ONNX To LLVM Pass Backend using linalg #3349

Open

kimm240 wants to merge 44 commits intoonnx:mainfrom

kimm240:feature/linalg-to-llvm-pipeline

Commits on Dec 15, 2025

Add ONNXToLinalg conversion pass with MatMul support
hyun gyu kim
committed
Add ONNX to Linalg lowering pipeline
hyun gyu kim
committed
Add bufferization to ONNX to Linalg pipeline
hyun gyu kim
committed
Add --use-linalg-path flag to choose between Krnl and Linalg pipelines
hyun gyu kim
committed
Split addONNXToLinalgPasses into addLinalgToAffinePasses and addLinalgToLLVMPasses
hyun gyu kim
committed
Refactor addPasses: translate comments to English and clarify pass execution order
hyun gyu kim
committed
Fix pass execution order: ensure addONNXToMLIRPasses is called for Linalg path
hyun gyu kim
committed
Refactor Linalg lowering pipeline: simplify to use pm.addPass only
hyun gyu kim
committed
Analyze PassManager execution mechanism and fix pass ordering
hyun gyu kim
committed
Try different pass ordering: convert-linalg-to-loops before bufferization
hyun gyu kim
committed
Register BufferizableOpInterface for arith, linalg, and tensor dialects
hyun gyu kim
committed
Register BufferizableOpInterface for func dialect
hyun gyu kim
committed
Add one-shot-bufferize to Linalg pipeline and fix pass ordering
hyun gyu kim
committed
Refactor Linalg lowering pipeline to match Krnl structure
hyun gyu kim
committed
Add comparison test between Krnl and Linalg pipelines
hyun gyu kim
committed
Remove onnx.EntryPoint before LLVM conversion in Linalg path
hyun gyu kim
committed
Apply coding practices: Use C++ style casting
hyun gyu kim
committed
Remove debug logs and apply clang-format
hyun gyu kim
committed
Apply clang-format to ConvertONNXToLinalg.cpp
hyun gyu kim
committed
Refactor Linalg to LLVM pipeline to use createConvertKrnlToLLVMPass
hyun gyu kim
committed
Add analysis of signature generation issue in Linalg pipeline
hyun gyu kim
committed
Fix ONNXEntryPointLowering pattern in Linalg pipeline
hyun gyu kim
committed
Translate comments in driver.cpp to English
hyun gyu kim
committed
Add test driver for Linalg pipeline MatMul execution
hyun gyu kim
committed
Refactor ONNXEntryPointLowering to use existing implementation
hyun gyu kim
committed
Apply clang-format to modified files
hyun gyu kim
committed
Clean up conversion directory
hyun gyu kim
committed
Remove LINALG_PIPELINE_PASSES.md documentation file
hyun gyu kim
committed
Clean up directory structure
Hyun Gyu Kim
committed
Clang Format
Hyun Gyu Kim
committed
Support empty inputs to the compiled model and remove input arguments in check-onnx-backend-constant (#3344)

tungld
authored and
Hyun Gyu Kim
committed
Make dynamic dimension analysis for Reshape more general (#3341)

authored and
Hyun Gyu Kim
committed
added comment to script and doc (#3312)

AlexandreEichenberger
authored and
Hyun Gyu Kim
committed
Fix a bug in saving a json file and a bug in onnx to zhigh for quantization (#3340)

authored and
Hyun Gyu Kim
committed
Make verifyInputTensor true by default (#3079)

chentong319
authored and
Hyun Gyu Kim
committed
fix osx build for deprecation of git image (#3348)

Sunny-Anand
authored and
Hyun Gyu Kim
committed
Add ONNXToLinalg conversion pass with MatMul support (#3343)

committed
Clang Format
Hyun Gyu Kim
committed
delete test file that doesn't have RUN script
Hyun Gyu Kim
committed
Chore: trigger ci
Hyun Gyu Kim
committed
Chore: trigger ci
Hyun Gyu Kim
committed
Chore: trigger ci
Hyun Gyu Kim
committed
Chore: trigger ci
Hyun Gyu Kim
committed
Merge branch 'main' into feature/linalg-to-llvm-pipeline
kimm240
authored