Skip to content

installation.md

Latest commit

 

History

History
298 lines (179 loc) · 7.92 KB

installation.md

File metadata and controls

298 lines (179 loc) · 7.92 KB

Installation

The following pre-requisite libraries and software are required in order to run IMPaCT properly:

  • Python3
  • CMake
  • Boost
  • AdaptiveCPP (with clang++ and OpenMP for compiler and parallelization)
  • HDF5
  • Armadillo (including enabling HDF5 usage)
  • GSL
  • NLopt
  • GLPK
  • optional GPU specific extras, e.g. llvm, cuda, etc.

Most of these can be installed simply using either sudo apt-get install or sudo brew install for Linux or MacOS respectively. Some have been provided in zip files in this repository for ease of installation. Due to updates to specific different libraries and tools, please check online if any of the below installation points no longer work and we will try and update them quickly.

Inside this repository you can find a Dockerfile with IMPaCT pre-installed which is the easiest way to use the tool. Additionally we have included an install script which automatically installs IMPaCT and has been tested to work on Ubuntu 22.

MacOS

I have personally got IMPaCT running on MacOS with an Intel chip, for the newer M1 and M2 chips you may need to see the AdaptiveCPP page here to see the support.

Open your machines terminal, not the the terminal of an application e.g. PyCharm.

If you have never run or installed code on your machine before then its a good idea to install XCode and this will download many packages automatically.

Python3

Check if already installed: python3 --version

Install: brew install python

CMake

Check if already installed: cmake --version

Install: brew install cmake

Boost

Check if already installed: which boost

Install: brew install boost

AdaptiveCPP

See the most up to date installation instructions here.

either clone from AdaptiveCPP GitHub or unzip the folder directly from this repository here: unzip AdaptiveCpp-develop.zip

go into the directory: cd AdaptiveCpp-develop

build the files sudo cmake . (note this dot is important to signify the current folder)

If errors occur then adding flags will help with the compilation such as:

sudo cmake -DCMAKE_INSTALL_PREFIX=/usr/local -DCMAKE_CXX_COMPILER=/path/to/clang++ -DOpenMP_ROOT=/path/to/libomp/include

You may need to find these files yourself to link appropriately.

install the files: sudo make install

HDF5

Check if already installed: h5dump --version

Install: brew install hdf5

Armadillo

should be able to install everything with: brew install armadillo

Important! Enable HDF5 usage in Armadillo by going to the folder armadillo_bits where Armadillo is installed:

e.g. cd usr/local/../armadillo_bits

you can find the location of this file using (if multiple locations exist you can try all of them to be safe!):

sudo find / -type d -name 'armadillo_bits'

open config file:

sudo vi config.hpp

uncomment the line ARMA_USE_HDF5 by removing the # symbol, press i to edit the file!

save and exit the file: Esc then :x and enter

GSL

Install: brew install gsl

NLopt

Install: brew install nlopt

GLPK

Install: brew install glpk

Linux

Ubuntu (tested on Ubuntu 22)

For your local machine if it runs ubuntu (tested on ubuntu 22, but may work for some other linux systems too) an installation script install_ubuntu22.sh can be easily used by running in the terminal the commands:

chmod +x install_ubuntu22.sh to make the file executable followed by sudo ./install_ubuntu22.sh to install all the packages and dependencies.

Other Linux Machines

Open your machines terminal, not the the terminal of an application e.g. PyCharm.

If you have never run code on your PC before, it's a good idea to run sudo apt-get install build-essential which will download many essential packages for your PC.

Python3

Check if already installed: python3 --version

Install (can replace star with version number, e.g. 8): sudo apt-get install python 3.*

CMake

Check if already installed: cmake --version

Install: sudo apt-get install cmake

Boost

Check if already installed: dpkg -s libboost-dev | grep 'version'

Install: sudo apt-get install libboost-all-dev

AdaptiveCPP

See the most up to date installation instructions here, this installation can be tricky and so there much more info on the official page. The most common problem that is faced is with correctly linking AdaptiveCpp with the correct installed files and compilers.

either clone from AdaptiveCPP GitHub or unzip the folder directly from this repository here: unzip AdaptiveCpp-develop.zip

go into the directory: cd AdaptiveCpp-develop

you may need to install clang and openMP: sudo apt install -y libclang-16-dev clang-tools-16 libomp-16-dev

build the files: sudo cmake . (note this dot is important to signify the current folder)

if issues occur add some flags such as -DCMAKE_CXX_COMPILER=/path/to/clang++-16 can help with the correct compilation, check the official page for more details.

install the files sudo make install

HDF5

Check if already installed: dpkg -s libhdf5-dev

Install: sudo apt-get install libhdf5-serial-dev

Armadillo

install pre-requisites: sudo apt install libopenblas-dev, liblapack-dev, libarpack2-dev, libsuperlu-dev

either download armadillo here or unzip the folder: tar -xvf armadillo-12.6.4.tar.xz

go into the directory: cd armadillo-12.6.4

build the files: sudo cmake .

install: sudo make install

Important! Enable HDF5 usage in Armadillo by going to the folder armadillo_bits where Armadillo is installed:

e.g. cd usr/local/../armadillo_bits

you can find the location of this file using (if multiple locations exist you can change all of them to be safe!):

sudo find / -type d -name 'armadillo_bits'

open config file: sudo vi config.hpp

uncomment the line ARMA_USE_HDF5 by removing the # symbol, press i to edit the file!

save and exit the file: Esc then :x and enter

GSL

either download GSL here or unzip the file: tar -xzf gsl-2.7.1.tar.gz

configure the files: ./configure

build the files: make

install the files: sudo make install

NLopt

either download NLopt here or unzip the file: tar -xzf nlopt-2.7.1.tar.gz

create a folder for the build: mkdir build

go into the folder: cd build

build the files: cmake ..

make the files: make

install the files: sudo make install

GLPK

Check if already installed: dpkg -l | grep glpk

Install: sudo apt-get install glpk-utils libglpk-dev glpk-doc

Windows

The AdaptiveCPP community have managed to get their software working for Windows, but we have not attempted it ourselves. Feel free to try IMPaCT on Windows and share with us your experience!

GPU

Check AdaptiveCPP instructions for different GPU cards and operating systems here.

Below are some basic commands for GPU installation for Linux (on top of the previous installation instructions).

Find your GPU card using

lspci

To install LLVM

llvm-config --version

sudo apt-install clang

wget https://apt.llvm.org/llvm.sh

chmod +x llvm.sh

sudo ./llvm.sh 16

sudo apt install -y libclang-16-dev clang-tools-16 libomp-16-dev llvm-16-dev

To install CUDA

nvcc --version

sudo apt install nvidia-cuda-toolkit

Add Flags to AdaptiveCPP installation

go into directory

sudo cmake . -DCUDAToolkit_LIBRARY_ROOT=/usr/lib/cuda

sudo make install