docker-compose.yml

version: '3.9'

x-gpu-base-service: &gpu_service
  #runtime: nvidia
  privileged: true
  devices:
    - /dev/nvidia0:/dev/nvidia0
    - /dev/nvidiactl:/dev/nvidiactl
    - /dev/nvidia-caps:/dev/nvidia-caps
    - /dev/nvidia-modeset:/dev/nvidia-modeset
    - /dev/nvidia-uvm:/dev/nvidia-uvm
    - /dev/nvidia-uvm-tools:/dev/nvidia-uvm-tools
  deploy:
    resources:
      reservations:
        devices:
          - driver: nvidia
            device_ids: ['0']
            capabilities: [gpu]

x-base_service: &base_service
  user: "${UID:-0}:${GID:-0}"
  #network_mode: "host"
  ports:
    - "7860:7860"
  build:
    context: ./services/AUTOMATIC1111
    args:
      # Compile time args
      pyver: "3.10"
      pyimage: python:3.10-slim
      #
      #XFORMERS_COMMAND: /bin/bash /docker/install-container-dep.sh --upgrade-strategy only-if-needed  /docker/xformers-*.whl
      XFORMERS_COMMAND: /bin/bash /docker/install-container-dep.sh xformers==0.0.23.post1
      #
      #TORCH_COMMAND: /bin/bash /docker/install-container-dep.sh /docker/tensorflow-*.whl /docker/torch-*.whl /docker/torchvision-*.whl /docker/torchaudio-*.whl
      TORCH_COMMAND: /bin/bash /docker/install-container-dep.sh torch==2.1.2+cu118 torchvision==0.16.2+cu118 torchaudio==2.1.2+cu118
      PYTORCH_CUDA_ALLOC_CONF: "garbage_collection_threshold:0.9,max_split_size_mb:256"
      TORCH_CUDA_ARCH_LIST: 7.5+PTX
      #
      #TENSORFLOW_COMMAND: /bin/bash /docker/install-container-dep.sh /docker/tensorflow-*.whl
      TENSORFLOW_COMMAND: /bin/bash /docker/install-container-dep.sh tensorflow[and-cuda]==2.15.0.post1
      #
      # General configuration
      PIP_REPOSITORY: "https://download.pytorch.org/whl/cu118"
      #
      TRITON_VERSION: "2.1.0"
      #
      DEEPSPEED: False
      DEEPSPEED_VERSION: "0.12.6"
      # https://www.tensorflow.org/install/pip#software_requirements
      # https://developer.nvidia.com/tensorrt-download
      CUDNN_VERSION: "8.6.0.163"
      CUDA_VERSION: cuda-11-8
      #CUDA_DRIVERS: cuda-drivers
      CUDA_KEYRING: https://developer.download.nvidia.com/compute/cuda/repos/debian11/x86_64/cuda-keyring_1.1-1_all.deb
      #
      DS_BUILD_OPS: 1
      NVCC_FLAGS: --use_fast_math
      JAX: False
      TPU: False
      #
      # History time args
      stableDiffusionV1_SHA: "21f890f9da3cfbeaba8e2ac3c425ee9e998d5229"     # https://github.com/P2Enjoy/stable-diffusion.git
      stableDiffusionV2_SHA: "cf1d67a6fd5ea1aa600c4df58e5b47da45f6bdbf"     # https://github.com/P2Enjoy/stable-diffusion-v2.git
      stableDiffusionXL_SHA: "9d759324e914de6c96dbd1468b3a4a50243c6528"     # https://github.com/P2Enjoy/stable-diffusion-xl.git
      codeFormer_SHA: "c5b4593074ba6214284d6acd5f1719b6c5d739af"            # https://github.com/P2Enjoy/CodeFormer.git
      BLIP_SHA: "3a29b7410476bf5f2ba0955827390eb6ea1f4f9d"                  # https://github.com/P2Enjoy/BLIP.git
      latentDiffusion_SHA: "abf33e7002d59d9085081bce93ec798dcabd49af"       # https://github.com/P2Enjoy/latent-diffusion.git
      tamingTransformers_SHA: "24268930bf1dce879235a7fddd0b2355b84d7ea6"    # https://github.com/P2Enjoy/taming-transformers.git
      kDiffusion_SHA: "cc49cf6182284e577e896943f8e29c7c9d1a7f2c"            # https://github.com/P2Enjoy/k-diffusion.git
      clip_interrogator_SHA: "bc07ce62c179d3aab3053a623d96a071101d11cb"     # https://github.com/P2Enjoy/clip-interrogator.git
      MiDas_SHA: "1645b7e1675301fdfac03640738fe5a6531e17d6"                 # https://github.com/P2Enjoy/MiDaS.git
      GFPGAN_SHA: "2eac2033893ca7f427f4035d80fe95b92649ac56"                # https://github.com/P2Enjoy/GFPGAN.git
      CLIP_SHA: "a1d071733d7111c9c014f024669f959182114e33"                  # https://github.com/P2Enjoy/CLIP.git
      open_clip_SHA: "73ad04ae7fb93ede1c02dc9040a828634cb1edf1"             # https://github.com/P2Enjoy/open_clip.git
      AUTO1111_SHA: "5ef669de080814067961f28357256e8fe27544f4"              # https://github.com/P2Enjoy/stable-diffusion-webui.git
  volumes:
    - &v1 ./data:/data
    - &v2 ./output:/output
    - &v3 ./data/config/auto/extensions:/stable-diffusion-webui/extensions
    - &v4 /tmp/.X11-unix:/tmp/.X11-unix
  deploy:
    restart_policy:
      delay: 5s
      max_attempts: 10
      window: 120s

name: webui-docker

services:
  download:
    build: ./services/download/
    profiles: ["download"]
    volumes:
      - *v1

  xformers:
    <<: *gpu_service
    build: ./services/xformers/
    profiles: ["xformers"]
    volumes:
      - ./services/xformers/data:/deploy
    environment:
      - TORCH_CUDA_ARCH_LIST=7.5+PTX
      - NVCC_FLAGS=--use_fast_math

  tensorflow:
    <<: *gpu_service
    build: ./services/tensorflow/
    profiles: ["tensorflow"]
    environment:
      - LIBNVINFER_VERSION=7.2.2
      - LIBNVINFER_MAJOR_VERSION=7
      - LIBNVINFER_CUDA_VERSION=11.1
      - CUDNN_VERSION=8.1.1.33
      - CUDNN_MAJOR_VERSION=8
      - CUDNN_CUDA_VERSION=11.2
      - CUDA_NVRTC=cuda-nvrtc-11-1
      - CUDA_NVRTC_DEV=cuda-nvrtc-dev-11-1
      - CUDA_NVRTC_VERSION=11.1.105-1
      - COMPUTE=7.5
      - MAX_GCC_VERSION=9
      - GIT_BRANCH=v2.11.0
    volumes:
      - ./services/tensorflow/data:/deploy

  auto: &automatic
    <<: [*base_service, *gpu_service]
    profiles: ["auto"]
    environment:
      - ACCELERATE=False
      - TF_ENABLE_ONEDNN_OPTS=1
      - USE_MEMORY_EFFICIENT_ATTENTION=0
      - DISPLAY=unix$DISPLAY
      - CLI_ARGS=--medvram --medvram-sdxl --opt-sdp-attention --xformers --opt-sub-quad-attention --opt-split-attention 
      - RUN_ARGS=/stable-diffusion-webui/webui.py --listen --enable-insecure-extension-access --port 7860 --allow-code --api
      - RUNNER=/docker/run.sh
      
  auto_debug:
    <<: *automatic
    profiles: ["auto_debug"]
    stdin_open: true
    tty: true
    environment:
      - ACCELERATE=False
      - TF_ENABLE_ONEDNN_OPTS=1
      - USE_MEMORY_EFFICIENT_ATTENTION=0
      - DISPLAY=unix$DISPLAY
      - NO_TCMALLOC="True"
      - NVIDIA_VISIBLE_DEVICES=all
      - NVIDIA_DRIVER_CAPABILITIES=compute,utility
      - CLI_ARGS=--always-batch-cond-uncond --opt-sdp-attention --xformers --opt-sub-quad-attention --opt-split-attention 
      - RUN_ARGS=/stable-diffusion-webui/webui.py --listen --enable-insecure-extension-access --port 7860 --allow-code --api
      - DEBUG=/docker/run.sh $$RUN_ARGS --ckpt-dir $${ROOT}/models/Stable-diffusion $$CLI_ARGS
      - RUNNER=/docker/debug.sh

  auto-cpu:
    <<: *base_service
    profiles: ["auto-cpu"]
    environment:
      - CLI_ARGS=--no-half --precision full