| commit | f577f5bb0beb1993e782e2a29a1f3accde5b1903 | [log] [tgz] |
|---|---|---|
| author | Tommy Chiang <ototot@chromium.org> | Tue Apr 08 07:30:50 2025 |
| committer | Chromeos LUCI <chromeos-scoped@luci-project-accounts.iam.gserviceaccount.com> | Wed Apr 16 18:43:36 2025 |
| tree | 8a995754684b1b97a883f91898a2ee48e2ef2ad3 | |
| parent | 41ed4ec34fcbc6660509e954dd15c68e62e6502e [diff] |
tflite: mtk_neuron: Add ParallelInference test Rauru NPU has multiple cores (MDLAs, and MVPUs), we want to see if Rauru NPU can utilize this feature to allow multiple model running at the same time. This CL add a test case trying to send two models to NPU at the same time. These two models are expected to utilize only one MDLA and one MVPU, so, in theory, they can be run in parallel. My tests show that the time difference between the two inference is usually around 11%, which should indicate that we can run models in parallel. BUG=b:374245241 TEST=neuron_delegate_test on navi Change-Id: I905fa9efbeafed24a182814ce3edefd5372e0ca4 Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/tflite/+/6437794 Commit-Queue: Tommy Chiang <ototot@google.com> Reviewed-by: Shik Chen <shik@chromium.org> Tested-by: Tommy Chiang <ototot@google.com>
This repository hosts the core ChromeOS TFLite components, enabling on-device machine learning (ODML) workloads accelerated by NPU.
The corresponding ebuild can be found at: tensorflow-9999.ebuild
Patches are stored in the patch/ directory and explicitly listed in WORKSPACE.bazel. A helper script, ./script/patcher.py, is included to facilitate patch management within a TFLite workspace.
The typical workflow:
Eject (Download) TensorFlow Source Code
Download the TensorFlow source code into a local git repository with patches applied as individual commits:
./script/patcher.py eject
This creates a new local git repository at tensorflow/.
Modify the TensorFlow Repository
Make changes to the tensorflow/ repository as needed, following standard git workflows. Optionally, include a PATCH_NAME= tag in commit messages to specify the filename of the corresponding patch.
Seal the Repository
Regenerate the patch files and update the WORKSPACE.bazel file:
./script/patcher.py seal
This updates the patches in the patch/ directory and reflects the changes in WORKSPACE.bazel.
It's preferred to submit changes to upstream TensorFlow first and cherry-pick them as patches. This helps minimize divergence and makes TensorFlow updates easier.