Advanced Research Computing delivers new dedicated researcher expansion
28 January 2022
ARCCA recently completed work to extend Hawk through the integration of three new dedicated researcher-funded expansions further capitalising on the “pluggable infrastructure” architectural design of the system.
The new nodes are the first to take advantage of the Hawk extended network fabric delivering a new dedicated Intel Cascade Lake partition and two new dedicated AMD Rome partitions. Furthermore, a fourth installation of a new subsystem to support cancer genomics research has also passed acceptance testing and will be featured in a future article.
The new Intel sub-system, comprising of both storage and compute, will support research led by Professor Erminia Calabrese from the School of Physics and Astronomy. Professor Calabrese’s cosmological research combines theoretical work with statistical data analysis requiring the use of Hawk to help answer fundamental questions about the universe, previously delivering new insight into the age of the universe.
In addition, two new dedicated AMD partitions have been installed for Dr Alberto Roldan Martinez and Andrew Logsdail respectively from the School of Chemistry. Dr Andrew Logsdail was awarded funding for a prestigious UK Research and Innovation (UKRI) Future Leaders Fellowship and requires the use of Hawk to support the computational modelling of catalytic materials. Research carried out by Dr Alberto Roldan Martinez was awarded funding as part of a National Interdisciplinary Centre for Circular Chemical Economy (NIC3E) requiring the use of Hawk to support work to accelerate the design of durable and efficient catalysts.
In collaboration with partners Supercomputing Wales, ARCCA led the design, implementation and acceptance testing of these new sub-systems ensuring suitability with the computational research requirements of the researchers.
This latest expansion continues the growth of Hawk from its initial 8,040 computing cores at installation in August 2018 to beyond 20,000 computing cores.