This page has the links to the presentations of the workshop.
| Tuesday 26 January 2016 |
| 09:00-09:30 |
Uli Schättler |
Introduction
- Overview on current developments
|
pdf |
| 09:30-10:30 |
MCH, all |
New Coding Style for the COSMO-Model
- Abstraction vs. explicit coding:
Several new features have been implemented wich provide abstraction
for some functionalities, e.g. the tracer structure or the block
copy for the blocked data structure. Is this a valid or desired approach
for the COSMO-Model? Is this the way to go for a better understanding
of the code?
|
pdf |
| 10.30-11:00 |
Coffee Break |
| 11:00-12:15 |
Uli Schättler, all |
COSMO-ICON Physics
- status
- blocked data structure
- allocation of local memory
- coarse radiation grid
- impacts on COSMO-ART, CLM, 2-moment scheme
|
pdf |
| 12:15-12:30 |
Xavier Lapillonne |
CLAW: A Code Transformation Tool to optimize for GPU / CPU |
pdf |
| 12.30-13:30 |
Lunch Break |
| 13:30-15:00 |
Xavier Lapillonne |
OpenACC Tutorial |
|
| 15:00-15:30 |
Coffee Break |
| 15:30-17:00 |
Xavier Lapillonne |
OpenACC Tutorial |
|
|
Wednesday 27 January 2016 |
| 09:00-10:30 |
MCH, all |
Workflow for Preparing new Versions |
pdf |
| 10.30-11:00 |
Coffee Break |
| 11:00-12:00 |
MCH, all |
GridTools Update; Python Interface; Discussion |
pdf |
| 12:00-12:30 |
MCH |
Migration to GIT at MCH and Code Validation |
pdf |
| 12.30-13:30 |
Lunch Break |
| 13:30-15:00 |
All |
Discussions
- SMC Requirements:
SMC wants that we give some recommendations on how to write or modify
the model code in order to efficiently implement GPU capabilities,
including C++ and OpenACC standards. What do you think is necessary
for such recommendations? What would you expect from such recommendations?
- Unit-Testing in COSMO:
Some modules provide build-in testing functionalities (src_lbc,
src_block_fields, mo_random). Should this be exposed in COSMO? Perhaps with
compiler flag switches? Should this be encouraged?
|
pdf |