Accelerating SD Turbo and SDXL Turbo Inference with ONNX Runtime and Olive

AI Impact Summary

Stable Diffusion Turbo and SDXL Turbo models are being accelerated using ONNX Runtime and Olive, significantly improving inference speeds on NVIDIA GPUs. This optimization leverages ONNX Runtime’s CUDA and TensorRT execution providers, achieving throughput gains of up to 229% for SDXL Turbo and 120% for SD Turbo compared to PyTorch. This shift allows for faster image generation, particularly beneficial for workflows requiring rapid iteration and experimentation with these models.