NVIDIA / NVIDIA/TensorRT

A request for a more practical loop integration

Open
#3,326 5 comments 0 reactions 3 assignees View on GitHub

@rajeevsrao is already working on this.

Since Sep 20, 2023.

triaged
Dominant language
C++
Stars
13.4k
Forks
2.4k
Avg merge
5d 3h
Merged PRs (30d)
2

Description

With the rise of diffusion models, it feels sad not being able to put the full diffusion loop inside a TensorRT model.

Unfortunately managing control-flow and loops is a nightmare in TensorRT. Sorry for the harsh word.

It would be great to be able to have an easier integration for loops that wouldn't require tracing the full loop in torch and sending a huge graph to ONNX.

Maybe this is possible at the moment, if so could you provide a simple example on how to do so ?

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.