| commit | f0fac24c6037984b9d4b76f14eac47be897770ed | [log] [tgz] |
|---|---|---|
| author | Jerry Wu <cheyuw@google.com> | Wed Apr 26 01:36:33 2023 +0000 |
| committer | GitHub <noreply@github.com> | Tue Apr 25 21:36:33 2023 -0400 |
| tree | 8f551383b3381b43916a9adceb12e614032f4383 | |
| parent | 7a86df40bca9ccb5ae84a467922063de8d414a96 [diff] |
Move benchmark config generation to build_e2e_test_artifacts (#13291) When working on https://github.com/openxla/iree/pull/13273, I noticed that benchmark configs and modules are uploaded in 3 different jobs. These artifacts can be uploaded to different GCS dirs when some of them are re-run due to failures (because re-run jobs will create [GCS_DIR with new run attempts](https://github.com/openxla/iree/blob/main/.github/workflows/benchmark_compilation.yml#L44-L50)). As a result, users might not be able to download all benchmark artifacts from a single GCS dir URL, which can be confusing. This PR changes the workflow to generate all benchmark modules and configs in `build_e2e_test_artifacts` to avoid such issue. All files are uploaded to `${GCS_URL}/e2e-test-artifacts`, a single path to download all benchmark artifacts. Besides the reason above, I think during the CMake build we should generate the benchmark configs under `${IREE_BUILD_DIR}/e2e_test_artifacts` instead of calling `export_benchmark_config.py` separately. There are some questions about how to pass the benchmark presets/filters through CMake configuration, so I decided to defer that as the next step.
IREE (Intermediate Representation Execution Environment, pronounced as “eerie”) is an MLIR-based end-to-end compiler and runtime that lowers Machine Learning (ML) models to a unified IR that scales up to meet the needs of the datacenter and down to satisfy the constraints and special considerations of mobile and edge deployments.
See our website for project details, user guides, and instructions on building from source.
IREE is still in its early phase. We have settled down on the overarching infrastructure and are actively improving various software components as well as project logistics. It is still quite far from ready for everyday use and is made available without any support at the moment. With that said, we welcome any kind of feedback on any communication channels!
See our website for more information.
IREE is licensed under the terms of the Apache 2.0 License with LLVM Exceptions. See LICENSE for more information.