| commit | e3f2d47cd7ae68d3c4ee68d064d1be725d5a6d55 | [log] [tgz] |
|---|---|---|
| author | Vivek Khandelwal <vivekkhandelwal1424@gmail.com> | Wed Oct 23 18:33:01 2024 +0530 |
| committer | GitHub <noreply@github.com> | Wed Oct 23 18:33:01 2024 +0530 |
| tree | 3f80db3130941381e26d3dd05c67d76f9f664a0b | |
| parent | 81c8b257cbd8e5b51517e9c696210a71945f4d4b [diff] |
Bump torch-mlir to 140cad5 and update TorchOnnxToTorch conversion pipeline (#18867) This commit bumps torch-mlir to https://github.com/llvm/torch-mlir/commit/140cad5659bb779bb1f5de1888566db5b5d21236. This commit also replaces the `TorchOnnxToTorchPass` used to convert torch-onnx IR to torch IR with `TorchOnnxToTorchBackendPipeline` which was introduced here: https://github.com/llvm/torch-mlir/commit/fa4794dae2057876ec8ad2a6464e2668f6a2ea0c. This pipeline consists of passes like `TorchOnnxToTorch`, `ScalarizeShapes`, `DecomposeComplexOps`, `ShapeRefinementPipeline`, etc., to convert the torch-onnx IR to torch IR consistent with the other lowering paths for torch backend IR. With the changes made in this PR, some of the tests which were failing earlier during compilation now passes compilation, some of them even passes inference, while there are some tests which were passing earlier are now failing. The list is as follows: - For tests/external/iree-test-suites/onnx_ops/onnx_ops_cpu_llvm_sync.json: a.) Earlier failing compilation, now passing inference: - onnx/node/generated/test_convtranspose_kernel_shape - onnx/node/generated/test_einsum_sum b.) Earlier failing compilation, now passing compilation but failing inference - onnx/node/generated/test_convtranspose_output_shape c.) Earlier passing but now failing inference - onnx/node/generated/test_scan_sum - onnx/node/generated/test_scan9_sum - For tests/external/iree-test-suites/onnx_ops/onnx_ops_gpu_rocm_rdna3.json: a.) Earlier failing compilation, now passing inference: - onnx/node/generated/test_convtranspose_kernel_shape - onnx/node/generated/test_einsum_sum b.) Earlier failing compilation, now passing compilation but failing inference - onnx/node/generated/test_convtranspose_output_shape c.) Earlier passing but now failing inference - onnx/node/generated/test_scan_sum - onnx/node/generated/test_scan9_sum - onnx/node/generated/test_slice_default_axes - For tests/external/iree-test-suites/onnx_ops/onnx_ops_gpu_vulkan.json: a.) Earlier failing compilation, now passing inference: - onnx/node/generated/test_convtranspose_kernel_shape b.) Earlier failing compilation, now passing compilation but failing inference - onnx/node/generated/test_convtranspose_output_shape - onnx/node/generated/test_einsum_sum c.) Earlier passing but now failing inference - onnx/node/generated/test_scan_sum - onnx/node/generated/test_scan9_sum This commit https://github.com/llvm/torch-mlir/commit/55ff110dc29cab7e2495ccdbec9a60512c29c665 is expected to fix the newly introduced failures which will be included in IREE in the next Torch-MLIR bump. --------- Signed-off-by: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
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
| Package | Release status |
|---|---|
| GitHub release (stable) |  |
| GitHub release (nightly) |  |
| Python iree-compiler |  |
| Python iree-runtime |  |
| Host platform | Build status |
|---|---|
| Linux |   |
| macOS |  |
| Windows |  |
For the full list of workflows see https://iree.dev/developers/general/github-actions/.
See our website for more information.
Community meeting recordings: IREE YouTube channel
IREE is licensed under the terms of the Apache 2.0 License with LLVM Exceptions. See LICENSE for more information.