Marconi Fusion 6th Cycle: 18.753.600 core hours for reflectometry simulation
A total of 390,700 node hours (18.753.600 core hours) on A3 conventional partition has been obtained on the Marconi Fusion 6th Cycle. An additional 11,000 node hours to run in the C1 'GPU' partition have also been attributed, reflecting the many and plural reflectometry activities and interests presently running at IPFN and as part of continuing work on modelling and simulation of different aspects of reflectometry.
Reflectometry simulations have been supporting the development of models and the advancing of numerical techniques, comparisons of theory/simulation/experiments, neutronics assessments, development of new experimental techniques and the planning of reflectometry systems for future machines. The Marconi Fusion projects are inserted in the EUROfusion call categories affiliated with Plasma turbulence and related transport processes; Edge physics; Integrated modelling of fusion plasmas; and Reactor technology.
The European Fusion Research 'Roadmap to the realisation of fusion energy' considers theory and modelling efforts as decisive to validate experimental physics results and to prepare the design of the next fusion power plant DEMO. In accordance, EUROfusion places the Calls for Proposals for the use of the EUROfusion High Performance Computer , presently for the 6th cycle, to give access, computational time and resources on the HPC Infrastructure at Cineca, one of the Large Scale Facilities in Europe and a PRACE Tier-0 hosting site.
The fusion community has at its service the MARCONI-A3 partition with Intel SkyLake (2x Intel Xeon 8160 @2.1GHz 24 cores) in a total of 3216 nodes (48×3216 cores) and MARCONI100 C1 partition, an accelerated Marconi partition, with IBM Power9 AC922 @3.1GHz 32 cores HT 4, each node equipped with 4 NVIDIA Volta V100 GPUs, in a total of 980 nodes, paving the way to the accelerated pre-exascale Leonardo supercomputer.
Participating on these projects are Emanuel Ricardo, Jorge Santos, António Silva, Jorge Ferreira, Raul Luís, Filipe da Silva, José Vicente, Yohanes Nietiadi, Rennan Bianchetti Morales (UKAEA/CCFE) and Stéphane Heuraux (Université de Lorraine).
The 6th cycle will run from the 15th March 2022 until the 28th February 2023. The attributed hours will be distributed among the following projects:
- DPPRTRB-Effect of turbulence in the DEMO PPR measurements (PI Emanuel Ricardo): 104,000 node hours to run on the A3 conventional partition and 2,000 node hours to run in the C1 'GPU' partition.
- DTTsimul3-3D reflectometry simulation for DTT PPR (PI Jorge Santos): 52,000 node hours to run on the A3 conventional partition.
- EnE_RRMP-Reflectometry plasma tracking during ramp-up/rampdown: Application to DEMO (PI António Silva): 52,000 node hours to run on the A3 conventional partition and 3,000 node hours to run in the C1 'GPU' partition.
- EnR_RTOP-Advances in real-time reflectometry for next-generation machines (PI Jorge Ferreira): 52,000 node hours to run on the A3 conventional partition and 3,000 node hours to run in the C1 'GPU'
- WPDCNeu3-Neutronics Simulations for DEMO Diagnostics (PI Raul Luís): 50,700 node hours to run on the A3 conventional partition.
- REFMULx-Parallel 3D reflectometry simulation (PI Filipe da Silva): 40,000 node hours to run on the A3 conventional partition and 3,000 node hours to run in the C1 'GPU' partition.
- REFTURB-3D full-wave simulations of 3D turbulence: application to ASDEX Upgrade (PI José Vicente): 40,000 node hours on the A3 conventional partition.
