Changes between Version 21 and Version 22 of 2013WP/2013Action_institutions_CMCC
- Timestamp:
- 2013-10-30T09:20:04+01:00 (11 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
2013WP/2013Action_institutions_CMCC
v21 v22 36 36 37 37 === '''CMCC.1''' Scalability Optimization === 38 ''' Motivation:''' Improve NEMO scalability. [[BR]] ''' Status:''' Partly closed. The feasibility study related to the introduction of the PETSc library has been concluded. From this study it has been deduced that a consistent manpower is needed. Since this effort is not available using the currently funded projects, the activity related to the introduction of PETSc into the NEMO code will be sustained when new project funds will be achieved. A report on the results of this study is available and will be provided at the November merge party. [[BR]]XIOS has been validated on the CMCC Sandy Bridge systems (ATHENA). [[BR]] ''' Main tasks:'''[[BR]] * Introduction of a new software layer in the NEMO stack software exploiting numerical optimized and parallel libraries such as pBLAS, ScaLAPACK and PETSc. [[BR]] * Continuation of the implementation of hybrid parallel approach based on the functional parallelism for thetracers, momentum equations and 3D domain decomposition.[[BR]] * Test performances of XIOS on CMCC architectures[[BR]] '''System Reviewer:''' [[BR]] ''' Deadline:''' 2013[[BR]] ''' Priority: '''Medium [[BR]] ''' Principal investigator:''' Italo Epicoco [[BR]]38 ''' Motivation:''' Improve NEMO scalability. [[BR]] ''' Status:''' Partly closed. The feasibility study related to the introduction of the PETSc library has been concluded. From this study it has been deduced that a consistent manpower is needed. Since this effort is not available using the currently funded projects, the activity related to the introduction of PETSc into the NEMO code will be sustained when new project funds will be achieved. A report on the results of this study is available and will be provided at the November merge party. [[BR]]XIOS has been validated on the CMCC Sandy Bridge systems (ATHENA). [[BR]] ''' Main tasks:'''[[BR]] * Introduction of a new software layer in the NEMO stack software exploiting numerical optimized and parallel libraries such as pBLAS, ScaLAPACK and PETSc. [[BR]] * Continuation of the implementation of hybrid parallel approach based on the functional parallelism for the tracers, momentum equations and 3D domain decomposition.[[BR]] * Test performances of XIOS on CMCC architectures[[BR]] '''System Reviewer:''' [[BR]] ''' Deadline:''' 2013[[BR]] ''' Priority: '''Medium [[BR]] ''' Principal investigator:''' Italo Epicoco [[BR]] 39 39 40 40 === '''CMCC.2 ''' Modification of masks === 41 ''' Motivation:''' Expand the modification of masks to the EEN scheme [[BR]] ''' Status:''' started in 2012, continued in 2013[[BR]] ''' Main tasks:''' Evaluate the impact of fmask on the EEN scheme and eventually make modifications. [[BR]] '''Science Reviewer:''' Gurvan Madec[[BR]] ''' Deadline:''' 2013[[BR]] ''' Priority:''' High[[BR]] ''' Principal investigator:''' Srdjan Dobricic [[BR]]41 ''' Motivation:''' Expand the modification of masks to the EEN scheme [[BR]] ''' Status:''' The modification of masks has been implemented for the energy and for the enstrophy conserving schemes. The implementation for the energy and enstrophy (Arakawa) conserving scheme has not been tested and therefore still cannot be implemented officially. The PI left CMCC and the development is suspended.[[BR]] ''' Main tasks:''' Evaluate the impact of fmask on the EEN scheme and eventually make modifications. [[BR]] '''Science Reviewer:''' Gurvan Madec[[BR]] ''' Deadline:''' 2013[[BR]] ''' Priority:''' High[[BR]] ''' Principal investigator:''' Srdjan Dobricic [[BR]] 42 42 43 43 === '''CMCC.3 ''' Direct interface with data assimilation === 44 ''' Motivation:''' Interface with data assimilation systems [[BR]] ''' Status:''' to be started[[BR]] ''' Main tasks:''' Develop software interface that will avoid the writing to the disk, provide the possibility to use a different domain decomposition in Nemo and the data assimilation, reducethe memory usage of Nemo when it is inactive. [[BR]] '''Science Reviewer:''' !MetOffice[[BR]] ''' Deadline:''' 2013 [[BR]] ''' Priority: '''Medium [[BR]] ''' Principal investigator:''' Srdjan Dobricic [[BR]]44 ''' Motivation:''' Interface with data assimilation systems [[BR]] ''' Status:''' The direct interface with the data assimilation requires a simple method for the deallocation of the memory once it is not necessary. This should be done in order to reduce the size of the executable. In NEMO this is possible only for he working arrays, but when coupled with a data assimilation scheme the deallocation of the memory of many arrays that are not necessary for restarting Nemo becomes very complicated. The solution for this problem would completely change the form of the Nemo code and currently is not feasible. [[BR]] ''' Main tasks:''' Develop software interface that will avoid the writing to the disk, provide the possibility to use a different domain decomposition in Nemo and the data assimilation, reduce the memory usage of Nemo when it is inactive. [[BR]] '''Science Reviewer:''' !MetOffice[[BR]] ''' Deadline:''' 2013 [[BR]] ''' Priority: '''Medium [[BR]] ''' Principal investigator:''' Srdjan Dobricic [[BR]] 45 45 46 46 === '''CMCC.4 '''Divergence damping filter operator === 47 ''' Motivation:''' Additional diffusion on only the divergent component of the velocity could reduce the numerical noise at the high model topography (step topography, deep convection, etc.). [[BR]] ''' Status:''' continuation of 2012 activity. Tested but not documented[[BR]] ''' Main tasks:''' Add a routine that will apply the divergence filter in addition to the horizontal diffusion of the velocity.[[BR]] ''' System Reviewer:''' [[BR]]'''Science Reviewer:''' '''[[BR]]Deadline: '''2013[[BR]] ''' Priority:''' High [[BR]] ''' Principal investigator:''' Srdjan Dobricic[[BR]]47 ''' Motivation:''' Additional diffusion on only the divergent component of the velocity could reduce the numerical noise at the high model topography (step topography, deep convection, etc.). [[BR]] ''' Status:''' The divergence damping operator is implemented in the ASM package. It has also been implemented and tested in the dynamics package in Version 3.4, but the documentation is still missing. Therefore, it was not implemented. The PI left CMCC and the development is suspended.[[BR]] ''' Main tasks:''' Add a routine that will apply the divergence filter in addition to the horizontal diffusion of the velocity.[[BR]] ''' System Reviewer:''' [[BR]]'''Science Reviewer:''' '''[[BR]]Deadline: '''2013[[BR]] ''' Priority:''' High [[BR]] ''' Principal investigator:''' Srdjan Dobricic[[BR]] 48 48 49 49 === '''CMCC.5 '''Vertical stability of increment fields === 50 ''' Motivation:''' Modify the increment field in order to reduce the vertical instabilities. [[BR]] ''' Status:''' to be started[[BR]] ''' Main tasks:''' Modify the ASM code to add the diagnostics of the vertical stability and the modificatyions to the increment fields.[[BR]] ''' System Reviewer:''' [[BR]] ''' Science Reviewer:''' Met Office [[BR]]'''Deadline: ''' 2013[[BR]] ''' Priority:''' Medium[[BR]] ''' Principal investigator: '''Srdjan Dobricic50 ''' Motivation:''' Modify the increment field in order to reduce the vertical instabilities. [[BR]] ''' Status:'''The vertical stability of the increments is implemented in Version 3.3 and has not been updated in the newest version. Also the documentation is still missing.[[BR]] ''' Main tasks:''' Modify the ASM code to add the diagnostics of the vertical stability and the modificatyions to the increment fields. The PI left CMCC and the development is suspended.[[BR]] ''' System Reviewer:''' [[BR]] ''' Science Reviewer:''' Met Office [[BR]]'''Deadline: ''' 2013[[BR]] ''' Priority:''' Medium[[BR]] ''' Principal investigator: '''Srdjan Dobricic 51 51 52 52 === '''CMCC.6 '''Surface and lateral boundary conditions for tracers === 53 ''' Motivation:''' Add generic interfaces for surface and lateral boundary conditions for BGC tracers (river input, atmospheric deposition, open boundaryconditions). [[BR]] ''' Status:''' started in 2012, first implementation waiting for the final choice of BDY[[BR]] ''' Main tasks:''' Port the interface used for transport routines to the surface and boundary condition routines currently used for T and S. [[BR]] ''' System Reviewer:''' C. Ethe'[[BR]] ''' Science Reviewer:''' C. Ethe', ?? [[BR]] '''Deadline:''' 2013[[BR]] ''' Priority:''' High [[BR]]''' Principal investigator: '''Marcello Vichi53 ''' Motivation:''' Add generic interfaces for surface and lateral boundary conditions for BGC tracers (river input, atmospheric deposition, open boundary conditions). [[BR]] ''' Status:''' started in 2012, first implementation waiting for the final choice of BDY[[BR]] ''' Main tasks:''' Port the interface used for transport routines to the surface and boundary condition routines currently used for T and S. [[BR]] ''' System Reviewer:''' C. Ethe'[[BR]] ''' Science Reviewer:''' C. Ethe', ?? [[BR]] '''Deadline:''' 2013[[BR]] ''' Priority:''' High [[BR]]''' Principal investigator: '''Marcello Vichi 54 54 55 55 === '''CMCC.7 '''Testing of AGRIF: scalability and usage with biogeochemical variables === 56 ''' Motivation:''' evaluate AGRIF performances in realistic domains and utilization with the BFM[[BR]] ''' Status:''' this activity has not been funded and will not be undertaken for this year[[BR]] ''' Main tasks:''' check performances of AGRIF with an Atlantic refinement; assess thecompatibility with BFM variables and the related computational overhead[[BR]] ''' System Reviewer:''' [[BR]] ''' Science Reviewer:''' Dorotea Iovino[[BR]] '''Deadline:''' 2013-2014[[BR]] ''' Priority:''' Medium [[BR]]''' Principal investigator: ''' Marcello Vichi and Italo Epicoco56 ''' Motivation:''' evaluate AGRIF performances in realistic domains and utilization with the BFM[[BR]] ''' Status:''' this activity has not been funded and will not be undertaken for this year[[BR]] ''' Main tasks:''' check performances of AGRIF with an Atlantic refinement; assess the compatibility with BFM variables and the related computational overhead[[BR]] ''' System Reviewer:''' [[BR]] ''' Science Reviewer:''' Dorotea Iovino[[BR]] '''Deadline:''' 2013-2014[[BR]] ''' Priority:''' Medium [[BR]]''' Principal investigator: ''' Marcello Vichi and Italo Epicoco