Alkali

KO | EN

gitlite — search

#llm #ai #python #typescript #nextjs #javascript #plugin #mcp #tailwindcss #minecraft #rust #fastapi

★ 29 Open GitHub ↗

Alkali is a MLIR-based compiler infrastructure for SmartNICs. It allows developers to write target-independent programs, with the compiler automatically managing cross-NIC porting and performance optimization.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Portable and High-Performance SmartNIC Programs with Alkali

![Status](https://img.shields.io/badge/Version-Experimental-green.svg)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)

Alkali is a compiler infrastructure for SmartNICs, delivering both functional and performance portability across a wide range of SmartNIC hardware. It centers on a unified intermediate representation (IR), a common set of optimization and transformation passes, and an automated network-application parallelization pipeline.

Currently, this repo contains Alkali frontend for C and αIR, the definition and implementation of IR, optimization & transformation passes, and code generation for following backends: Verilog(FPGA), MicroC(Netronome), and LLVM(ARM DPDK/RiscV).

- [Portable and High-Performance SmartNIC Programs with Alkali](#portable-and-high-performance-smartnic-programs-with-alkali)
  - [Building](#building)
    - [Building - Alkali Compiler Build](#building---alkali-compiler-build)
  - [Running Alkali Example](#running-alkali-example)
  - [Reference](#reference)
  - [Contact](#contact)

## Building

Alkali is built and tested on Ubuntu 22.04.2 LTS. Alkali depends on Boost and Z3. You need to install the external dependencies first, and build MLIR libraries using the following command:

```sh
# current z3 version is Z3 4.8.7
sudo apt install build-essential libz3-dev ninja-build libboost-dev cmake

# build MLIR
cd llvm-project
git apply patches/translateToCpp.patch
mkdir build && build
cmake -G Ninja ../llvm \
 -DLLVM_ENABLE_PROJECTS="mlir;clang" \
 -DLLVM_TARGETS_TO_BUILD="AArch64;AMDGPU;ARM;RISCV;X86" \
 -DLLVM_ENABLE_ASSERTIONS=ON \
 -DCMAKE_BUILD_TYPE=Release 
ninja
```

### Building - Alkali Compiler Build

This setup assumes that you have built LLVM and MLIR in `$BUILD_DIR` and installed them to `$PREFIX`. To build and launch the tests, run
```sh
mkdir build && cd build
 cmake -G Ninja .. \
    -DLLVM_EXTERNAL_LIT=$PWD/../llvm-project/build/bin/llvm-lit \
    -DMLIR_DIR=$PWD/../llvm-project/build/lib/cmake/mlir \
    -DCLANG_DIR=$PWD/../llvm-project/build/lib/cmake/clang \
    -DCMAKE_BUILD_TYPE=Debug
cd .. &&  ninja -C build/
```

Alkali binaries, `ep2c` and `ep2c-opt`, will be generated `build/bin`.

## Running Alkali Example

The source files are located in `tests`.
Alkali framework will first convert the input file (C or αIR) into MLIR representation, and then apply optimization, mapping and code generation.

After that is built, you run example with
```sh
./scripts/alkalic tests/experiments_c/nfchain.c --target rtl
```
This will compile the nfchain example for FPGA NICs. The generated Verilog code can be found in `./fpga_out`. It transforms the oringinal RTC handler into pipeline and data parallel FPGA codes.

Similarly, if you want to compile the nfchain example for Agilio NICs, run:
```sh
./scripts/alkalic tests/experiments_c/nfchain.c --target netronome
```
The generated Agilio Micro C code can be found in `./netronome_out`.

More example applications can be found in ./tests/

## Reference

If you use Alkali in your research or projects, please cite our paper [Portable and High-Performance SmartNIC Programs with Alkali](https://www.usenix.org/conference/nsdi25/presentation/lin-jiaxin):

```
@inproceedings{alkali,
  title={Portable and High-Performance SmartNIC Programs with Alkali},
  author={Lin, Jiaxin and Guo, Zhiyuan and Shah, Mihir and Ji, Tao and Zhang, Yiying and Kim, Daehyeok and Akella, Aditya}
  booktitle={22st USENIX Symposium on Networked Systems Design and Implementation (NSDI 25)},
  year={2025}
}
```

## Contact

If you encounter any issues, want to report a bug, or request a new feature, please open an issue on this repository.
For additional assistance, you can also reach out to
`z9guo [at] ucsd [dot] edu` and `jxlin [at] utexas [dot] edu`.

Alkali

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README