Boosting Brain Simulations for the World’s Fastest Supercomputers

Research theses
Abstract

Arbor is a brain simulation library that enables the multiscale exploration of large neural networks with detailed biophysical modeling. However, the execution of biomedically relevant experiments requires vast computational resources, putting the scalability of both HPC systems and software to the test. This internship project will explore the optimization of Arbor to improve its performance and scalability on flagship exascale systems. The intern will contribute to the analysis and improvement of parallel execution, including communication patterns, load balancing strategies, and the efficient use of modern GPU architectures. The work will emphasize performance, scalability, and best practices in high‑performance scientific software development.

Keywords
Parallel programming
GPU programming
High Performance Computing
Performance optimisation
ERC sector(s)
PE Physical Sciences and Engineering
Fields of study
Math and Computer Science
Others
Thesis supervisor
Name supervisor
Esteban Stafford
E-mail
eunice@unican.es
Department/Faculty/School/Institute/Area/Division NAME
Department of Computer and Electronic Engineering
Name of the host University
University of Cantabria (UC)
EUNICE partner e-mail of destination Research
area.eunice@unican.es
Country
Spain
Student profile
Minimal language knowledge requisite
English B2
Additional info
Length of the research internship
3 months
Financial support available (other than E+)
No
Research interests for cooperation opportunities
Optimisation of computationally intensive applications at multiple levels (vector, thread, cluster, GPU) and exploration of architectural techniques to improve performance and efficiency.