Installation of COMEDI

Precompiled binaries

Before of installing

Standalone version of COMEDI are available for Windows, Linux & Mac OS (only for 64 bit OS). With the standalone executable, there is no need have Matlab installed.

It is recommended to install COMEDI on a computer with high-performance specs:
  1. multi-core CPU

  2. RAM > 16 GB

  3. dedicated graphics cards / GPU

If you already tried to run the executable ‘COMEDI’ before having read this part of the manual, you may have encountered the error message.

The standalone version of COMEDI is currently being developed in Matlab R2021b. The executable, ‘COMEDI’, only works after you installed the Matlab Common Runtime (MCR) libs of that Matlab version – you can download the one your OS of interest here:

Windows

https://www.comedi.io/matlab/64/MCR_R2021b_win64_installer.zip

Linux

https://www.comedi.io/matlab/64/MCR_R2021b_glnxa64_installer.zip

MacOS

https://www.comedi.io/matlab/64/MCR_R2021b_maci64_installer.zip

Download and install

You can download the precompiled binaries from

Binaries

https://www.comedi.io/download

To install the MCR libs file, simply follow the instructions.Note that MCR libraries from other Matlab versions are probably not compatible with these standalone versions of COMEDI.Again, notethat you do NOT need a Matlab license to run COMEDI! For Linux/Mac, if you choose an installation directory for the MCR that is different from the default one, you need to use this path in the “run_COMEDI.sh” as well

Release/Preview source

Release

https://github.com/diffusionmri/COMEDI/releases

Release

https://github.com/diffusionmri/comedi

  1. Clone the COMEDI repository:

$ git clone https://github.com/diffusionmri/COMEDI.git

or download the released COMEDI:

$ wget https://github.com/diffusionmri/COMEDI/archive/refs/tags/v1.0.zip
  1. Configure the COMEDI install in Matlab:

>> cmdInstall
  1. Keeping COMEDI up to date

You can update your installation at any time by opening at Matlab in the COMEDI folder, and typing:

>> cmdUpdate

If this does’t work immediately, it may be that you need to re-setup the configure script:

>> cmdInstall

HPC cluster

These instructions outline a few issues specific to high-performance computing (HPC) systems.

Without more specific! With HCP cluster, it is recommended to add a few line code in your job to be submission, for example

$ export SLURM_JOBTMP=/path/to/tmp/folder
$ export MATLAB_PREFDIR=$(mktemp -d $SLURM_JOBTMP/matlab-XXXX)
$ echo "Hostname: $(hostname)"
$ matlab -nodesktop -nosplash -nodisplay -r "yourfunction();exit;"