Pasar al contenido principal

DESAFIOS ACTUALES EN HPC: ARQUITECTURAS, SOFTWARE Y APLICACIONES

PID2019-104184RB-I00

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal de I+D+i Orientado a los Retos de la Sociedad
Convocatoria Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Año convocatoria 2019
Plan Estatal 2017-2020 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSIDADE DA CORUÑA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 11
Encontrada(s) 1 página(s)

RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Parallel Construction of RNA Databases for Extensive lncRNA-RNA Interaction Prediction

  • Amatria Barral, Iñaki
  • González-Domínguez, Jorge
  • Touriño, Juan
Poster presentado en: 38th ACM/SIGAPP Symposium on Applied Computing (SAC ’23), March 27-April 2, 2023, Tallinn, Estonia © Autors | ACM 2023. This is the author's version of the work. It is posted here for your personal use. Not for redistribution. The definitive Version of Record was published in Proceedings of the 38th ACM/SIGAPP Symposium on Applied Computing, https://doi.org/10.1145/3555776.3577772, [Abstract] Long non-coding RNA sequences (lncRNAs) have completely changed how scientists approach genetics. While some believe that many lncRNAs are results of spurious transcriptions, recent evidence suggests that there exist thousands of them and that they have functions and regulate key biological processes. For the experimental characterization of lncRNAs, many tools that try to predict their interactions with other RNAs have been developed. Some of the fastest and more accurate tools, however, require a slow database construction step prior to the identification of interaction partners for each lncRNA. This paper presents a novel and efficient parallel database construction procedure. Benchmarking results on a 16-node multicore cluster show that our parallel algorithm can build databases up to 318 times faster than other tools in the market using just 256 CPU cores. All the code developed in this work is available to download at GitHub under the MIT License (https://github.com/UDC-GAC/pRIblast)., This work was supported by the Ministry of Science and Innovation of Spain (PID2019-104184RB-I00 / AEI / 10.13039 / 501100011033), and by Xunta de Galicia and FEDER funds of the European Union (Centro de Investigación de Galicia accreditation 2019-2022, ref. ED431G 2019/01)., Xunta de Galicia; ED431G 2019/01, https://github.com/UDC-GAC/pRIblast
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Self-adaptive Cooperation Scheme in a Parallel ACO Algorithm for Binary Combinatorial Problems

  • Prado-Rodríguez, Roberto
  • González, Patricia
  • Banga, Julio R.
  • Doallo, Ramón
Cursos e Congresos, C-155, [Abstract] The ant colony optimization (ACO) is widely used for combinatorial optimization problems, although it can suffer from fast convergence to local minima. In order to provide a versatile implementation of ACO, we present a parallel multicolony strategy with an improved cooperation scheme for binary combinatorial problems. Our proposal is based on a self-adaptive method, which assigns appropriate run-time cooperation levels to each problem based on its size and available computational resources. We evaluate this proposal with problems with different levels of cooperation and number of processes. All these configurations combined show its flexibility as a versatile solver for this type of problems, Xunta de Galicia; ED431G 2019/01, Xunta de Galicia; ED431C 2021/30, RPR, PGG and RDB acknowledges funding from Grants PID2019-104184RB-I00 and PID2022-136435NB-I00, funded by MCIN/AEI/ 10.13039/501100011033, PID2022 also funded by ”ERDF. A way of making Europe”, EU; Xunta de Galicia and FEDER funds of the EU (Centro de Investigación de Galicia accreditation 2019–2022, ref. ED431G 2019/01; Consolidation Program of Competitive Reference Groups, ref. ED431C 2021/30). JRB acknowledges funding from the Ministry of Science and Innovation of Spain MCIN /AEI / 10.13039/501100011033 through grant PID2020-117271RB- C22 (BIODYNAMICS). Authors also acknowledge the Galician Supercomputing Center (CESGA) for the access to its facilities
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

An approach to support generic topologies in distributed PSO algorithms in Spark

  • Pardo, Xoán C.
  • González, Patricia
  • Banga, Julio R.
  • Doallo, Ramón
[Abstract] Particle Swarm Optimization (PSO) is a popular population-based search algorithm that has been applied to all kinds of complex optimization problems. Although the performance of the algorithm strongly depends on the social topology that determines the interaction between the particles during the search, current Metaheuristic Optimization Frameworks (MOFs) provide limited support for topologies. In this paper, we present an approach to support generic topologies in distributed PSO algorithms within a framework for the development and execution of population- based metaheuristics in Spark, which is currently under development., Xunta de Galicia; ED431C 2021/30, This work was supported by the Ministry of Science and Innovation of Spain (PID2019-104184RB-I00, AEI, 10.13039/501100011033) and by Xunta de Galicia (Consolidation Program of Competitive Reference Groups, ref. ED431C 2021/30).
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Performance Optimization of a Parallel Error Correction Tool

  • Martínez-Sánchez, Marco
  • Expósito, Roberto R.
  • Touriño, Juan
Presented at the 4th XoveTIC Conference, A Coruña, Spain, 7–8 October 2021., [Abstract] Due to the continuous development in the field of Next Generation Sequencing (NGS) technologies that have allowed researchers to take advantage of greater genetic samples in less time, it is a matter of relevance to improve the existing algorithms aimed at the enhancement of the quality of those generated reads. In this work, we present a Big Data tool implemented upon the open-source Apache Spark framework that is able to execute validated error-correction algorithms at an improved performance. The experimental evaluation conducted on a multi-core cluster has shown significant improvements in execution times, providing a maximum speedup of 9.5 over existing error correction tools when processing an NGS dataset with 25 million reads., This research was funded by the Ministry of Science and Innovation of Spain (PID2019-104184RB-I00/AEI/10.13039/501100011033), and by Xunta de Galicia and FEDER funds of the European Union (Centro de Investigación de Galicia accreditation 2019-2022, ref. ED431G2019/01; Consolidation Program of Competitive Reference Groups, ref. ED431C 2021/30), Xunta de Galicia; ED431G2019/01, Xunta de Galicia; ED431C 2021/30
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

The New UPC++ DepSpawn High Performance Library for Data-Flow Computing with Hybrid Parallelism

  • Fraguela, Basilio B.
  • Andrade, Diego
This versión of the contribution has been accepted for publication, after peer review but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: https://doi.org/10.1007/978-3-031-08751-6_55. Use of this Accepted Version is subject to the publisher’s Accepted Manuscript terms of use https://www.springernature.com/gp/open-research/policies/accepted-manuscript-terms, [Abstract] Data-flow computing is a natural and convenient paradigm for expressing parallelism. This is particularly true for tools that automatically extract the data dependencies among the tasks while allowing to exploit both distributed and shared memory parallelism. This is the case of UPC++ DepSpawn, a new task-based library developed on UPC++ (Unified Parallel C++), a library for parallel computing on a Partitioned Global Address Space (PGAS) environment, and the well-known Intel TBB (Threading Building Blocks) library for multithreading. In this paper we present and evaluate the evolution of this library after changing its engine for shared memory parallelism and adapting it to the newest version of UPC++, which differs very strongly from the original version on which UPC++ DepSpawn was developed. As we will see, while keeping the same high level of programmability, the new version is on average 9.3% faster than the old one, the maximum speedup being 66.3%., This research was supported by the Ministry of Science and Innovation of Spain (PID2019-104184RB-I00/AEI/10.13039/501100011033), and by the Xunta de Galicia co-founded by the European Regional Development Fund (ERDF) under the Consolidation Programme of Competitive Reference Groups (ED431C 2021/30). We acknowledge also the support from the Centro Singular de Investigación de Galicia “CITIC”, funded by Xunta de Galicia and the European Union (European Regional Development Fund- Galicia 2014–2020 Program), by grant ED431G 2019/01. Finally, we acknowledge the Centro de Supercomputación de Galicia (CESGA) for the use of their computers, Xunta de Galicia; ED431C 2021/30, Xunta de Galicia; ED431G 2019/01
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Learning Crop Management by Reinforcement: Gym-DSSAT

  • Gautron, Romain
  • Padrón, Emilio J.
  • Preux, Philippe
  • Bigot, Julien
  • Maillard, Odalric-Ambrym
  • Hoogenboom, Gerrit
  • Teigny, Julien
Traballo presentado no: 2nd AAAI Workshop on AI for Agriculture and Food Systems (AIAFS 2023), 12-14 Febr. 2023, Washington DC, United States / Online. Poster dispoñible en: https://hdl.handle.net/2183/48134 ., [Abstract]: We introduce gym-DSSAT, a gym environment for crop management tasks, that is easy to use for training Rein- forcement Learning (RL) agents. gym-DSSAT is based on DSSAT, a state-of-the-art mechanistic crop growth simulator. We modify DSSAT so that an external software agent can in- teract with it to control the actions performed in a crop field during a growing season. The RL environment provides pre- defined decision problems without having to manipulate the complex crop simulator. We report encouraging preliminary results on a use case of nitrogen fertilization for maize. This work opens up opportunities to explore new sustainable crop management strategies with RL, and provides RL researchers with an original set of challenging tasks to investigate., The authors acknowledge the PDI team, in particular Karol Sierocinski, for their help. They also thank the DSSAT development team, especially Cheryl Porter for her continuous support. We acknowledge the Consultative Group for International Agricultural Research (CGIAR) Platform for Big Data in Agriculture and we especially thank Brian King. Ph. Preux, and O-A. Maillard acknowledge the support of the Métropole Européenne de Lille (MEL), ANR, Inria, Université de Lille, through the AI chair Apprenf number R-PILOTE-19-004-APPRENF. We acknowledge the support of the AIDA team at CIRAD and the outstanding working environment provided by Inria in the Scool research group. Emilio J. Padrón’s work was partially supported through the research projects PID2019-104184RBI00 funded by MCIN/AEI/10.13039/501100011033, and ED431C 2021/30, ED431F 2021/11 and ED431G 2019/01 funded by Xunta de Galicia. We thank Debabrota Basu for his help in the writing of this paper., Xunta de Galicia; ED431C 2021/30, Xunta de Galicia; ED431F 2021/11, Xunta de Galicia; ED431G 2019/01
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Probing the Efficacy of Hardware-Aware Weight Pruning to Optimize the SpMM routine on Ampere GPUs

  • Castro, Roberto L.
  • Andrade, Diego
  • Fraguela, Basilio B.
[Abstract]: The Deep Learning (DL) community found in pruning techniques a good way to reduce the models' resource and energy consumption. These techniques lead to smaller sparse models, but sparse computations in GPUs only outperform their dense counterparts for extremely high levels of sparsity. However, pruning up to such sparsity levels can seriously harm the accuracy of the Neural Networks (NNs). To alleviate this, novel performance-aware pruning techniques favor the generation of more regular sparse matrices that can improve the exploitation of the underlying hardware. Nevertheless, an important drawback is that these techniques heavily condition the location of the non-pruned values, which can strongly degrade the accuracy of the models. This paper focuses on improving the performance of the SpMM routine on DL workloads by combining performance-aware pruning with pruning-independent SpMM kernels to relax input-format constraints. We start with a microarchitecture-level performance study of SOTA SpMM implementations to identify their main bottlenecks and flaws. Then, the paper centers on maximizing the performance of the routine by adjusting the parameters of performance-aware pruning techniques to the hardware properties. This second study explains the intrinsic causes of the observed performance results. We show that, following this approach, a generic SpMM routine can perform up to 49% and 77% better for half and single precision, respectively, than using non-performance-aware pruning, providing speedups over cuBlas of up to 1.87× and 4.20×, respectively. Additionally, the performance achieved on half precision is boosted with a new Ampere-ready specialized implementation for the columnvector sparse format, CLASP, which achieves a 2.42× speedup over cuBlas. Finally, we also introduce ad-colPrune, a novel pruning technique that widens the design space of possible trade-offs between performance and accuracy. © 2022 Association for Computing Machinery., Xunta de Galicia; ED431C 2021/30, Xunta de Galicia; ED431G 2019/01, This research was supported by the Ministry of Science and Innovation of Spain (PID2019-104184RB-I00, AEI, 10.13039/501100011033), the Ministry of Education (predoctoral grant of Roberto L. Castro, FPU19/03974), and by Xunta de Galicia under the Consolidation Program of Competitive Reference Groups (ED431C 2021/30). We also acknowledge the support from CITIC, funded by Xunta de Galicia and FEDER funds of the EU (Centro de Investigación de Galicia accreditation 2019-2022, ED431G 2019/01). Finally, we acknowledge the Centro de Supercomputación de Galicia (CESGA) for the use of their computers.
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

VENOM: A Vectorized N:M Format for Unleashing the Power of Sparse Tensor Cores

  • Castro, Roberto L.
  • Ivanov, Andrei
  • Andrade, Diego
  • Ben-Nun, Tal
  • Fraguela, Basilio B.
  • Hoefler, Torsten
© 2023 Autores | ACM. This is the author's version of the work. It is posted here for your personal use. Not for redistribution. The definitive Version of Record was published in International Conference for High Performance Computing, Networking, Storage and Analysis, https://doi.org/10.1145/3581784.3607087, [Abstract]: The increasing success and scaling of Deep Learning models demands higher computational efficiency and power. Sparsification can lead to both smaller models as well as higher compute efficiency, and accelerated hardware is becoming available. However, exploiting it efficiently requires kernel implementations, pruning algorithms, and storage formats, to utilize hardware support of specialized sparse vector units. An example of those are the NVIDIA's Sparse Tensor Cores (SPTCs), which promise a 2× speedup. However, SPTCs only support the 2:4 format, limiting achievable sparsity ratios to 50%. We present the V:N:M format, which enables the execution of arbitrary N:M ratios on SPTCs. To efficiently exploit the resulting format, we propose Spatha, a high-performance sparse-library for DL routines. We show that Spatha achieves up to 37× speedup over cuBLAS. We also demonstrate a second-order pruning technique that enables sparsification to high sparsity ratios with V:N:M and little to no loss in accuracy in modern transformers., This research was supported by the Ministry of Science and Innovation of Spain (grants PID2019-104184RB-I00 and PID2022-136435NB-I00, funded by MCIN/AEI/ 10.13039/501100011033, PID2022 also funded by "ERDF A way of making Europe", EU), the Ministry of Education (predoctoral grant of Roberto L. Castro, FPU19/03974), by Xunta de Galicia under the Consolidation Program of Competitive Reference Groups (ED431C 2021/30), and ERC grant PSAP, no. 101002047. We also acknowledge the support from CITIC, funded by Xunta de Galicia and FEDER funds of the EU (Centro de Investigación de Galicia accreditation 2019-2022, ED431G 2019/01). Finally, we thank the Swiss National Supercomputing Center (CSCS) and the Centro de Supercomputación de Galicia (CESGA) for the use of their computers., Xunta de Galicia; ED431C 2021/30, Xunta de Galicia; ED431G 2019/01
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Applying dynamic balancing to improve the performance of MPI parallel genomics applications

  • Fernández-Fraga, Alejandro
  • González-Domínguez, Jorge
  • Martín, María J.
© ACM 2024. This is the author's version of the work. It is posted here for your personal use. Not for redistribution. The definitive Version of Record was published in Proceedings of the 39th ACM/SIGAPP Symposium on Applied Computing (SAC '24)., [Absctract]: Genomics applications are becoming more and more important in the field of bioinformatics, as they allow researchers to extract meaningful information from the huge amount of data generated by the new sequencing technologies. The analysis of these data is a very time consuming task and, therefore, the use of High Performance Computing (HPC) and parallel processing techniques is essential. Although the structure of these applications can be easily adapted to parallel systems by distributing the data to be processed among the available processors, load imbalance is a usual cause of performance degradation. In this paper we propose a dynamic load balancing method based on MPI RMA one-sided communications to minimize the synchronization among processes and the overhead due to communications while improving the workload balance. The strategy is applied, as a case study, to ParRADMeth, an MPI/OpenMP parallel application for the identification of Differential Methylated Regions (DMRs). Results show that the new version of the tool outperforms the previous one in all cases, achieving high performance and scalability. For example, our approach is up to 243 times faster than the sequential version and 1.74 times faster than the previous parallel version when processing a real dataset on a cluster with 8 nodes, each one with 32 CPU cores., This work has been supported by grants PID2019-104184RB-I00 and PID2022-136435NB-I00, both grants funded by MCIN/AEI/ 10.13039/501100011033, PID2022 also funded by "ERDF A way of making Europe", EU; the Ministry of Universities of Spain under grant FPU21/03408; and by Xunta de Galicia and FEDER funds (Centro de Investigación de Galicia accreditation 2019-2022 and Consolidation Program of Competitive Reference Groups, under Grants ED431G 2019/01 and ED431C 2021/30, respectively), Xunta de Galicia; ED431G 2019/01, Xunta de Galicia; ED431C 2021/30
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Accelerating the quality control of genetic sequences through stream processing

  • Castellanos Rodríguez, Óscar
  • Expósito, Roberto R.
  • Touriño, Juan
This is the author's version of the work. It is posted here for your personal use. Not for redistribution. The definitive Version of Record was published in Proceedings of the 38th ACM/SIGAPP Symposium on Applied Computing (SAC '23). Association for Computing Machinery, New York, NY, USA, 398–401. https://doi.org/10.1145/3555776.3577785., [Abstract]: Quality control of DNA sequences is an important data preprocessing step in many genomic analyses. However, all existing parallel tools for this purpose are based on a batch processing model, needing to have the complete genetic dataset before processing can even begin. This limitation clearly hinders quality control performance in those scenarios where the dataset must be downloaded from a remote repository and/or copied to a distributed file system for its parallel processing. In this paper we present SeQual-Stream, a Big Data tool that allows performing quality control on genomic datasets in a fast, distributed and scalable way. To do so, our tool relies on the Apache Spark framework and the Hadoop Distributed File System (HDFS) to fully exploit the stream paradigm and accelerate the preprocessing of large datasets as they are being downloaded and/or copied to HDFS. The experimental results have shown significant improvements when compared to a batch processing tool, providing a maximum speedup of 2.7x., This work was supported by the Ministry of Science and Innovation of Spain (PID2019-104184RB-I00 / AEI / 10.13039 / 501100011033), and by Xunta de Galicia and FEDER funds of the European Union (Centro de Investigación de Galicia accreditation 2019-2022, ref. ED431G 2019/01; Consolidation Program of Competitive Reference Groups, ref. ED431C 2021/30)., Xunta de Galicia; ED431G 2019/01, Xunta de Galicia; ED431C 2021/30
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Real-time classification of handball game situations

  • Cabado, Bruno
  • Guijarro-Berdiñas, Bertha
  • Padrón, Emilio J.
Presented at: 34th IEEE International Conference on Tools with Artificial Intelligence, ICTAI 2022, Virtual, Online, 31 October 2022 through 2 November 2022, This version of the article has been accepted for publication, after peer review. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. The Version of Record is available online at: https://doi.org/10.1109/ICTAI56018.2022.00106, [Abstract]: During the broadcast of sporting events, certain situations such as a penalty or a time-out occur, for which a specific action is required. In traditional broadcasting, many people are implied in making decisions based on what is happening at any given moment. To broadcast quality and entirely automatically matches it is necessary to be able to classify the important situations and then make decisions based on them. This paper presents a solution based on deep learning which is able to classify the main states of a handball match. The generated model has been trained using 127,600 images of 13 local team matches. On a test set of 118,129 images of other 7 matches, it is able to classify these situations with an accuracy of 98.6% in only 4 milliseconds, allowing to analyze the state of the game in real time. The full pipeline takes only 34.04 milliseconds using GPU acceleration, processing more than 25 frames per seconds., This work was supported in part by Spanish Agencia Estatal de Investigación (PID2019-109238GB-C2/AEI/ 10.13039/501100011033 and PID2019-104184RBI00/AEI/10.13039/501100011033) and the Xunta de Galicia (ED431C 2018/34, ED431C 2021/30 and ED431F 2021/11). Bruno Cabado wish to thanks the Axencia Galega de Innovación the grant received through its Industrial Doctorate program (23/IN606D/2021/2612054). CITIC is funded by Xunta de Galicia (ED431G 2019/01) and ERDF funds., Xunta de Galicia; ED431C 2018/34, Xunta de Galicia; ED431C 2021/30, Xunta de Galicia; ED431F 2021/11, Xunta de Galicia; IN606D/2021/2612054, Xunta de Galicia; ED431G 2019/01