Marta García-Gasulla

ORCID: 0000-0003-3682-9905
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Parallel Computing and Optimization Techniques
  • Advanced Data Storage Technologies
  • Distributed and Parallel Computing Systems
  • Cloud Computing and Resource Management
  • Computational Fluid Dynamics and Aerodynamics
  • Combustion and flame dynamics
  • Advanced Numerical Methods in Computational Mathematics
  • Advanced Memory and Neural Computing
  • Neural dynamics and brain function
  • Lattice Boltzmann Simulation Studies
  • Neuroscience and Neural Engineering
  • Particle Dynamics in Fluid Flows
  • Electromagnetic Simulation and Numerical Methods
  • Magnetic confinement fusion research
  • Simulation Techniques and Applications
  • Embedded Systems Design Techniques
  • Peer-to-Peer Network Technologies
  • Advanced Combustion Engine Technologies
  • Advanced Neural Network Applications
  • Fluid Dynamics and Turbulent Flows
  • Wind and Air Flow Studies
  • Nuclear reactor physics and engineering
  • Matrix Theory and Algorithms
  • Aerodynamics and Fluid Dynamics Research
  • Radiation Effects in Electronics

Universitat Politècnica de Catalunya
2012-2024

Barcelona Supercomputing Center
2011-2024

Argonne National Laboratory
2024

Universitat Ramon Llull
2018

Centre Européen de Recherche et de Formation Avancée en Calcul Scientifique
2008-2010

Efficient numerical tools coupled with high-performance computers, have become a key element of the design process in fields energy supply and transportation. However flow phenomena that occur complex systems such as gas turbines aircrafts are still not understood mainly because models needed. In fact, most computational fluid dynamics (CFD) predictions found today industry focus on reduced or simplified version real system (such periodic sector) usually solved steady-state assumption. This...

10.1088/1749-4699/2/1/015003 article EN Computational Science & Discovery 2009-11-12

A current trend in HPC systems is the utilization of architectures with SIMD or vector extensions to exploit data parallelism. There are several ways take advantage such modern architectures, each a different impact on code and its portability. For example, use intrinsics, guided vectorization via pragmas, compiler autovectorization. Our objectives maximize efficiency minimize specialization. To achieve these objectives, we rely We leverage set hardware software tools that allow us analyze...

10.1109/ipdps57955.2024.00047 article EN 2022 IEEE International Parallel and Distributed Processing Symposium (IPDPS) 2024-05-27

Since the appearance of Docker in 2013, container technologies for computers have evolved and gained importance cloud data centers. However, adoption containers High-Performance Computing (HPC) centers is still under discussion: on one hand, ease portability very well accepted; other performance penalties security issues introduced by added software layers are often scrutiny. little evaluation large production HPC codes running available, we provide this paper a comparative study using...

10.1109/ipdps.2019.00066 article EN 2022 IEEE International Parallel and Distributed Processing Symposium (IPDPS) 2019-05-01

In this study, we develop new reverse engineering (RE) techniques to identify the organization of synaptic inputs generating firing patterns populations neurons. We tested these in silico allow rigorous evaluation their effectiveness, using remarkably extensive parameter searches enabled by massively-parallel computation on supercomputers. chose spinal motoneurons as our target neural system, since process all motor commands and have well-established input-output properties. One set...

10.7554/elife.90624.3 article EN cc-by eLife 2024-10-16

Arm-based system in HPC are a reality since more than decade. However, when new chip enters the market always implies challenges, not only at ISA level, but also with regards to SoC integration, memory subsystem, board node interconnection, and finally OS all layers of software (compiler libraries). Guided by procurement an NVIDIA Grace cluster within deployment MareNostrum 5, emulating approach scientist who needs migrate its scientific research system, we evaluated five complex...

10.1145/3636480.3637284 article EN 2024-01-08

In this work, we evaluate the performance of SeisSol, a simulator seismic wave phenomena and earthquake dynamics, on RISC-V-based system utilizing vector processing unit. We focus GEMM libraries address their limited ability to leverage long architectures by developing batched DGEMM library in plain C. This achieves speedups ranging from approximately 3.5x 32.6x compared reference implementation. then integrate approach into SeisSol application, ensuring portability across different CPU...

10.48550/arxiv.2501.06175 preprint EN arXiv (Cornell University) 2025-01-10

This work presents a parallel numerical strategy to transport Lagrangian particles in fluid using dynamic load balance strategy. Both and particle solvers are parallel, with two levels of parallelism. The first level is based on substructuring technique uses message passing interface (MPI) as the communication library; second consists OpenMP pragmas for loop parallelisation at node level. When dealing transient flows, there exist main alternatives address coupling these solvers. On one hand,...

10.1080/10618562.2016.1227070 article EN International journal of computational fluid dynamics 2016-07-02

We present LeWI: a novel load balancing algorithm, that can balance applications with very different patterns of imbalance. Our algorithm fine grain imbalances, non iterative and irregular To achieve this LeWI reassigns the computational resources blocked processes to other more loaded. have implemented within DLB Dynamic Load Balancing Library developed by us. helps parallel programming models make most power available minimum effort. It solves imbalance among in two levels parallelism...

10.1109/icpp.2009.56 article EN International Conference on Parallel Processing 2009-09-01

The fluid-structure interaction (FSI) problem has received great attention in the last few years, mainly because it is present many physical systems, industrial applications, and almost every biological system. In parallel computational field, outstanding advances have been achieved for individual components of problem, allowing, instance, simulations around complex geometries at very high Reynolds numbers or contraction a beating heart. However, not an easy task to combine both fields,...

10.1137/17m1138868 article EN SIAM Journal on Scientific Computing 2018-01-01

The recent rapid growth of the data-flow programming paradigm enabled development specific architectures, e.g., for machine learning. most known example is Tensor Processing Unit (TPU) by Google. Standard data-centers, however, still can not foresee large partitions dedicated to learning architectures. Within High-Performance Computing (HPC) clusters are highly parallel machines targeting a broad class compute-intensive workflows, as such they be used tackling challenges. On top this, HPC...

10.1109/ccgrid.2019.00067 article EN 2019-05-01

Computational fluid and particle dynamics (CFPD) simulations are of paramount importance for studying improving drug effectiveness. requirements CFPD codes demand high-performance computing (HPC) resources. For these reasons, we introduce evaluate in this article system software techniques performance tolerating load imbalance on a state-of-the-art production code. We demonstrate benefits Intel-, IBM- Arm-based HPC technologies ranked the Top500 supercomputers, showing using mechanisms...

10.1177/1094342019842919 article EN The International Journal of High Performance Computing Applications 2019-04-23

AI is one of the biggest megatrends towards 4th industrial revolution. Although these technologies promise business sustainability as well product and process quality, it seems that ever-changing market demands, complexity fair concerns about privacy, impede broad application reuse Artificial Intelligence (AI) models across industry. To break entry barriers for unleash its full potential, knowlEdge project will develop a new generation methods, systems, data management infrastructure....

10.1109/indin45523.2021.9557410 article EN 2022 IEEE 20th International Conference on Industrial Informatics (INDIN) 2021-07-21
Coming Soon ...