Can we achieve data parallelism and model parallelism in onnxruntime?

[deng-ShiFu]

Hi,
We want to implement parallel inference on multiple FPGAs, such as data parallelism or model parallelism.We would like to know whether onnxruntime supports this kind of functionality.

If it is not supported, can we achieve this requirement by extending onnxruntime? If you have relevant ideas, can you briefly describe them so that we can be inspired?Thank you.