New URL for NEMO forge!

Since March 2022 along with NEMO 4.2 release, the code development moved to a self-hosted GitLab.
This present forge is now archived and remained online for history.
[training] XIOS training session (4-6 Nov. 2020) – Message List – Discussion – NEMO

News (#8) - [training] XIOS training session (4-6 Nov. 2020) (#95) - Message List

[training] XIOS training session (4-6 Nov. 2020)


The XIOS team will open a training session on XIOS and dr2xml from November 4th to 6th 2020 at IDRIS.

This training session will include presentations of XIOS & dr2xml and most importantly several hands-on exercises.

A dedicated webpage for this training is available here and information will be updated regularly including a more detailed program of the training and slides.

This training is free. Coffee and lunches are also provided.

What is XIOS:

The climate simulation models launched on super-computers produce a significant amount of data. At this scale, the management of output files and post-treatment on data become the performance bottleneck for those climate models. In order to generate efficiently the output from models while taking advantage of the calculate resources, Institut Pierre Simon Laplace has launched the XIOS project.

XIOS is a library dedicated to generate efficiently and easily the parallel input/output of climate models. It uses a client-server methodology in which clients (or models) will simulate the physics and send data asynchronously to servers. It is the XIOS servers who are in charge of writing data to files. In this way, we can smooth the I/O peaks by sending to servers a constant data flux all along the simulations and the I/O is overlapped with calculates.

What's more, XIOS provides multiple functionalities which allow an "in situ" data post-treatment. For example, we can read the data from models, operate transformations on those data, combine data from different source models, and send results to multiple output files. All these operations are defined from an external XML file with a very compact and flexible way. Any modifications in this XML file will not require re-compilation of XIOS or models.

XIOS is written in C++ and uses MPI for asynchronous communications between clients and servers. NETCDF, HDF5, and MPIIO are also used in XIOS for efficient file writing.

What is dr2xml:

DR2XML is a python package useful for climate models whose I/O are handled by XIOS. It has been originally designed to facilitate the configuration of XIOS (automatic generation of XML files) so as to fully satisfy the CMIP6 DataRequest?. With DR2XML, the NetCDF files output by the model are directly CMOR3 compliant. Some extended functionalities were also developed so that DR2XML enables the user to set up his own Data Request. Consequently DR2XML can be used to customize the official CMIP6 DR or even to work off the CMIP6 framework, but benefiting from the CF-NetCDF, CMOR, and CMIP6 standards (file naming convention, file structure, variable definitions, file and variable attributes, etc.).

Registration to the training:

If you are interested in following the XIOS&dr2xml training, please fill up the registration form.

About the training:

For the moment, we have the possibility to give the tutorial onsite. However, this may be changed to a online tutorial if more strict sanitary condition applied. In case of online training, there will be no hands-on exercise. Due to the current health crisis of COVID-19, this onsite training is limited to 12 participants.

We are maintaining the training session onsite for several reasons. Firstly, a face-to-face tutorial is more efficient and interactive than reading slides online. Secondly, learning XIOS requires lots of hands-on exercises. For the moment we do not have the necessary material nor the human resources to ensure the online exercise. Last but not least, we are aware of the numerous demands on XIOS training sessions. So XIOS team are planning to do XIOS trainings more regularly. The next session is planned for the first trimester of 2021. Others will also be scheduled.

If you have any questions about this training, please contact Yushan Wang (…) or XIOS team.

Regards, XIOS team


No attachments created.