[Feature]: AutoDeploy: Revisit the flashinfer Attention API

[nvchenghaoz]

🚀 The feature, motivation and pitch

Looks like the flashinfer attention api is a bit old in autodeploy. With new backends (e.g. trtllm-gen) and different layout (HND), we might want to update the flashinfer API calls to make sure that AutoDeploy uses the best kernel.

Alternatives

No response

Additional context

No response

Before submitting a new issue...

• Make sure you already searched for relevant issues, and checked the documentation and examples for answers to frequently asked questions.

[lucaslie]

this is probably blocked by flashinfer upgrade in Tensorrt-LLM right?