torch-mlir/.github/workflows/buildAndTest.yml

136 lines
5.2 KiB
YAML
Raw Normal View History

name: Build and Test
on:
push:
branches:
- main
pull_request:
workflow_dispatch:
jobs:
build-validate:
strategy:
fail-fast: true
matrix:
os: [macos-12, ubuntu-20.04]
targetarch: [x86_64, AArch64]
python-version: ["3.10"]
torch-binary: [ON, OFF]
llvmtype: [source, binary]
llvmbuildtype: [in-tree, out-of-tree]
exclude:
# No need for "out-of-tree LLVM and PyTorch source"
- llvmtype: source
llvmbuildtype: in-tree
- llvmtype: binary
llvmbuildtype: out-of-tree
# Disable M1 builds until https://github.com/llvm/torch-mlir/issues/1094 is fixed
- targetarch: AArch64
# macOS we only do source builds to reduce options
- os: macos-12
torch-binary: OFF
- os: macos-12
llvmtype: source
- os: macos-12
llvmtype: out-of-tree
runs-on: ${{ matrix.os }}
steps:
2021-09-28 05:07:24 +08:00
- name: Get torch-mlir
uses: actions/checkout@v2
with:
submodules: 'true'
- uses: ./.github/actions/setup-build
with:
cache-suffix: ${{ matrix.os }}-${{ matrix.targetarch }}-${{ matrix.buildtype }}
- name: llvm-binary-torch-src-or-binary
# Should be the fastest builds for CI and fails fast
# OSX CMake flags are ignore on Linux
if: matrix.llvmtype == 'binary'
run: |
cd $GITHUB_WORKSPACE
cmake $GITHUB_WORKSPACE/externals/llvm-project/llvm -B build -GNinja \
-DCMAKE_BUILD_TYPE=Release \
-DCMAKE_LINKER=lld \
-DCMAKE_C_COMPILER_LAUNCHER=ccache -DCMAKE_CXX_COMPILER_LAUNCHER=ccache \
-DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ \
-DPython3_EXECUTABLE=$(which python) \
-DLLVM_ENABLE_ASSERTIONS=ON \
2021-09-28 05:07:24 +08:00
-DLLVM_ENABLE_PROJECTS=mlir \
-DLLVM_EXTERNAL_PROJECTS="torch-mlir;torch-mlir-dialects" \
2021-09-28 05:07:24 +08:00
-DLLVM_EXTERNAL_TORCH_MLIR_SOURCE_DIR="$GITHUB_WORKSPACE" \
-DLLVM_EXTERNAL_TORCH_MLIR_DIALECTS_SOURCE_DIR="${GITHUB_WORKSPACE}/external/llvm-external-projects/torch-mlir-dialects" \
2021-09-28 05:07:24 +08:00
-DMLIR_ENABLE_BINDINGS_PYTHON=ON \
-DTORCH_MLIR_ENABLE_MHLO=ON \
-DTORCH_MLIR_USE_INSTALLED_PYTORCH=${{ matrix.torch-binary }} \
-DCMAKE_OSX_ARCHITECTURES=${{ matrix.taregetarch }} \
-DMACOSX_DEPLOYMENT_TARGET=10.15 \
[torch-mlir earthmoving (2/N)] Python code movement. This moves the bulk of the Python code (including the Torch interop) from `frontends/pytorch` into `torch-mlir/TorchPlugin`. This also required reconciling a bunch of other Python-related stuff, like the `torch` dialects. As I did this, it was simpler to just remove all the old numpy/basicpy stuff because we were going to delete it anyway and it was faster than debugging an intermediate state that would only last O(days) anyway. torch-mlir has two top-level python packages (built into the `python_packages` directory): - `torch_mlir_dialects`: `torch` dialect Python bindings (does not depend on PyTorch). This also involves building the aggregate CAPI for `torch-mlir`. - `torch_mlir`: bindings to the part of the code that links against PyTorch (or C++ code that transitively does). Additionally, there remain two more Python packages in npcomp (but outside `torch-mlir`): - `npcomp_torch`: Contains the e2e test framework and testing configs that plug into RefBackend and IREE. - `npcomp_core`: Contains the low-level interfaces to RefBackend and IREE that `npcomp_torch` uses, along with its own `MLIR_PYTHON_PACKAGE_PREFIX=npcomp.` aggregation of the core MLIR python bindings. (all other functionality has been stripped out) After all the basicpy/numpy deletions, the `npcomp` C++ code is now very tiny. It basically just contains RefBackend and the `TorchConversion` dialect/passes (e.g. `TorchToLinalg.cpp`). Correspondingly, there are now 4 main testing targets paralleling the Python layering (which is reflective of the deeper underlying dependency structure) - `check-torch-mlir`: checks the `torch-mlir` pure MLIR C++ code. - `check-torch-mlir-plugin`: checks the code in `TorchPlugin` (e.g. TorchScript import) - `check-frontends-pytorch`: Checks the little code we have in `frontends/pytorch` -- mainly things related to the e2e framework itself. - `check-npcomp`: Checks the pure MLIR C++ code inside npcomp. There is a target `check-npcomp-all` that runs all of them. The `torch-mlir/build_standalone.sh` script does a standalone build of `torch-mlir`. The e2e tests (`tools/torchscript_e2e_test.sh`) are working too. The update_torch_ods script now lives in `torch-mlir/build_tools/update_torch_ods.sh` and expects a standalone build. This change also required a fix upstream related to cross-shlib Python dependencies, so we also update llvm-project to 8dca953dd39c0cd8c80decbeb38753f58a4de580 to get https://reviews.llvm.org/D109776 (no other fixes were needed for the integrate, thankfully). This completes most of the large source code changes. Next will be bringing the CI/packaging/examples back to life.
2021-09-11 02:44:38 +08:00
-DLLVM_TARGETS_TO_BUILD=host
- name: llvm-source-out-of-tree-torch-src-or-binary
2022-04-19 22:07:25 +08:00
# This build takes a while but is expected to almost always be cached.
# A cache invalidation occurs when the committed LLVM version is changed.
if: matrix.llvmtype == 'source'
2022-04-04 23:20:56 +08:00
run: |
cd $GITHUB_WORKSPACE
cmake -Bllvm-build -GNinja \
-DCMAKE_BUILD_TYPE=Release \
-DCMAKE_LINKER=lld \
-DCMAKE_C_COMPILER_LAUNCHER=ccache -DCMAKE_CXX_COMPILER_LAUNCHER=ccache \
-DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ \
-DPython3_EXECUTABLE=$(which python) \
-DLLVM_ENABLE_ASSERTIONS=ON \
-DLLVM_ENABLE_PROJECTS=mlir \
-DMLIR_ENABLE_BINDINGS_PYTHON=ON \
-DLLVM_TARGETS_TO_BUILD=host \
externals/llvm-project/llvm
cmake --build llvm-build
2022-04-04 23:20:56 +08:00
cmake -GNinja -Bbuild \
-DCMAKE_LINKER=lld \
-DCMAKE_C_COMPILER_LAUNCHER=ccache -DCMAKE_CXX_COMPILER_LAUNCHER=ccache \
-DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ \
-DMLIR_DIR="$(pwd)/llvm-build/lib/cmake/mlir/" \
-DLLVM_DIR="$(pwd)/llvm-build/lib/cmake/llvm/" \
-DMLIR_ENABLE_BINDINGS_PYTHON=OFF \
-DTORCH_MLIR_ENABLE_MHLO=ON \
-DTORCH_MLIR_USE_INSTALLED_PYTORCH=${{ matrix.torch-binary }} \
2022-04-04 23:20:56 +08:00
-DPython3_EXECUTABLE=$(which python) \
.
- name: Build and run check-torch-mlir-all
if: matrix.llvmtype == 'binary'
run: |
cd $GITHUB_WORKSPACE
export PYTHONPATH="$GITHUB_WORKSPACE/build/tools/torch-mlir/python_packages/torch_mlir"
cmake --build build --target check-torch-mlir-all
- name: RefBackend - TorchScript end-to-end tests
if: matrix.llvmtype == 'binary'
run: |
cd $GITHUB_WORKSPACE
export PYTHONPATH="$GITHUB_WORKSPACE/build/tools/torch-mlir/python_packages/torch_mlir"
python -m e2e_testing.torchscript.main --config=refbackend -v
- name: EagerMode - TorchScript end-to-end tests
if: matrix.llvmtype == 'binary'
run: |
cd $GITHUB_WORKSPACE
export PYTHONPATH="$GITHUB_WORKSPACE/build/tools/torch-mlir/python_packages/torch_mlir"
python -m e2e_testing.torchscript.main --config=eager_mode -v
- name: TOSA backend - TorchScript end-to-end tests
if: matrix.llvmtype == 'binary'
run: |
cd $GITHUB_WORKSPACE
export PYTHONPATH="$GITHUB_WORKSPACE/build/tools/torch-mlir/python_packages/torch_mlir"
python -m e2e_testing.torchscript.main --config=tosa -v
- name: Lazy Tensor Core - TorchScript end-to-end tests
if: matrix.llvmtype == 'binary'
run: |
cd $GITHUB_WORKSPACE
export PYTHONPATH="$GITHUB_WORKSPACE/build/tools/torch-mlir/python_packages/torch_mlir"
python -m e2e_testing.torchscript.main --config=lazy_tensor_core -v