Machine Learning : regression tool

This project has been developped in order to predict the performance of the DGETRF core from the LAPACK library

Build with docker :

Docker :

You will need docker installed on your machine

On Debian/Ubuntu with apt :

$ sudo apt-get install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin

#Make sure install went right :
$ sudo docker run hello-world

Docker documentation is available here

Build the project :

#Clone this repository
$ git clone https://github.com/Nxirda/PPN_Projet_MEM.git

#First build the image, once in the project directory :
$ docker build . -t project_mem 

#Then start the container, the project will be built at the same time
$ docker run --rm -it project_mem

Then you can skip to the Usage section below

Build : Standard way

Dependencies :

You will need :

Gtest :

sudo apt install libgtest-dev

MPI :

Boost :

Build :

$ git clone https://github.com/Nxirda/PPN_Projet_MEM.git
$ cd PPN_Projet_MEM && mkdir build
$ cd build && cmake ..
$ make -j

If your installation of MPI isnt in your path you can run the following command instead of "cmake .."

$ cmake -DCMAKE_CXX_COMPILER=/path/to/your/mpi/c++/compiler -DCMAKE_PREFIX_PATH=/path/to/your/mpi/installation ..

$ cmake -S . -B <BUILD_DIR> <CMAKE OPTIONS>

Usage :

# Ensure the project has been built properly
cd build
Usage is : ./lngbm [-h] 

Tests :

If you want to make sure everything has been done properly you can run the tests :

• Refer to the Readme in the test directory

Benchmarks :

If you want to benchmark your model you can run the benchmarks :

• Refer to the Readme in the Benchmark directory

Authors :

Adrien Henrot

Thanks to

Mathys Jam, for guiding us through the project