LER-IR is an frontenc compiler and MLIR dialect for compiling LER notation into executables. It is based of the paper: ‘GLORE: Generalized Loop Redundancy Elimination upon LER-Notation’ (Shen, Ding 2017).
To get started with this project, ensure you have the following installed on your system:
- CMake >= v3.22
- LLVM/MLIR Libraries == v18.1.6 (19.1.4 Produces bugs)
- Instructions for build LLVM with MLIR can be found here.
- Ninja build system, this is optional but recommended.
The LER optimizer in the optimizer directory requires the Java ANTLR parser library to be compiled:
cd /usr/local/lib
sudo curl -O https://www.antlr.org/download/antlr-4.13.1-complete.jar
export CLASSPATH=".:/usr/local/lib/antlr-4.13.1-complete.jar:$CLASSPATH"
git clone https://github.com/bradenhelmer/LER-IR
mkdir build
cmake -S . -B build
cmake --build build
git clone https://github.com/bradenhelmer/LER-IR
mkdir build
cmake -S . -G Ninja -B build
cmake --build build
An executable named ler-compile
will be in the build
directory.