A5102992099- Parallel Computing and Optimization Techniques
- Distributed and Parallel Computing Systems
- Advanced Data Storage Technologies
- Cloud Computing and Resource Management
- Interconnection Networks and Systems
- Matrix Theory and Algorithms
- Green IT and Sustainability
- Scientific Computing and Data Management
- Embedded Systems Design Techniques
- Low-power high-performance VLSI design
- Distributed systems and fault tolerance
- VLSI and FPGA Design Techniques
- IoT and Edge/Fog Computing
- Transportation Planning and Optimization
- Optimization and Search Problems
- Algorithms and Data Compression
- Advanced Neural Network Applications
- Transportation and Mobility Innovations
- Data Management and Algorithms
- Advanced Numerical Methods in Computational Mathematics
- Image Retrieval and Classification Techniques
- Advanced Optimization Algorithms Research
- Radiation Effects in Electronics
- Medical Image Segmentation Techniques
- Machine Learning in Materials Science
Universidad de La Laguna
2015-2025
Universidade de Santiago de Compostela
1994-2001
Abstract This study evaluates the performance and portability of Weather Research Forecasting model in high-performance computing environments, comparing traditional baremetal deployments with containerized executions using Singularity. Experiments were conducted on TeideHPC system, focusing execution time, impact different compilers (Intel vs. GCC), parallelization strategies (MPI OpenMP). The results indicate that while Singularity introduces a overhead 11 to 15%, it offers significant...
Abstract Over the past few years, ARM has been dominant player in embedded systems and System-on-Chips (SoCs). With emergence of hardware platforms based on RISC-V architecture, a practical comparison focusing their energy efficiency performance is needed. In this study, our goal to comprehensively evaluate SoCs three different systems. We will conduct benchmark tests measure power consumption overall system performance. The results study are valuable developers researchers looking for most...
Abstract This document addresses some inherent problems in Machine Learning (ML), such as the high computational and energy costs associated with their implementation on IoT devices. It aims to study analyze performance efficiency of quantization an optimization method, well possibility training ML models directly device. Quantization involves reducing precision model weights activations while still maintaining acceptable levels accuracy. Using representative networks for facial recognition...
Abstract This work tackles the performance and energy consumption analysis of a legacy scientific application, VASP (Vienna Ab-initio Simulation Package), an application commonly used by physicists chemists for modeling materials at atomic scale. Many these applications have been implemented in Fortran, where metrics instrumentation is not straightforward. We obtained figures (execution time consumption) instrumenting source code using EML. measurement library has modified to introduce...
Modern standards in High Performance Computing (HPC) have started to consider energy consumption and power draw as a limiting factor. New more complex architectures been introduced HPC systems afford these new restrictions, include coprocessors such GPGPUs for intensive computational tasks. As increase heterogeneity, workload distribution becomes core problem achieve the maximum efficiency every component. We present Multi-Objective Dynamic Load Balancing (DLB) approach where several...
The advent of multicore systems, joined to the potential acceleration graphics processing units, alleviates some well known important architectural problems at expense a considerable increment programmability wall. heterogeneity, both and programming level same time, raises difficulties. Adapting existing code libraries such systems is fundamental problem. performance this affected by large interdependence between parallel architecture. We have developed dynamic load balancing library that...
Parallel computing in heterogeneous environments is drawing considerable attention due to the growing number of these kind systems. Adapting existing code and libraries such systems a fundamental problem. The performance this affected by large interdependence between parallel architectures. We have developed dynamic load balancing library that allows be adapted for wide variety problems. overhead introduced our system minimal cost programmer negligible. strategy was applied Dynamic...
Abstract Rapid growth of demand for remote computational power, along with high energy costs and infrastructure limits, has led to treating power usage as a primary constraint in data centers. Especially, recent challenges related development exascale systems or autonomous edge require tools that will limit consumption. This paper presents capping method allows operators quickly adjust the external conditions and, at same time, reduce consumption negative impact on performance applications....
Comparable to time performance models, it is now possible estimate based upon energy consumption for HPC systems. The predictive ability of the analytical modeling an interesting feature that motivates us approach this methodology case consumption. In paper, we present model predicting High Performance Linpack (HPL). derived can be used know in advance consumed by HPL over a target architecture, and integrated into schedulers operating systems or queue managers. We established experimental...
We studied potential changes in the sub-ventricular zone (SVZ) stem cell niche of senescence-accelerated mouse prone-8 (SAM-P8) aging model. Bromodeoxyuridine (BrdU) assays with longtime survival revealed a lower number label-retaining cells SAM-P8 SVZ compared SAM-Resistant 1 (SAM-R1) control strain. also found that signaling is attenuated and pool less responsive to self-renewal factor pigmented epithelium-derived (PEDF). Protein analysis demonstrated stable amounts PEDF ligand niche;...
This paper presents a new LogP-based model, called LoOgGP, which allows an accurate characterization of MPI applications based on microbenchmark measurements. model is extension LogP for long messages in both overhead and gap parameters perform linear dependency with message size. The LoOgGP has been fully integrated into modelling framework to obtain statistical models parallel applications, providing the analyst easy automatic tool parameter set assessment characterize communications. use...
Energy consumption is one of the main limiting factors for designing and deploying ultrascale systems. Therefore, this paper presents challenges trends associated with energy efficiency systems based on current activities working group "Energy Efficiency" in European COST Action Nesus IC1305. The analysis contains major areas that are related to studies systems: heterogeneous low power hardware architectures, monitoring at large scale, modeling simulation systems, energy-aware scheduling...
With energy efficiency one of the main challenges on way towards ultrascale systems, there is great need for access to high-quality consumption data. Such data would enable researchers and designers pinpoint inefficiencies at all levels computing stack, from whole nodes down critical regions code. However, measurement capabilities are often missing, significantly differ between platforms where they exist. A standard yet be established. To that end, this paper attempts an extensive survey...