News and Announcements
NCAS-CMS will be running a 1-day Rose/Cylc UM conversion course on Friday 5th May at Reading University. Rose & Cylc are the replacement user interface and job submission/control system for the UMUI for UM version 10.x. The course is aimed at those already familiar with running the UM under the UMUI and will not cover how the UM itself works. (See our UM Training course page for details of other courses we run: http://cms.ncas.ac.uk/wiki/UmTraining). The majority of the course will be hands-on exercises supplemented with lectures.
- Overview of Rose/Cylc
- Creating & managing suites
- How to configure, build and run a UM application
- Examining Rose suite output
- Common pitfalls and troubleshooting
Please only attend this course if you are planning on using Rose within the next few months. We will be running this course again later in the year, subject to demand.
We still have a few spaces left. To register for this course please email: cms-support@… including a brief summary of what model configurations/versions you intend to be working with.
Course attendees are responsible for paying their own travel costs.
A new version of the CF Checker has been released and is now available for download from Github (https://github.com/cedadev/cf-checker/releases). A distribution will also be available on PyPI soon.
This is a major update to remove the dependency on CDAT; moving to use netcdf4-python instead.
NCAS-CMS will be delivering a 3-day course introducing new users to the Met Office Unified Model with the new Rose/cylc system. The course is a mix of practical exercises and informal lectures covering a range of topics including running suites, code management, file formats and troubleshooting common problems.
The course will be held at the University of Reading from 29-31 March 2017. There is no fee for attending the course, and lunch will be provided. However attendees are responsible for their own travel, accommodation and other subsistence costs.
For further details and a link to the registration form see:
Please note, this course is intended for new UM users only. CMS will also be running a 1 day Rose/cylc conversion course for users familiar with running the UM through the UMUI. Details of this will follow soon.
We recently ran the first, this academic year, of our popular 3 day UM Introduction courses. The course was attended by researchers from around the UK, as well as 3 participants from the US.
For the first time the course focused entirely on running the UM through the Rose/cylc interface. Attendees worked through a series of exercises designed to help users become familiar with the UM software, including how to diagnose common problems. We also had 2 very interesting science talks from Carol Halliwell and Sylvia Bohnenstengel of the Met Office.
The course was very well received with everyone who responded to our request for feedback (11 out of 19 participants) rating the lectures, practicals and facilities good or excellent.
The next course will run 29-31 March 2017, with registration opening in January 2017. For more details see: http://cms.ncas.ac.uk/wiki/UmTraining
The Met Office recently released version 1.3 of Mule. Mule is a Python API for reading & writing UM format files, and a set of utilities to replace the UM executables such as cumf and pumf.
For more information on Mule, see the user guide: https://code.metoffice.gov.uk/doc/um/index.html. (Note you will need a MOSRS account to view this page.)
Mule is available on ARCHER for the serial and login nodes. To use, a one-off setup step is required to add the Mule library to your Python path. For instructions see: wiki:ToolsAndUtilities/mule
Work undertaken to optimise Pier Luigi Vidale's N512 GA7 runs by NCAS-CMS has resulted in a 15-20% speed up of the model with no loss of bit-comparison. The savings in ARCHER resource resulting from this effort are very significant - at ARCHER partner rate this is ~£60k, and twice that at the non-partner rate. Another way of looking at this is that we save ~100M AU (for these experiments) which enables a raft of ARCHER projects which may struggle otherwise.
In the standard GA7, stochastic physics is turned on. This holds a field in spectral space distributed over the PEs. Every timestep this field is gathered by PE0, converted to grid space and then distributed over the domain. This requires a gather and many scatters at everytime timestep. A month long N512 high resolution job was examined with drhoook and it was noted that FOR_PATTERN, the routine which does the gather/scatter and spectral→grid space transformation takes a significant proportion of the run time, 513s for a 3000s run.
FOR_PATTERN has been rewritten to remove all gathers and scatters of the spectral field. Instead of having the spectral rows distributed over all PEs, each PE holds the spectral rows equivalent to its own rows in grid space. It then does a local fourier transform to get back to grid space, and then extracts the its own longitude domain from the resultant field.
This requires extra compute as every PE over a latitude band does the same spectral→grid space transformation, but the savings communication times can far outweigh this. Gather/scatters when running on a large number (1000s) of PEs are best avoided as these can take many milliseconds to perform.
With it new code FOR_PATTERN takes 23s, compared with 513s.
Two test jobs have been run. A high resolution N512 and an AMIP N96 configuration. For both there is full bit compariability with the previous version.
The N512 GA7 is now being run on Archer with this branch. The speed increase, from Pier Luigi Vidale:
So, for two domain decompositions and for a 2-month dump at N512: 48x48: ~7 hours, but a few times as short as 6hrs40mins (down from ~8 hours) 48x72: 4hrs50m to 5hrs02m (down from ~6hours)
So a ~15%-20% speed up
For the AMIP GA7 runs, there was no discernable speed change, but this isn't surprising as the gather/scatter would be much faster whilst running on a low number of PEs
We recently delivered the second of our bi-annual 3-day UM Training courses. This time we hosted 16 attendees from a wide range institutions from across the UK. Three days of hard work running and breaking UM jobs on ARCHER with specialist presentations from MO, and NCAS experts. This was our final UMUI-centric training course, next time Rose/Cylc will be the main focus. Many thanks to those attending and presenting.
Course details and presentations are available here.
Dates: 13-15th April 2016
Location: University of Reading
Applications for this course are now open, please apply for places through the application form
This course, run by NCAS-CMS, introduces new users to the Met Office Unified Model systems and provides practical experience of setting up and running experiments. It comprises a mix of lecture-style delivery covering a range of topics, including UM overview, Running the UM, UM file formats, post processing tools, etc and a significant practical component with exercises run on ARCHER to introduce new UM users to HPC and to help users past the initial steep learning usually involved with UM modelling.
Closing date for applications: 5th April
Accommodation, subsistence costs and reasonable travel within the UK will be funded for eligible students and scientists.
The Met Office/NERC Cloud Model (MONC) is a large eddy simulation. It's predecessor was the Met Office Large Eddy Model (LEM). MONC includes all the LEM science, but is written in modern Fortran, using modern programming techniques. It is much faster than the LEM, much more scalable and can deal with larger domains.
MONC can be run on a variety of computers, for example Linux boxes, MONSooN and ARCHER.
MONC has a BSD license and can be accessed at the Met Office Science Repository Service.
December saw the successful delivery of one of our bi-annual 3-day UM Training courses. This time we hosted 25 attendees from a wide range institutions from across the UK. Three days of hard work running and breaking UM jobs on ARCHER with specialist presentations from EPCC, MO, and NCAS experts left us all ready for the Christmas break. Many thanks to those attending and presenting.
Course details and presentations are available here.
Those of us feeling photogenic appear in the course photo: