to around 7.5K unit of memory. I increased it to 12GB and it seems to work fine. However for 32 iterations it takes around 3 mins to generate.
wsl2 docs for increasing limit
Had this error too, increasing ram did remove the error and threw another: #93
I was having a similar issue with SIGKILL killing my process, usually after the "Downloading..." step. Bumping the swapfile up to 6GB resolved this for me:
swapoff /swapfile # May take a minute
fallocate -l 6G /swapfile
mkswap /swapfile
swapon /swapfile
(Ubuntu 22.04, zsh, Python 3.10, 16GB RAM, Intel i7-6600U CPU @ 2.60GHz × 4)
I get a persistent error when my job finishes... the process is killed, and no image is outputted.
Here's an example using 2 inference steps only (so it runs quickly):
The same thing happens in both
fish
andbash
.What is causing this, and how can I get the job to complete?