Hugging Face· PlatformStory of the week

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

Hugging Face blog post describing how to accelerate SD Turbo and SDXL Turbo inference using ONNX Runtime and Olive, with performance benchmarks showing up to 229% throughput gains for SDXL Turbo and 120% for SD Turbo.

From the source

ONNX Runtime outperformed PyTorch for all (batch size, number of steps) combinations tested, with throughput gains as high as 229% for the SDXL Turbo model and 120% for the SD Turbo model.

huggingface.co

Announced

15 Jan 2024 00:00 UTC