[Feature Request] TRT_MoE Support for older GPU architectures (sm90, sm89, sm80)
Open
Nobody has claimed this yet.
Feature Request
- Dominant language
- C++
- Stars
- 13.4k
- Forks
- 2.4k
- Avg merge
- 5d 3h
- Merged PRs (30d)
- 2
Description
Currently, trtmoe only supports Blackwell or Thor architectures.
MoE requires Blackwell or Thor GPU architecture (SM 10.x or SM 11.x)
Are there any plans on the roadmap to add support for Hopper (sm90), Ada (sm89), or Ampere (sm80) in the future? Many developers in the community are still relying on these GPUs for local development and inference.
Thanks!
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
No file, test, or entry point is named. Start by locating TRT_MoE's GPU architecture checks and existing Blackwell/Thor support, then determine whether Hopper (sm90), Ada (sm89), and Ampere (sm80) are feasible; done would mean documented and tested support for the requested architectures.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- machine-learning, performance
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100