source: cardio/

Cardio is the HPC facility at the CEU.

CSD3 partition

scontrol show part
# cpu-o-1 for interactive jobs with no limit on cpus per user expecting fair usage
# cpu-o-[2-5] for long jobs with maximum cpus per user set to 40 and a maximum wall time of 7 days
# cpu-o-7 for short jobs with a maximum wall time of 24 hours

# Interactive jobs

#SBATCH -p cardio_intr
#SBATCH --qos=cardio_intr

# Long jobs

#SBATCH -p cardio
#SBATCH --qos=cardio

# Short jobs

#SBATCH -p cardio_short
#SBATCH --qos=cardio_short

Legacy materials


Automatic login can be enabled with ssh-copy-id


It is possible to fetch my-file-on-Cardio with

scp$USER/my-file-on-Cardio .

as with sftp, noting the -r option for both could be very useful. More generally, it is preferable to use rsync,

Old version on Cardio, rsync 3.0.6, gives errors,

rsync -av --partial mydir/$USER/hpc-work/mydir

Errors after successfully sending a number of files :
rsync: writefd_unbuffered failed to write 4 bytes to socket [sender]: Broken pipe (32)
rsync: connection unexpectedly closed (604 bytes received so far) [sender]
rsync error: error in rsync protocol data stream (code 12) at io.c(600) [sender=3.0.6]

Bram Prins ( has the latest version (3.1.3) that doesn't give this error here:


Note /scratch at Cardio is now /DO-NOT-MODIFY-SCRATCH – an example is

R package reinstallations

Package redeployment is illustrated with R below for building R package list from /home/$USER/R at Cardio to be resintalled to /rds/user/$USER/hpc-work/R at CSD3.

We can use screen copy of package list from Cardio since users do not have write permission.

On Cardio

  home <- Sys.getenv("HOME")
  from <- paste0(home,"/R")
  pkgs <- unname(installed.packages(lib.loc = from)[, "Package"])


  pkgs <- "mark and copy the list as given in c() above with Shift+Ins"
  user <- Sys.getenv("USER")
  to <- paste0("/rds/user/",user,"/hpc-work/R")
  install.packages(pkgs, lib=to, repos="")

If a package list is already built one can reinstall them as follows.

user <- Sys.getenv("USER")
location <- paste0("/rds/user/",user,"/hpc-work/R")
pkgs <- unname(installed.packages(lib.loc = location)[, "Package"])
install.packages(pkgs, lib=location, repos="")

A version by Scott Ritchie (, reinstall_r_pkgs.R, also touches upon Bioconductor, whose package installations and updates are described at