pytorch / pytorch/executorch

Improve qualcomm end-to-end user experience

Open
#18,126 0 comments 0 reactions 1 assignee View on GitHub

@abhinaykukkadapu is already working on this.

Since Mar 12, 2026.

  • #17989 by @abhinaykukkadapu — merged
  • #17990 by @abhinaykukkadapu — merged
Dominant language
Python
Stars
5k
Forks
1.2k
Avg merge
2d 10h
Merged PRs (30d)
581

Description

This is a task to capture all the things we need to improve in Qualcomm user flow:

  1. Improve AOT experience
  2. Improve build from source
  3. Improve building binaries workflow

Current top pain points:

  1. Fragmented entry points and user experience downloading QNN sdk and Android NDK, libc++ - unify the download paths, be transparent instead of silently downloading 2+GB sdks
  2. Retry downloads on failures
  3. Start from partial downloads on retry
  4. Cache downloaded sdks
  5. Build QNN on linux by default (enable via pybind.cmake) - enable download QNN_SDK during cmake
  6. In build from source path, enable QNN by default on linux
  7. Make glibc search and fallback transparent
  8. Manage all QNN sdk version config at a single point instead of hardcoded paths
  9. Set all env vars after downloading the sdk

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.