Versions Compared

Old Version 1

changes.mady.by.user Xavier Abellan

Saved on

compared with

New Version Current

changes.mady.by.user Xavier Abellan

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.


Warning

TEMS is to be retired at the end of October 2021. See more information here .

TEMS is a test platform where users and system administrators can prepare the migration to the final Atos systems.

This means that by nature TEMS is evolving with time, and may not be as feature complete as other production systems.

Tip

If you find any problem or any feature missing that you think should be present, and it is not listed here, please let us know  by reporting as a "Problem on computing" through the ECMWF Support Portal mentioning "TEMSAtos" in the summary.

Table of Contents

Missing Features

Comprehensive software stack

We have provided a basic software stack that should satisfy most users, but some software packages or libraries you require may not be present. If that is the case, let us know by reporting as a "Problem on computing" through the ECMWF Support Portal mentioning "TEMS" in the summary.

Atos HPCF is not operational platform yet, and many features or elements may be gradually added as complete setup is finalised. Here is a list of the known limitations, missing features and issues.

Table of Contents

Missing Features

End of job information

A basic report is provided at the end of the job with information about its execution.

No Format
## INFO[ECMWF-INFO -ecepilog] --------------------------------------------------------------------------------------------- 
## [ECMWF-INFO -ecepilog] This is the ECMWF job Epilogue.
[ECMWF-INFO -ecepilog] +++ Please report problems to ServiceDesk, servicedesk@ecmwf.int
## INFO issues using the Support portal +++
[ECMWF-INFO -ecepilog] +++ https://support.ecmwf.int                     +++
[ECMWF-INFO -ecepilog] ---------------------------------------------------------------------------------------------
## INFO
## INFO[ECMWF-INFO -ecepilog]
[ECMWF-INFO -ecepilog] Run at 2021-0409-08T1128T06:1921:4225 on TEMS
## INFOaa
[ECMWF-INFO -ecepilog] Job Name                  : myjob
## INFOeci
[ECMWF-INFO -ecepilog] Job ID                    : 66894
## INFO1009559
[ECMWF-INFO -ecepilog] Submitted                 : 2021-0409-08T1128T06:1905:38
## INFO23
[ECMWF-INFO -ecepilog] Dispatched                : 2021-0409-08T1128T06:1905:38
## INFO23
[ECMWF-INFO -ecepilog] Completed                 : 2021-0409-08T1128T06:1921:42
## INFO25
[ECMWF-INFO -ecepilog] Waiting in the queue      : 0.0
## INFO[ECMWF-INFO -ecepilog] Runtime                   : 4.0
## INFO962
[ECMWF-INFO -ecepilog] Exit Code                 : 0:0
## INFO Account[ECMWF-INFO -ecepilog] State                     : ecus
## COMPLETED
[ECMWF-INFO Queue-ecepilog] Account                    : np
## INFO Ownermyaccount
[ECMWF-INFO -ecepilog] Queue                     : uid(9999)
## INFO STDOUTnf
[ECMWF-INFO -ecepilog] Owner                     : myjob.66894.out
## INFO STDERRuser
[ECMWF-INFO -ecepilog] STDOUT                    : myjobslurm-1009559.66894.out
## [ECMWF-INFO Nodes-ecepilog] STDERR                    : 1
## INFO Tasksslurm-1009559.out
[ECMWF-INFO -ecepilog] Nodes                     : 16
## INFO CPUs/Task   1
[ECMWF-INFO -ecepilog] Logical CPUs              : 4
## INFO8
[ECMWF-INFO -ecepilog] SBU                       : 220.722460 units
## INFO Logical CPUs              : 256
## INFO[ECMWF-INFO -ecepilog]


Warning
iconfalse
  • There is no charge made to the project accounts for any SBUs used on the TEMS system
  • The SBU cost provided in the end of job information for jobs run on TEMS reflects the expected SBU cost on the final Atos system.
  • We are unable to provide a figure for the memory used at this time.

Alternatively, you may use sacct to get some of the statistics from SLURM once the job has finished.

SSD disks on GPIL nodes

The GPIL nodes have local SSDs with some 950GB capacity. These have not been mounted yet, as we need to develop a service model for their use.

Known issues

Intel MKL

...

greater than 19.0.5: performance issues on AMD chips

Recent versions of MKL do not use the AVX2 kernels for certain operations on non-intel chips, such as the AMD Rome on TEMSour HPCF. The consequence is a significant drop in performance.