Skip to end of metadata
Go to start of metadata

System Billing Unit (SBU)

To ensure that computing resources are distributed equitably and to discourage irresponsible use users' jobs are charged for the resources they have used against a project account. The project accounts have computer usage units allocated at the beginning of each accounting year. From 2022 the accounting year, up to now following the ISO 8601 calendar, will follow the Gregorian calendar, i.e. start on 1 January and finish on 31 December.

The accounting on the HPCF is currently based on the total elapsed time for which cores are used. With that, the HPCF unit of account, the System Billing Unit (SBU), for both serial and parallel jobs is defined as follows.

According to the current migration plan the future Atos system will take over from the Cray as operational HPCF on 1 May 2022. Before that date only Cray usage will be charged against any SBU allocations. From 1 May 2022 the Cray will no longer be available and only Atos HPC usage will be charged against SBU allocations.

ECMWF reserves the right to alter the formulas below in the light of experience.

Cray XC40 (cca,ccb)



P = 50365140 / (3617 * 36 * 86400)

A fixed proportionality factor, adjusted so that one cluster will deliver 50,365,140 units per day on a "typical"
ECMWF job mix on the HPC system (3617 * 36 is the number of cores of one Cray XC 40 cluster and 86400
the number of seconds in a day)


The elapsed time (real time in seconds) recorded for a job, multiplied by the number of cores used by the job


  1. A serial job using 1 hour (3600 seconds) of elapsed time will be charged about 16.11 SBU.
  2. A parallel job requesting 2 nodes, running for 3 hours elapsed time will be charged 3481.14 SBU.

More information on the Cray HPC system, which might be helpful to compare it to other platforms and to estimate computational cost in terms of SBUs, is available at and the HPCF user documentation.

Atos HPC

Preliminary formula for the future Atos HPC system. Please note that the system will have hyperthreading enabled, providing two logical CPUs per core.



P = 115750592 / (2000 * 128 * 86400)

is a fixed proportionality factor, adjusted so that one cluster will deliver 115,750,592 units per day on a "typical"
ECMWF job mix on the HPC system (2000 * 128 is the number of cores of one Atos clusters and 86400
the number of seconds in a day)


is the elapsed time (real time in seconds) recorded for a job, multiplied by the number of cores used by the job


  1. A serial job using 1 hour (3600 seconds) of elapsed time will be charged about 18.84 SBU.
  2. A parallel job requesting 2 nodes, running for 3 hours elapsed time will be charged 14468.82 SBU.

More information on the Atos HPC system, which might be helpful to compare it to other platforms and to estimate computational cost in terms of SBUs, is available in a Newsletter article  and in the Atos HPCF Documentation.

User Usage

The following two commands are available on ecgate to help determine your day to day usage.

account [-l uid]

This lists the accounts under which a user-id is registered. If more than one account is listed, then by default, job usage will be debited to the first account listed, the primary account. If you wish to debit the usage of a job to an alternative account (out of your list of assigned accounts) you may do so by using the following PBS statement

#PBS -l EC_billing_account=<other_account>


The acct_status command allows users to check the usage of their project accounts. The usage is:

acct_status [-p week/period] [-r] [-u] [-h hostname] [-y year] [project(s)]

-p : gives the usage for the current week or 4-week period, e.g. -p week

-r : gives information for sub-projects

-u : gives information for each user of the project account

-h : gives the usage for the specified hostname (e.g. cca). If omitted the cumulative total for all supercomputers is reported

-y : accounting year YYYY (default: current ISO year)

project(s) : the project(s) or node(s) to be reported on. The default is the project account of the logged in identifier.

The reply shows the usage in terms of SBUs (System Billing Units), it is largely self explanatory. You can only check the usage of those accounts which you are registered to use. To see the list of those accounts, use the command "account".

Note the following:

  1. "Total use" is the cumulative usage since the beginning of the year.
  2. The "current" week begins on a Monday
  3. The reply for the current week or period (using -p) shows:
    • the SBUs used for the current week/period (SBUs:)
    • the total used to date this year (total:)
    • the current allocation (alloc:)
    • the percentage use made to date of the allocation.
  4. The node report may list "Responsible User(s) and/or Privileged User(s)" - these are the persons who are permitted to see information at this node level and all levels below it. A "Privileged User" is a person who can change allocations for nodes and projects below this level. The "Privileged User" for a Member State could be its Computing Representative.
  5. The usage figures are usually updated daily (before 0700) and include all jobs that finished before 00:00 on the current day. Operational problems may cause this update to be delayed for any given day.

Some examples:

acct_status abc xyz

will report on the two projects or nodes abc, xyz

acct_status -u

will report on the individual usage of the account by the logged in identifier

acct_status -p week -r ms

will report the current week's usage for the node ms and sub-projects under it. Only a "Responsible User" or the "Privileged User" for a node can obtain this report.


The eoj (end of job) command gives some basic statistics about your job including the usage of SBUs. It has been included in the system epilogues but can also be used interactively at any time to check your running job. Login to the HPC and type:

   eoj job-ID

To find the job ID of your job use the qscan command. The eoj command without argument can be placed at various points in an HPC batch job to show job accounting information up to that point within a job. For example, eoj will provide a summary report from the beginning of the job up to that eoj command. To obtain such a report just add the line


to your job script. It is also possible to obtain a report on a specific sequence of commands by getting an eoj report before executing the sequence, another one when the sequence is finished, and computing the difference of the relevant terms. This information can then be used to identify blocks of commands within a job which take a lot of SBUs to execute.

Please note that

  • The information given by eoj is updated every 10 seconds at the earliest
  • For parallel jobs the reported CPU time is not correct

For more information on eoj use eoj -h on the HPC system.

Accounting and account management

Every user of the ECMWF computer system is given a separate user identifier. Because of both security requirements and the way ECFS files are stored and identified, a given identifier must not be shared by two or more users. The first 2 characters of the user identifier are common to a given section or group of people, (e.g. all User Support identifiers begin US). The following characters are then assigned by the ECMWF manager responsible for the user, by the Member State Computing Representative, or by the Principal Investigator of a Special Project, as appropriate.

Groups of users are assigned to a computer project (or account). Each computer project has a unique account name. Reports listing usage against each computer project are generated at regular intervals (daily, weekly, 4-weekly, annually), the 4-weekly ones are available, via e-mail, upon request to User Support.

Computing resource allocation rules

Member States are allocated computing resources for their own purposes according to the Council guidelines on distribution of computer resources. Member State Computing Representatives may distribute those resources among projects registered at ECMWF as they feel appropriate. Should a redistribution of resources between the projects of a Member State be required during the course of a year, this may be carried out by the Member State Computing Representative or another designated "privileged user" (see section Administration of project accounts in the Member States   below).

Special Projects

Special Projects are experiments or investigations of a scientific or technical nature undertaken by one or more Member States, likely to be of interest to the general scientific or technical community. In the spring of each year, Directors of national meteorological services are invited to submit requests for Special Project resources to be available for use in the following year. Requests are considered by the Council at its winter session, in conjunction with recommendations from the Scientific Advisory Committee and Technical Advisory Committee. Since Special Project accounts are separate from Member State resources, their account administration is performed by ECMWF. For more details see

Daily Accounting Controls

Information is extracted daily from HPCF report files to produce a cumulative total of resources used per user and per account. Use of each account is checked daily and once usage reaches 90% of its resources available, a warning message is sent out, to enable users to arrange for the transfer of additional resources, if necessary. Once usage reaches 100% no further use of that account should be made.

Period (4-weekly) account reporting

Member State Computing Representatives or Special Project Principal Investigators, receive a regular summary of accounting information relating to all their Member State's projects or their Special Project(s) respectively at the end of each four week accounting period. The format of the report is shown below. If a user does not use the system in any period, no entry is made.

Format of HPC account report
1.     Member State:
2.     Combined 4-week report for period ending (MMDD) : 0128
3.     Node:     ms        SBUs: 31.9     Total: 51.9    Limit: 20000.0 (0%)
4.     Acct:     msacct1   SBUs: 31.9     Total: 31.9    Limit: 10000.0 (0%)
5.               User: ms1 SBUs: 21.1
5.               User: ms3 SBUs: 10.2
5.               User: msd SBUs:  0.6
4.     Acct:     msacct2   SBUs:  0.0     Total: 20.0    Limit: 10000.0 (0%)


Line  1       Name of Member State

Line  2       Period concerned, where the period is identified by its last day 
              (in this example January 28).

Line  3       Overall node for the Member State "ms" showing
              SBUs:     total used this period
              Total:    cumulative total for this year
              Limit:    overall allocation, with percentage used to date

Line  4       The summary for a given account (msacct1 and msacct2 in this 
              example). The entries have the same meaning as in line 3.

Lines 5       Usage per user (ms1, ms3, msd in this example) during the period 
              in question.

              User:          user identifier
              SBUs:          total used this period

Registering a new account

New accounts are registered by ECMWF only on the authority either of an ECMWF Department Head (for ECMWF staff accounts) or of a Member State Computing Representative (for Member State accounts) or the Principal Investigator (for Special Projects). For further information contact User Support.

Administration of project accounts in the Member States

Member States may nominate "privileged users" to perform limited administration of Member State accounts using the acct_admin program, see below. By default, the Centre will grant this status to all Member State Computing Representatives with a registered user identifier, but Member State Computing Representatives may propose an alternative person and/or nominate a backup if they wish.

A "privileged user" may be authorised to modify all or a reduced set of Member States' accounts. For instance, Member States are free to allocate control to several "privileged" users, each having control of a particular sub-node. It should be noted, however, that the nomination of privileged users should be kept to a minimum.

For security reasons, the establishment of new project accounts and the addition of new users remains under the control of ECMWF.

Running the account administration program

The ECMWF accounting administration program is started (on ecgate) by the command:


The following list of commands is available on entering acct_admin:

    read allows only testing and checking 
    edit to modify account allocations (privileged users only) 
    exit to terminate having made permanent changes 
    quit to terminate without any change

The modes "edit" or "read" may be specified at command invocation, e.g.

    acct_admin edit

or after entering the acct_admin program.

Within "edit" or "read" mode the prompt is

    Type command:

A list of subcommands may be displayed by the subcommand "help" and is given below.


help [commands]
? [commands] (help)
add-responsible-user account user
remove-responsible user account user
set-sbu-allocation account allocation
add-sbu-allocation account allocation
remove-sbu-allocation account allocation
transfer-sbu-allocation from-account to-account allocation
list-account account [and-sub-accounts] [and-users]
list-all-account account [and-users]
list-users account [all]
list-tree account

Notes regarding these subcommands:

    1. Account can be a node or a sub-node name.
    2. Allocation is replaced by an actual value (likewise allocation-increment, allocation-decrement and allocation-change).
    3. Items within [ ] are optional qualifiers.
    4. The value to be inserted for [and-sub-accounts] & [and-users] are yes or no. The value yes means that that option will be turned on (no is default).
    5. Additional commands besides these may be listed but these are only available to the ECMWF Account administrator.

As the invocation of the edit subcommand will lock the system, please make sure to leave the acct_admin command with exit or quit as otherwise it will not be available to others.

Listing allocations for Member State projects/nodes

To list all users on all projects for Member State Sweden (country code SE):

    list-account se yes yes

To list all users on Special Project spmsabcd

     list-users spmsabcd all

Changing allocations for Member State projects

To change the allocation of a project, use the 'edit' menu and the add/remove/transfer command

     acct_admin edit 
    1. To decrease the allocation of a project/node use:
            remove-sbu-alloc project amount
      If sub-nodes are used to sub-divide the allocation of a Member State, the allocation for the sub-node may also be decreased.
    2. To increase the allocation of a project/node use:
            add-sbu-alloc project amount
      It must be ensured that the node above the project/node has enough resources 'not allocated'. If not, some resources will have to be freed (as under A) before the increase can be done.
    3. To transfer an allocation from one project/node to another use:
            transfer-sbu-alloc from-account to-account amount
      This command will transfer resources from one project/node and it's ancestor nodes to the destination project/node and its ancestors.

Example 1

Country 'mo' uses the following nodes and accounts:

Node:          mo        Allocation: 10000,  Not allocated:  3000
          Acct:     moabcd         Allocation:  5000,  Total use:      2000
          Acct:     moxyz          Allocation:  2000,  Total use:      2239

The account 'moxyz' has used more than its allocation.

To add an extra 2000 units:

     acct_admin edit <<EOF
     add-sbu-alloc moxyz 2000

Supposing this project runs over its allocation again, the status becomes:

     Node:          mo        Allocation: 10000,  Not allocated: 1000
          Acct:     moabcd         Allocation:  5000,  Total use:     2000
          Acct:     moxyz          Allocation:  4000,  Total use:     4239

Now, the 2000 units must be transferred from the other account, as only 1000 units are unallocated.

     acct_admin edit <<EOF
     remove-sbu-alloc moabcd 2000
     add-sbu-alloc moxyz  2000

The following command would also have accomplished the last change more simply:

     transfer-sbu-alloc moabcd moxyz 2000 

Check status using:

     list-acc mo yes

Note that resources must be freed first, or else the error message "not enough unallocated resources to change allocation" is given.

Example 2

Country 'su' has 2 nodes subalt and suukr each with 2 projects:

     Node:          su          Allocation: 80000,  Not allocated:    0
        Node:       subalt      Allocation: 10000,  Not allocated:    0
          Acct:     subaest     Allocation:  5000,  total use:        2000
          Acct:     subalit     Allocation:  5000,  total use:        3000
        Node:       suukr       Allocation: 70000,  Not allocated:    0
          Acct:     suuksib     Allocation: 20000,  total use:        3000
          Acct:     suukrus     Allocation: 50000,  total use:        48000

The project 'suukrus' is running close to its limit. To increase the allocation of this project, some resources have to be freed first, as the node 'suukr' has no free allocation.

     acct_admin edit <<EOF

To free allocation from the project 'subaest':

     remove-sbu-alloc subaest 2000 

To give the allocation to 'suukrus':

     add-sbu-alloc suukr   2000
     add-sbu-alloc suukrus 2000

Check status using:

     list-acc su yes

To first run it as a test, replace the 'exit' command with 'quit'.

N.B. An easier way of doing the series of removes and adds is to just use:

     transfer-sbu-alloc subaest suukrus 2000

The transfers at the node levels, above the projects, are then taken care of automatically.

  • No labels