Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

split model between host memory and GPU #113

Open
pommedeterresautee opened this issue Jul 7, 2022 · 0 comments
Open

split model between host memory and GPU #113

pommedeterresautee opened this issue Jul 7, 2022 · 0 comments
Labels
enhancement New feature or request ONNX Runtime

Comments

@pommedeterresautee
Copy link
Member

pommedeterresautee commented Jul 7, 2022 •
edited
Loading

With Triton it's certainly possible to move a model between host and device memory.
Combined with pipeline parallelization we may support model bigger than the available memory.

inspiration: https://arxiv.org/pdf/2207.00032.pdf
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
enhancement New feature or request ONNX Runtime
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
1 participant
@pommedeterresautee