tlwu/sd-turbo-onnxruntime

Stable Diffusion Turbo for ONNX Runtime CUDA

Introduction

This repository hosts the optimized ONNX models of SD Turbo to accelerate inference with ONNX Runtime CUDA execution provider for Nvidia GPUs. It cannot run in other providers like CPU and DirectML.

The models are generated by Olive with command like the following:

python stable_diffusion.py --provider cuda --model_id stabilityai/sd-turbo --optimize --use_fp16_fixed_vae

See the usage instructions for how to run the SDXL pipeline with the ONNX files hosted in this repository.

Model Description

• Developed by: Stability AI
• Model type: Diffusion-based text-to-image generative model
• License: STABILITY AI NON-COMMERCIAL RESEARCH COMMUNITY LICENSE
• Model Description: This is a conversion of the SD-Turbo model for ONNX Runtime inference with CUDA execution provider.

Performance

Latency

Below is average latency of generating an image of size 512x512 using NVIDIA A100-SXM4-80GB GPU:

Engine	Batch Size	Steps	ONNX Runtime CUDA
Static	1	1	38.2 ms
Static	4	1	120.2 ms
Static	1	4	68.7 ms
Static	4	4	192.6 ms

Static means the engine is built for the given batch size and image size combination, and CUDA graph is used to speed up.

Usage Example

Following the demo instructions. Example steps:

0. Install nvidia-docker using these instructions.

1. Clone onnxruntime repository.

git clone https://github.com/microsoft/onnxruntime
cd onnxruntime

2. Download the ONNX files from this repo

git lfs install
git clone https://huggingface.co/tlwu/sd-turbo-onnxruntime

3. Launch the docker

docker run --rm -it --gpus all -v $PWD:/workspace nvcr.io/nvidia/pytorch:23.10-py3 /bin/bash

4. Build ONNX Runtime from source

export CUDACXX=/usr/local/cuda-12.2/bin/nvcc
git config --global --add safe.directory '*'
sh build.sh --config Release  --build_shared_lib --parallel --use_cuda --cuda_version 12.2 \
            --cuda_home /usr/local/cuda-12.2 --cudnn_home /usr/lib/x86_64-linux-gnu/ --build_wheel --skip_tests \
            --use_tensorrt --tensorrt_home /usr/src/tensorrt \
            --cmake_extra_defines onnxruntime_BUILD_UNIT_TESTS=OFF \
            --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=80 \
            --allow_running_as_root
python3 -m pip install build/Linux/Release/dist/onnxruntime_gpu-*-cp310-cp310-linux_x86_64.whl --force-reinstall

If the GPU is not A100, change CMAKE_CUDA_ARCHITECTURES=80 in the command line according to the GPU compute capacity (like 89 for RTX 4090, or 86 for RTX 3090). If your machine has less than 64GB memory, replace --parallel by --parallel 4 --nvcc_threads 1 to avoid out of memory.

5. Install libraries and requirements

python3 -m pip install --upgrade pip
cd /workspace/onnxruntime/python/tools/transformers/models/stable_diffusion
python3 -m pip install -r requirements-cuda12.txt
python3 -m pip install --upgrade polygraphy onnx-graphsurgeon --extra-index-url https://pypi.ngc.nvidia.com

6. Perform ONNX Runtime optimized inference

python3 demo_txt2img.py \
  "starry night over Golden Gate Bridge by van gogh" \
  --version sd-turbo   \
  --engine-dir /workspace/sd-turbo-onnxruntime