We have access to high-performance computing centres and dedicated support within Cardiff University.
Advanced Research Computing at Cardiff (ARCCA)
The ARCCA division at Cardiff University (with support from Supercomputing Wales) has a research computing resource, Hawk, which features 19,416 cores, an Infiniband interconnect and more than 1 PB of file storage.
A 3D stereo visualisation facility, featuring a 3.5m by 3.5m screen.
We undertake our Natural Environment Research Council (NERC) funded simulation research on ARCHER (a 118,080 processing core system), to be replaced in 2020 by ARCHER2, a 748,544 core system. The ARCHER National Supercomputer was funded by NERC and EPSRC, while its replacement National Supercomputer ARCHER2 was funded by UK Research and Innovation (UKRI).
How they help
The Mantle Geodynamics Research Group uses the facilities for its research. Their research is funded by Research Councils, Private Industry and Charities. This high-performance computing research involves high resolution numerical simulations of the dynamics of the deep Earth and how it affects the surface.
The facilities allow us to undertake sophisticated simulations of the dynamics, including simulations in 3D spherical geometry. These models include mantle circulation models (MCM) which incorporate plate motion history. The models can be directly compared, including by 3D visualisation, to actual observations for improved understanding.
The facilities have allowed us to better understand Earth evolution including influences on surface topography, magmatism and seismic structure:
- In Price et al., (2019) we showed that Earth’s deep water cycle has a strong level of self-regulation keeping the vast majority of the water in the interior over Earth's history.
- In Barry et al., (2017) we explained that the reason that the geochemistry of magmas from the ‘Indian’ and ‘Pacific’ domains remain distinct over 500 million years is because the subducted plates encircling the Pacific keep its heterogeneity within the basin, acting like curtains in the mantle.
- In Wolstencroft and Davies (2017) we showed that in the debate of how supercontinents are broken apart there is no need to choose between the active process of impinging underlying hot mantle plumes, and the passive process of them being pulled apart by distant plate extension forces, since we showed that both processes are simultaneously active.
- In Van Heck et al., (2016) we developed sophisticated modules to allow global-scale dynamic 3D spherical modelling to efficiently incorporate melting and isotope evolution, steps required for the previous publications.
- The dynamic simulations of subduction in Garel et al., (2014) demonstrated the importance of the rheology of both subducting and overriding lithosphere and asthenosphere in controlling the resulting subduction behaviour and ultimate subducted slab morphology.