Freie Universität Berlin Curta Configuration
NB: In order to run pipelines using this HPC cluster, you must first apply for access here.
Job limits are capped at 32 CPUs, 772 GB and a maximum run time of 14 days.
This profile is configured to run with Apptainer, which is pre-loaded and globally available to all users on the cluster.
First, Nextflow should be loaded into your interactive session or added to your Slurm script using:
module load NextflowNote: There exists different version modules for Nextflow on the cluster. This command ensures loading the latest version.
Using -profile fub_curta will download the config file, which has been pre-configured with a setup suitable for the Curta cluster.
The default partition for job submissions is main. The user has access to 166 nodes on main. The profile automatically detects whether the user belongs to the BeGenDiv working group, and gives the user access to 4 more nodes by making use of the begendiv partition together with main, giving precedence to the former.
In addition, job submissions are assigned the appropriate quality of service (QOS) as such:
hipriofortask.time <= 3.hpriofortask.time <= 3.dstandardfortask.time <= 14.d
Institute-specific pipeline profiles exists for:
debug profile
Deactivating the cleanup of intermediate files is also possible. It is done by specifying -profile fub_curta,debug. This simply turns off automatic clean up of intermediate files, which can be useful for debugging.
Publications
If computing time on Curta has contributed to a publication or the completion of a degree course, please send us the corresponding DOI or BibTeX entry and, if possible, an appropriate image to hpc@zedat.fu-berlin.de. This is the ONLY way we can document the usefulness of our service.
The current list of publications is available here.
Please also acknowledge the use of our service in any publications you produce, using the following reporting guidelines, including the DOI 10.17169/refubium-26754.
Config file
// Config profile metadata
params {
config_profile_contact = 'Wassim Salam (@wassimsalam01)'
config_profile_contact_github = '@wassimsalam01'
config_profile_contact_email = 'TODO'
config_profile_name = 'FUB Curta'
config_profile_description = 'Freie Universität Berlin HPC (Curta) profile'
config_profile_url = 'https://www.fu-berlin.de/en/sites/high-performance-computing/index.html'
max_cpus = 32
max_memory = 772.GB
max_time = 14.d
}
def membership = ['/bin/bash', '-c', 'groups'].execute().text.trim()
// Slurm configuration parameters
process {
executor = 'slurm'
maxRetries = 3
clusterOptions = { (membership.startsWith("b") ? '--partition=begendiv,main ' : '--partition=main ') << ( task.time <= 3.h ? '--qos=hiprio' : ( task.time <= 3.d ? '--qos=prio' : '--qos=standard' ) ) }
}
// Singularity parameters
singularity {
enabled = true
autoMounts = true
}
cleanup = true
executor
contact
