Stable diffusion tests on the N100 Mini PC.

apt update
apt full-upgrade
apt install build-essential git mercurial
apt install google-perftools libgl1 libglib2.0-0

apt install python3-full
update-alternatives --remove-all python
update-alternatives --install /usr/bin/python python /usr/bin/python3 1
update-alternatives --config python
python -V
Python 3.11.2

The system has 16 GB RAM so make at least 24 GB swap space.

swapoff /var/swap
dd if=/dev/zero of=/var/swap bs=4M count=6144
mkswap /var/swap
echo "/var/swap none swap sw 0 0" >> /etc/fstab
swapon /var/swap

Github

mkdir -p ~/src && cd ~/src
git clone --recursive https://github.com/AUTOMATIC1111/stable-diffusion-webuii
cd stable-diffusion-webui stable-diffusion-webui.automatic1111
echo 'export PYTORCH_TRACING_MODE=TORCHFX' >> webui-user.sh
echo 'export COMMANDLINE_ARGS="--skip-torch-cuda-test --precision full --no-half --listen"' >> webui-user.sh

./webui.sh

Github

mkdir -p ~/src && cd ~/src
git clone --recursive https://github.com/openvinotoolkit/stable-diffusion-webui stable-diffusion-webui.openvino
cd stable-diffusion-webui
wget "https://raw.githubusercontent.com/openvinotoolkit/stable-diffusion-webui/refs/heads/master/scripts/openvino_accelerate.py" -O scripts/openvino_accelerate.py

diff -u requirements.txt.bak requirements.txt
--- requirements.txt.bak        2024-12-08 10:05:07.531033313 +0000
+++ requirements.txt    2024-12-08 10:21:51.939748587 +0000
@@ -28,7 +28,8 @@
 scikit-image>=0.19
 timm
 tomesd
-torch
+torch==2.1.0
+torchvision==0.16.0
 torchdiffeq
 torchsde
 diffusers

diff -u webui-user.sh.orig webui-user.sh
--- webui-user.sh.orig  2024-12-08 00:47:06.593715977 +0000
+++ webui-user.sh       2024-12-08 09:59:09.850730395 +0000
@@ -9,8 +9,11 @@
 # Name of the subdirectory
 #clone_dir="stable-diffusion-webui"

+export PYTORCH_TRACING_MODE=TORCHFX
+export USE_OPENVINO=1
+
 # Commandline arguments for webui.py, for example: export COMMANDLINE_ARGS="--medvram --opt-split-attention"
-#export COMMANDLINE_ARGS=""
+export COMMANDLINE_ARGS="--skip-torch-cuda-test --precision full --no-half --listen"

 # python3 executable
 #python_cmd="python3"

./webui.sh

Github

mkdir -p ~/src && cd ~/src
git clone --recursive https://github.com/vladmandic/automatic automatic.sd.next
cd automatic

The process gets killed with OOM even with 16 GB RAM and 24 GB swap space performing any operation.

./webui.sh --debug --use-openvino --listen --device-id 0

My SD WEBUI