KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#developer-tools #codex #typescript #ai-agents #claude-code #dsh #dsh-plugin #cli #deepseek-harness #macos #ai-agent #open-source

neura

★ 34 Open GitHub ↗

A dataflow compiler

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Neura -- a dataflow dialect.

Build LLVM & Neura
--------------------------------------------------------
 - Clone llvm-project.

- Clone this repo.

- Build LLVM:
   - Check out to commit `6146a88` (Our project is based on this version of LLVM).
   - Build:
```sh
 $ mkdir build && cd build
 # May need install ccache and lld.
 $ cmake -G Ninja ../llvm \
     -DLLVM_ENABLE_PROJECTS="mlir;clang" \
     -DLLVM_BUILD_EXAMPLES=OFF \
     -DLLVM_TARGETS_TO_BUILD="Native" \
     -DCMAKE_BUILD_TYPE=Release \
     -DLLVM_ENABLE_ASSERTIONS=ON \
     -DCMAKE_C_COMPILER=clang \
     -DCMAKE_CXX_COMPILER=clang++ \
     -DCMAKE_CXX_STANDARD=17 \
     -DCMAKE_CXX_FLAGS="-std=c++17 -frtti" \
     -DLLVM_ENABLE_LLD=ON \
     -DMLIR_INSTALL_AGGREGATE_OBJECTS=ON \
     -DLLVM_ENABLE_RTTI=ON \
     -DCMAKE_C_COMPILER_LAUNCHER=ccache \
     -DCMAKE_CXX_COMPILER_LAUNCHER=ccache
 $ cmake --build . --target check-mlir
 $ cmake --build . --target check-clang
```

- Build Neura:
   - Step into this repo, note that this repo is outside of llvm-project.
   - Build:
```sh
 $ mkdir build && cd build
 # Replace the path "/workspace" accordingly.
 $ cmake -G Ninja .. \
     -DLLVM_DIR=/workspace/llvm-project/build/lib/cmake/llvm \
     -DMLIR_DIR=/workspace/llvm-project/build/lib/cmake/mlir \
     -DMLIR_SOURCE_DIR=/workspace/llvm-project/mlir \
     -DMLIR_BINARY_DIR=/workspace/llvm-project/build \
     -DCMAKE_CXX_FLAGS="-std=c++17"
 $ ninja
```

- Test:
```sh
 $ cd ../test
 # a test with detailed middle output
 $ ../build/tools/mlir-neura-opt/mlir-neura-opt --debug test.mlir

# Or test with lit:
 $ /workspace/llvm-project/build/bin/llvm-lit test.mlir -v
```

Sync `test/e2e` outputs into Zeonica_Testbench (submodule)
--------------------------------------------------------
This repo vendors [`sarchlab/Zeonica_Testbench`](https://github.com/sarchlab/Zeonica_Testbench.git) as a git submodule at `test/benchmark/Zeonica_Testbench`.

After running e2e compilation/tests, you can sync the generated artifacts into the testbench repo:
```sh
$ ./tools/sync_e2e_outputs_to_zeonica_testbench.sh
```

Mapping (per kernel `K`):
- `test/e2e/K/tmp-generated-dfg.{dot,yaml}` -> `test/benchmark/Zeonica_Testbench/kernel/K/K-dfg.{dot,yaml}`
- `test/e2e/K/tmp-generated-instructions.{asm,yaml}` -> `test/benchmark/Zeonica_Testbench/kernel/K/K-instructions.{asm,yaml}`

Contributing
--------------------------------------------------------
Please refer to the [Contributing Guide](https://github.com/coredac/dataflow?tab=contributing-ov-file#contributing-to-neura) for code style, formatting, and contribution workflow.

neura

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README