ALF/INSTALL at master · RyanLeeHayes/ALF · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
Overview

In order to install ALF, you will need an environment with C, C++, MPI, and CUDA
compilers as well as cmake and python and pip. You can either to a manual
installation, setting up these libraries yourself, for example by loading
appropriate modules, and then install in a python virtual environment, OR you
can do a conda installation, where you install these libraries with a writeable
conda installation, and then pip install ALF into the active conda environment.
In both cases, cmake manages compilation of executables and automatically
performs pip installation in the active conda environment.


Installing Manually

Make sure paths to C, C++, MPI, CUDA, and cmake are in your environment. If you
are working on a cluster, this might be achievable with appropriate module load
commands. For example at hpc3.rcic.uci.edu the following modules work

module load cmake/3.20.5 cuda/11.7.1 gcc/11.2.0 openmpi/4.1.2/gcc.11.2.0 fftw/3.3.10/gcc.11.2.0-openmpi.4.1.2
module load anaconda

Then perform the following commands to create a new python virtual environment
named env-alf and build a copy of ALF in a directory named build

rm -r build
mkdir build
cd build
python -m venv env-alf
source env-alf/bin/activate
cmake .. -DCMAKE_INSTALL_PREFIX=`pwd`
make install

Any time you wish to use ALF in this environment, you should source the activate
command for this enviroment. For example, to activate with a relative path from
the ALF directory run

source build/env-alf/bin/activate


Installing with conda

With conda, you must have permission to write to the conda installation,
otherwise you should create a python virtual enviroment based on the conda
installation and proceed as above. If you need a writeable conda enviroment you
may follow instructions at

https://github.com/BrooksResearchGroup-UM/MSLD-Workshop/tree/main/0Install_Tools/Linux

If you have the conda enviroment set up you may activate the conda environment
in which you wish to install ALF and proceed as follows.

mkdir build
cd build
cmake .. -DCMAKE_INSTALL_PREFIX=$CONDA_PREFIX
make install


Testing

As a first test of whether your installation was successful, you can run

python -c "import alf"


Installing without CUDA

Installing without CUDA is not currently supported. The code that used to
support it is still present, but no longer compatible with the rest of ALF.
Contact the developers if this functionality is needed.