[psiddh]

QNN: construct partitioners from options and allocate LLM I/O in rpcmem

Exporters often have backend option dictionaries rather than ready-made compiler specs. QnnPartitioner.from_options() routes each setting to the correct compiler-spec builder or partitioner and rejects unknown names. The rpcmem allocator registers through the preceding PR's generic I/O-buffer hook, so LLM runner code remains independent of QNN.

Review order: QnnPartitioner.from_options, option-routing tests, llm/io_buffer_allocator.cpp, build / llama-main link wiring, then the allocator tests. The allocator forwards allocation alignment, binding and free calls; a failed allocation remains nullptr so the IOManager can use heap fallback.

This is PR 4 of the Stack 2 series. Its local parent is stack2/pr03-runner-contracts; its GitHub base is main, with earlier stack commits included. The QNN delegate output-order change is not in this PR; it follows separately for Qualcomm review.

Test Plan

On this PR's source tree:

python -m pytest -q backends/qualcomm/tests/test_partitioner_options.py
8 passed in 4.05s

python -m pytest -q backends/qualcomm/tests/test_passes.py \
  -k 'partitioner_carries or partitioner_runs or partitioner_from_options'
3 passed, 38 deselected in 4.42s

cmake -S backends/qualcomm/tests -B qnn-tests -DCMAKE_BUILD_TYPE=RelWithDebInfo \
  -DPYTHON_EXECUTABLE=$(command -v python)
cmake --build qnn-tests -j8 --target qnn_io_buffer_allocator_test
ctest --test-dir qnn-tests --output-on-failure
100% tests passed, 0 tests failed out of 2

QNN SDK 2.37 was used. HTP / GPU specs are checked; the LPAI test verifies the genuine >=2.39 SDK rejection rather than claiming LPAI compilation. Allocator tests compile the real allocator and stub only the rpcmem C APIs; hardware shared-memory registration was not rerun. The full host QNN build also compiled the allocator library. Buck parity could not start because the local dotslash wrapper requires a missing worktree .buck2; no Buck-green claim is made.

No external cache-memory-planning cherry-pick is included.

Stack (review in order)

1. Partitioner: run backend-contributed edge transform passes #24566 Partitioner: run backend-contributed edge transform passes
2. Export: avoid reused lifted-constant names #24567 Export: avoid reused lifted-constant names
3. LLM runner: select I/O contracts and accept custom buffer allocators #24568 LLM runner: select I/O contracts and accept custom buffer allocators
4. QNN: construct partitioners from options and allocate LLM I/O in rpcmem #24569 QNN: construct partitioners from options and allocate LLM I/O in rpcmem
5. QNN: bind delegate outputs in exported program order #24570 QNN: bind delegate outputs in exported program order
6. Export hf_static float-cache programs and register their Python runner #24571 Export hf_static float-cache programs and register their Python runner
7. LLM runner: validate and drive hf_static float-cache programs #24572 LLM runner: validate and drive hf_static float-cache programs
8. Document the version-1 hf_static I/O contract #24573 Document the version-1 hf_static I/O contract

Each PR includes the commits of the ones above it; review only its own top commits.

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/24569

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit df791aa with merge base 5f72739 ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.