Projects¶
If you wish to apply for a project which will have resources allocated at HPC2N, the Principal Investigator (PI) has to apply for one according to the rules described on the NAISS allocations web page. There are two types of projects - compute and storage - each of which have different sizes of projects you can apply for: small, medium, and large. You apply for all of them on the NAISS SUPR page, under rounds. To do so you need an account at SUPR.
When you start a project application, the progress will be saved as ‘In Preparation’ on your SUPR page, and you can continue working on it later. Just remember to save the changes you make to each of the sections!
Note
When applying for resources at HPC2N, be aware that the following are considered separate resources and needs to applied for separately:
- Largemem nodes
- CPU nodes
- GPU nodes
Compute projects¶
You will need to fill in more information for the larger projects, than for the smaller ones. However, there is some information you will need regardless of the type of project: Title, Abstract, Resource Usage (how do you plan to use the requested resources), Classification (VR classification codes - only in Swedish), length of project (1-12 months), which resource(s) you wish to use, and the amount of core hours/month you need.
Much of the information on the application will be filled in, but you should check that your contact information and such is correct.
Note
To apply for a project, you need to be the PI.
Local¶
HPC2N’s cluster “Kebnekaise” is now only open for local project requests. This means you can only apply for time on Kebnekaise if you (the PI) is affiliated with one of HPC2N’s partner sites (UmU, LTU, IRF, MiUN, SLU). You can still add members from anywhere.
- Login to SUPR (or create a SUPR account if you do not have one).
- Either go directly to the HPC2N local compute round.
- Or
- Click “Rounds” in the left hand side menu, then pick “Compute Rounds” and then “Centre Local Compute”.
- Pick “HPC2N Local Compute YYYY” and choose “Create New Proposal for HPC2N Local Compute YYYY”. (YYYY is the current year).
- You can either create a proposal from scratch or use an earlier proposal as a starting point.
- If you have existing storage from an earlier project and wish to continue using the same directory to avoid having to move your files, do “continuation” on the existing storage.
- IMPORTANT: The GPU partition is now considered a separate resource, so if you need the GPUs, then you must add that as well under “Resources”.
- Decide if you need more storage than the default amount (500 GB).
- If that amount is enough, accept it and pick a directory name during the compute proposal.
- If you need more, instead apply for a local storage project in the “Storage Rounds” -> “Centre Local Storage”. Pick a name for the directory. Link it to the compute project to have the members of the compute project automatically become members of the storage project.
- More information and application here, at Local Compute rounds.
NAISS resources¶
Small (NAISS resources)
- Max core hours/month/resource depends on the resource and can be seen on SUPR, for each resource’s compute round page. It is currently around 5000 core hours/month.
- Applications can be submitted at any time via the SUPR small project application page.
- Evaluation of applications are performed locally, approximately once a week.
- For small projects and new groups that want to gain experience in using HPC-systems
Medium (NAISS resources)
- Max core hours/month/resource depends on the resource and can be seen on SUPR, for each resource’s compute round page..
- Applications can be submitted at any time via the SUPR project allocation page.
- Evaluation of applications are performed locally and assesses the feasability of using the requested resources.
- Evaluations are handled monthly. Submission deadline for applications are the 15th of each month. If you sent in your application in time and have waited for more than a month on the evaluation, contact us through mail to support@hpc2n.umu.se.
Before filling out the form to apply for a medium level project, you should read the instructions.
Large (NAISS resources)
- Applications for more time than the maximum available for a medium project on the intended resource .
- Minor applications for supplementary resources can be included.
- Calls for applications for large level projects are issued by SNAC twice a year. There are instructions here and an applications form here (under NAISS Large Compute).
- Applications are evaluated by NAC and they also decide on the allocations, based on scientific merit, need for the resources, efficient use of the resources, and impact.
- Note that if a large level application is granted time, the PIs existing medium and small level projects will be terminated.
Before filling out the form to apply for a large level project, you should read the instructions.
You can read the (OLD) slides here from the seminar by Philipp Schlatter: “SNIC and SNAC: Handling of large-scale computer time allocations in Sweden.”. Note that some things have changed since then.
Make sure you have the following information handy before filling in the application form:
- your name and affiliation,
- which project you are taking part in.
Please, also read the
before you start applying for an account. Take a look at an overview of the available resources at HPC2N.
Storage projects¶
The basic storage in your home directory at HPC2N is 25GB, so you would almost certainly need to apply for a storage project as well as a compute project. (See more information about our file systems here.
Storage projects are applied for through SUPR. While storage projects needs to be applied for separately, they are usually linked to a compute project.
The exception to this is the default storage (500GB) you can get with the compute project. If this is not enough, you need to apply for a storage project.
There are three sizes of storage project in SUPR; small, medium, and large. In addition, there is the “local” project. You will need to fill in more information for the larger projects, than for the smaller ones. However, there is some information you will need regardless of the type of project:
- Title
- Abstract
- Resource Usage (how do you plan to use the requested resources, size of your expected data, expected number of files)
- Data Management Plan (type of data, processing, data flow, how will the data be backed up)
- Directory Name (you get to chose the name for the project storage at HPC2N, or if you want to use the default snicxxxx-yy-zz named for your project ID)
- Classification (VR classification codes - only in Swedish)
- Length of project (1-12 months)
- Which resource(s) you wish to use
Much of the information on the application will be filled in, but you should check that your contact information and such is correct.
Note
To apply for a project, you need to be the PI.
Local¶
HPC2N’s cluster “Kebnekaise” is now only open for local project requests.
- Information and application here, at HPC2N Local Storage round.
NAISS resources¶
Small (NAISS resources)
- Max storage depends on the resource.
- Applications can be submitted at any time via the SUPR storage project allocation page.
- Evaluation of applications are performed locally, approximately once a week.
- If you have or are applying for a Small compute project, this is the storage project you would usually apply for with it.
Medium (NAISS resources)
- Max storage depends on the resource.
- Applications can be submitted at any time via the SUPR storage allocation page.
- Evaluation of applications are performed locally and assesses the feasability of using the requested resources.
- Evaluations are handled monthly, if you have waited for more than a month, contact us through mail to support@hpc2n.umu.se.
Before filling out the form to apply for a medium level project, you should read the instructions. The requirements for applying for a medium storage project are listed on SUPR.
Large (NAISS resources)
- Applications for more storage than the maximum available for a medium storage project on the intended resource.
- Calls for applications for large level storage projects are issued by SNAC twice a year. There is more information as well as instructions here. You apply through SUPR, from their Storage project application page.
- Applications are evaluated by SNAC and they also decide on the allocations, based on scientific merit, need for the resources, efficient use of the resources, and impact.
- Minor applications for additional allocations to be added to an existing LARGE project allocation can be included.
Before filling out the form to apply for a large level project, you should read the instructions which also lists the requirements for applying for a large storage project.
Project and account policies¶
- Each user should have one, and only one user account with HPC2N.
- Each user account may be added to as many projects as the user is participating in.
- A user should not have multiple user accounts for different projects.
- After the project has been approved and assigned a project ID, existing user accounts can be added to the project.
- If you do not yet have a user account, you may apply for a new one as described on the User Account page.
- Having a user account is now dependent on being a member or PI of a project.
- Projects CANNOT be extended any longer, but a new project application can be created based on the old project.
Adding users to existing projects¶
All user administration for all types of projects are handled by the PI (or the appointed proxy-PI) in SUPR. The users must first have an account in SUPR and have signed the NAISS User agreement. Signing it can be done on the “Personal Information” page in SUPR.
While storage projects needs to be applied for separately, they are usually linked to a compute project. The storage project PI can link one or several compute projects to the storage project, thereby allowing users in the compute project access to the storage project without the PI having to explicitly handle access to the storage project.
Expiring compute projects¶
If you have jobs still in the queue when your compute project expires, you need to remove those jobs or change the project to an active one as they will otherwise never start.
Project Information¶
Sometimes you need to obtain information about current (or past) projects running at HPC2N and their recent CPUhours usage.
We strongly recommend you to use SUPR for looking up allocation and resource usage. It is the most updated and correct version. However we also provide a command line program you can use called:
which allows users to list:
- project name, principal investigator (PI) and duration;
- allocation in CPUhours/month to a specific cluster and the recent fairshare usage;
- real usage in CPUhours;
- user accounts (and names) related to a project.
Note
Used CPUhours may differ if you look it up in SUPR compared to what you get from projinfo. Usually the numbers in SUPR are “more correct” than those you get when using projinfo. projinfo get its information from the batch system and does not take into account that different resources are charged differently, for instance GPU- and KNL-resources on Kebnekaise.
Info
projinfo’s synopsis is as follows:
projinfo [-v][-q] [-o] [-u user] [-p project] [-c cluster] -v : prints authorized users of a project -d : also print user usage per month -q : quiet, only print data (intended for script use) -o : print information about all (old) projects, not just current -u user : print information about specific user -p project : print information about specific project -c cluster : only print allocations on specific cluster -h : prints this help
Used time on the clusters are only available on the login node of the cluster.
Project should be specified as in the submit file
I.E. -p naissXXXX-YY-ZZ or hpc2nXXXX-YYY)
- If run without any option, projinfo lists all active projects for the current user.
- If -v option is specified, authorized user accounts will be listed along with the corresponding real names.
- Sometimes usage information is not available in the job scheduler, in which case you will see N/A for usage.
- Note that the real usage is approximate but gives a quick indication of the time used. However, the final scheduling priority is determined (among other things) by comparing recent fairshare usage with fairshare utilization target (displayed by projinfo in percent).
Example
The following example lists information for an example project SNIC2020-5-28:
projinfo -p snic2020-5-28 -v Information for project SNIC2020-5-28: PIs name: Project title Active from 20200901 to 20210831 Allocations: kebnekaise: 5000 CPUhours/month, used 109.2% the past 30 days kebnekaise-lm: 500 CPUhours/month, used 94.2% the past 30 days Usage on kebnekaise 2020 Sep: 5046 CPUhours Authorised users: user1 User’s real name user2 User’s real name
Questions?¶
If you have any questions about how to apply for a project, add users, etc., contact us through the support form on SUPR or mail them to: support@hpc2n.umu.se.