ITT - High performance computing (HPC) cluster

The requirement of the National Physical Laboratory [NPL] is a high-performance computing (HPC) cluster to perform numerical simulations in the area of quantum technologies. This involves the emulation of noisy quantum computers, multi-scale device simulations, and automation of quantum technologies using machine learning (ML). Example software that will run on this HPC cluster includes quantum computing emulators (such as QuEST, Qiskit, Qulacs), machine learning software (such as PyTorch, TensorFlow), and materials modelling software (such as quantum espresso, questaal, siesta). This HPC cluster therefore needs to be able to run a wide range of open source and proprietary applications, and provide a parallel development and runtime environment for software built locally. The development languages to be used in include Fortran, C, C++, Python and Cuda.

To these aims the HPC cluster needs to include a combination of nodes without and nodes with Graphics Processing Units (GPUs) accelerator cards, with high-speed network connections between nodes. The solution is to include fully functioning installation at an appropriate facility in a dedicated computer room, to be provided by NPL. The detailed requirements are specified in Section 5, and include showing all cabinets, ethernet networking, PDUS, operating system etc required to build a complete solution, given that the NPL will be providing space, power, cooling, connection to the network infrastructure and remote console access to the head-node.