Skip to main content

Resultados relacionados

Resultados totales (Incluyendo duplicados): 55
Encontrada(s) 1 página(s)

DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

How2Sign: A large-scale multimodal dataset for continuous American sign language

  • Duarte, Amanda
  • Palaskar, Shruti
  • Ventura, Lucas
  • Ghadiyaram, Deepti
  • DeHaan, Kenneth
  • Metze, Florian
  • Torres, Jordi
  • Giró i Nieto, Xavier
Trabajo presentado en la IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), celebrada de forma virtual del 19 al 25 de junio de 2021, One of the factors that have hindered progress in the areas of sign language recognition, translation, and production is the absence of large annotated datasets. Towards this end, we introduce How2Sign, a multimodal and multiview continuous American Sign Language (ASL) dataset, consisting of a parallel corpus of more than 80 hours of sign language videos and a set of corresponding modalities including speech, English transcripts, and depth. A three-hour subset was further recorded in the Panoptic studio enabling detailed 3D pose estimation. To evaluate the potential of How2Sign for real-world impact, we conduct a study with ASL signers and show that synthesized videos using our dataset can indeed be understood. The study further gives insights on challenges that computer vision should address in order to make progress in this field. Dataset website: http://how2sign.github.io/, This work received funding from Facebook through gifts to CMU and UPC; through projects TEC2016-75976-R, TIN2015- 65316-P, SEV-2015-0493 and PID2019-107255GB-C22 of the Spanish Government and 2017-SGR-1414 of Generalitat de Catalunya. This work used XSEDE’s “Bridges” system at the Pittsburgh Supercomputing Center (NSF award ACI1445606). Amanda Duarte has received support from la Caixa Foundation (ID 100010434) under the fellowship code LCF/BQ/IN18/11660029. Shruti Palaskar was supported by the Facebook Fellowship program.
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Continuous-action reinforcement learning for memory allocation in virtualized servers

  • Garrido Platero, Luis Ángel
  • Nishtala, Rajiv
  • Carpenter, Paul Matthew
In a virtualized computing server (node) with multiple Virtual Machines (VMs), it is necessary to dynamically allocate memory among the VMs. In many cases, this is done only considering the memory demand of each VM without having a node-wide view. There are many solutions for the dynamic memory allocation problem, some of which use machine learning in some form. This paper introduces CAVMem (Continuous-Action Algorithm for Virtualized Memory Management), a proof-of-concept mechanism for a decentralized dynamic memory allocation solution in virtualized nodes that applies a continuous-action reinforcement learning (RL) algorithm called Deep Deterministic Policy Gradient (DDPG). CAVMem with DDPG is compared with other RL algorithms such as Q-Learning (QL) and Deep Q-Learning (DQL) in an environment that models a virtualized node. In order to obtain linear scaling and be able to dynamically add and remove VMs, CAVMem has one agent per VM connected via a lightweight coordination mechanism. The agents learn how much memory to bid for or return, in a given state, so that each VM obtains a fair level of performance subject to the available memory resources. Our results show that CAVMem with DDPG performs better than QL and a static allocation case, but it is competitive with DQL. However, CAVMem incurs significant less training overheads than DQL, making the continuous-action approach a more cost-effective solution., This research is part of a project that has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 754337 (EuroEXA) and the European Union’s 7th Framework Programme under grant agreement number 610456 (Euroserver). It also received funding from the Spanish Ministry of Science and Technology (project TIN2015-65316-P), Generalitat de Catalunya (contract 2014-SGR-1272), and the Severo Ochoa Programme (SEV-2015-0493) of the Spanish Government., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Micro-architectural characterization of Apache Spark on batch and stream processing workloads

  • Awan, Ahsan
  • Brorsson, Mats
  • Vlassov, Vladimir
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
While cluster computing frameworks are continuously evolving to provide real-time data analysis capabilities, Apache Spark has managed to be at the forefront of big data analytics for being a unified framework for both, batch and stream data processing. However, recent studies on micro-architectural characterization of in-memory data analytics are limited to only batch processing workloads. We compare the micro-architectural performance of batch processing and stream processing workloads in Apache Spark using hardware performance counters on a dual socket server. In our evaluation experiments, we have found that batch processing and stream processing has same micro-architectural behavior in Spark if the difference between two implementations is of micro-batching only. If the input data rates are small, stream processing workloads are front-end bound. However, the front end bound stalls are reduced at larger input data rates and instruction retirement is improved. Moreover, Spark workloads using DataFrames have improved instruction retirement over workloads using RDDs., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Improving accuracy and speeding up document image classification through parallel systems

  • Ferrando Monsonís, Javier|||0000-0002-2637-0961
  • Domínguez, Juan Luis
  • Torres Viñals, Jordi|||0000-0003-1963-7418
  • García Fuentes, Raul
  • García Doménech, David
  • Garrido Miñambres, Daniel
  • Cortada, Jordi
  • Valero Cortés, Mateo|||0000-0003-2917-2482
This paper presents a study showing the benefits of the EfficientNet models compared with heavier Convolutional Neural Networks (CNNs) in the Document Classification task, essential problem in the digitalization process of institutions. We show in the RVL-CDIP dataset that we can improve previous results with a much lighter model and present its transfer learning capabilities on a smaller in-domain dataset such as Tobacco3482. Moreover, we present an ensemble pipeline which is able to boost solely image input by combining image model predictions with the ones generated by BERT model on extracted text by OCR. We also show that the batch size can be effectively increased without hindering its accuracy so that the training process can be sped up by parallelizing throughout multiple GPUs, decreasing the computational time needed. Lastly, we expose the training performance differences between PyTorch and Tensorflow Deep Learning frameworks., This work was partially supported by the Spanish Ministry of Science and Innovation and the European Regional Development Fund under contract TIN2015- 65316-P, by the BSC-CNS Severo Ochoa program SEV-2015-0493, and grant 2017-SGR-1414 by Generalitat de Catalunya and by the research agreement CaixaBank-BSC 2016-2021., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

MUSA: a multi-level simulation approach for next-generation HPC machines

  • Grass, Thomas
  • Allande, César
  • Armejach, Adrià
  • Rico, Alejandro
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
  • Labarta Mancho, Jesús José|||0000-0002-7489-4727
  • Valero Cortés, Mateo|||0000-0003-2917-2482
  • Casas, Marc|||0000-0003-4564-2093
  • Moretó Planas, Miquel|||0000-0002-9848-8758
The complexity of High Performance Computing (HPC) systems is increasing in the number of components and their heterogeneity. Interactions between software and hardware involve many different aspects which are typically not transparent to scientific programmers and system architects. Therefore, predicting the behavior of current scientific applications on future HPC infrastructures is a challenging task. In this paper we present MUSA, an end-to-end methodology that employs a multi-level simulation infrastructure. By combining different levels of abstraction, MUSA is able to model the communication network, microarchitectural details and system software interactions, providing different trade-offs in terms of simulation cost and accuracy. We compare detailed MUSA simulations with native executions of up to 2,048 cores and find relative errors that are within 10% in the common case. In addition, we use MUSA to simulate up to 16,384 cores and successfully identify scalability bottlenecks due to different factors, e.g. memory contention or load imbalance. We also compare different system configurations, showing how MUSA can help system designers to assess the usefulness of future technologies in next-generation HPC machines., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

STT-MRAM for real-time embedded systems: performance and WCET implications

  • Asifuzzaman, Kazi
  • Fernández, Mikel
  • Radojković, Petar
  • Abella Ferrer, Jaume|||0000-0001-7951-4028
  • Cazorla Almeida, Francisco Javier
STT-MRAM is an emerging non-volatile memory quickly approaching DRAM in terms of capacity, frequency and device size. Intensified efforts in STT-MRAM research by the memory manufacturers may indicate a revolution with STT-MRAM memory technology is imminent, and therefore it is essential to perform system level research to explore use-cases and identify computing domains that could benefit from this technology. Special STT-MRAM features such as intrinsic radiation hardness, non-volatility, zero stand-by power and capability to function in extreme temperatures makes it particularly suitable for aerospace, avionics and automotive applications. Such applications often have real-time requirements --- that is, certain tasks must complete within a strict deadline. Analyzing whether this deadline is met requires Worst Case Execution Time (WCET) Analysis, which is a fundamental part of evaluating any real-time system. In this study, we investigate the feasibility of using STT-MRAM in real-time embedded systems by analyzing average system performance impact and WCET implications., This work was supported by BSC, Spanish Government through Programa Severo Ochoa (SEV-2015-0493), by the Spanish Ministry of Science and Technology through TIN2015-65316-P project and by the Generalitat de Catalunya (contracts 2014-SGR-1051 and 2014-SGR-1272). This work has also received funding from the European Union’s Horizon 2020 research and innovation programme under ExaNoDe project (grant agreement No 671578). Jaume Abella was partially supported by the Ministry of Economy and Competitive-ness under Ramon y Cajal postdoctoral fellowship RYC-2013-14717., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

TaskPoint: sampled simulation of task-based programs

  • Grass, Thomas Dieter
  • Rico, Alejandro
  • Casas, Marc|||0000-0003-4564-2093
  • Moretó Planas, Miquel|||0000-0002-9848-8758
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
Sampled simulation is a mature technique for reducing simulation time of single-threaded programs, but it is not directly applicable to simulation of multi-threaded architectures. Recent multi-threaded sampling techniques assume that the workload assigned to each thread does not change across multiple executions of a program. This assumption does not hold for dynamically scheduled task-based programming models. Task-based programming models allow the programmer to specify program segments as tasks which are instantiated many times and scheduled dynamically to available threads. Due to system noise and variation in scheduling decisions, two consecutive executions on the same machine typically result in different instruction streams processed by each thread. In this paper, we propose TaskPoint, a sampled simulation technique for dynamically scheduled task-based programs. We leverage task instances as sampling units and simulate only a fraction of all task instances in detail. Between detailed simulation intervals we employ a novel fast-forward mechanism for dynamically scheduled programs. We evaluate the proposed technique on a set of 19 task-based parallel benchmarks and two different architectures. Compared to detailed simulation, TaskPoint accelerates architectural simulation with 64 simulated threads by an average factor of 19.1 at an average error of 1.8% and a maximum error of 15.0%., This work has been supported by the Spanish Government (Severo Ochoa grants SEV2015-0493, SEV-2011-00067), the Spanish Ministry of Science and Innovation (contract TIN2015-65316-P), Generalitat de Catalunya (contracts 2014-SGR-1051 and 2014-SGR-1272), the RoMoL ERC Advanced Grant (GA 321253), the European HiPEAC Network of Excellence and the Mont-Blanc project (EU-FP7-610402 and EU-H2020-671697). M. Moreto has been partially supported by the Ministry of Economy and Competitiveness under Juan de la Cierva postdoctoral fellowship JCI-2012-15047. M. Casas is supported by the Ministry of Economy and Knowledge of the Government of Catalonia and the Cofund programme of the Marie Curie Actions of the EUFP7 (contract 2013BP B 00243). T.Grass has been partially supported by the AGAUR of the Generalitat de Catalunya (grant 2013FI B 0058)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Accelerating K-mer Frequency Counting with GPU and Non-Volatile Memory

  • Cadenelli, Nicola
  • Polo Cantero, José|||0000-0002-9917-1795
  • Carrera Pérez, David|||0000-0003-4898-3424
The emergence of Next Generation Sequencing (NGS) platforms has increased the throughput of genomic sequencing and in turn the amount of data that needs to be processed, requiring highly efficient computation for its analysis. In this context, modern architectures including accelerators and non-volatile memory are essential to enable the mass exploitation of these bioinformatics workloads. This paper presents a redesign of the main component of a state-of-the-art reference-free method for variant calling, SMUFIN, which has been adapted to make the most of GPUs and NVM devices. SMUFIN relies on counting the frequency of k-mers (substrings of length k) in DNA sequences, which also constitutes a well-known problem for many bioinformatics workloads, such as genome assembly. We propose techniques to improve the efficiency of k-mer counting and to scale-up workloads like SMUFIN that used to require 16 nodes of Marenostrum 3 to a single machine with a GPU and NVM drives. Results show that although the single machine is not able to improve the time to solution of 16 nodes, its CPU time is 7.5x shorter than the aggregate CPU time of the 16 nodes, with a reduction in energy consumption of 5.5x. © 2017 IEEE., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Improving HPC system throughput and response time using memory disaggregation

  • Vieira Zacarias, Felippe
  • Carpenter, Paul Matthew
  • Petrucci, Vinicius
© 2021 IEEE. 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., HPC clusters are cost-effective, well understood, and scalable, but the rigid boundaries between compute nodes may lead to poor utilization of compute and memory resources. HPC jobs may vary, by orders of magnitude, in memory consumption per core. Thus, even when the system is provisioned to accommodate normal and large capacity nodes, a mismatch between the system and the memory demands of the scheduled jobs can lead to inefficient usage of both memory and compute resources. Disaggregated memory has recently been proposed as a way to mitigate this problem by flexibly allocating memory capacity across cluster nodes. This paper presents a simulation approach for at-scale evaluation of job schedulers with disaggregated memories and it introduces a new disaggregated-aware job allocation policy for the Slurm resource manager. Our results show that using disaggregated memories, depending on the imbalance between the system and the submitted jobs, a similar throughput and job response time can be achieved on a system with up to 33% less total memory provisioning., This work is part of a project that has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 754337 (EuroEXA); it has been supported by the Spanish Ministry of Science and Innovation (project TIN2015-65316-P and Ramon y Cajal fellowship RYC2018-025628-I), Generalitat de Catalunya (contracts 2014-SGR-1051 and 2014-SGR-1272), and the Severo Ochoa Programme (SEV-2015-0493)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Large-memory nodes for energy efficient high-performance computing

  • Živanovič, Darko
  • Radulović, Milan
  • Llort, German
  • Zaragoza, David
  • Strassburg, Janko
  • Carpenter, Paul Matthew
  • Radojković, Petar
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
Energy consumption is by far the most important contributor to HPC cluster operational costs, and it accounts for a significant share of the total cost of ownership. Advanced energy-saving techniques in HPC components have received significant research and development effort, but a simple measure that can dramatically reduce energy consumption is often overlooked. We show that, in capacity computing, where many small to medium-sized jobs have to be solved at the lowest cost, a practical energy-saving approach is to scale-in the application on large-memory nodes. We evaluate scaling-in; i.e. decreasing the number of application processes and compute nodes (servers) to solve a fixed-sized problem, using a set of HPC applications running in a production system. Using standard-memory nodes, we obtain average energy savings of 36%, already a huge figure. We show that the main source of these energy savings is a decrease in the node-hours (node_hours = #nodes x exe_time), which is a consequence of the more efficient use of hardware resources. Scaling-in is limited by the per-node memory capacity. We therefore consider using large-memory nodes to enable a greater degree of scaling-in. We show that the additional energy savings, of up to 52%, mean that in many cases the investment in upgrading the hardware would be recovered in a typical system lifetime of less than five years., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Optimization of condensed matter physics application with OpenMP tasking model

  • Criado, Joel
  • Garcia Gasulla, Marta|||0000-0003-3682-9905
  • Labarta Mancho, Jesús José|||0000-0002-7489-4727
  • Chatterjee, Arghya
  • Hernández, Óscar
  • Sirvent, Raül
  • Álvarez, Gonzalo
The Density Matrix Renormalization Group (DMRG++) is a condensed matter physics application used to study superconductivity properties of materials. It’s main computations consist of calculating hamiltonian matrix which requires sparse matrix-vector multiplications. This paper presents task-based parallelization and optimization strategies of the Hamiltonian algorithm. The algorithm is implemented as a mini-application in C++ and parallelized with OpenMP. The optimization leverages tasking features, such as dependencies or priorities included in the OpenMP standard 4.5. The code refactoring targets performance as much as programmability. The optimized version achieves a speedup of 8.0 × with 8 threads and 20.5 × with 40 threads on a Power9 computing node while reducing the memory consumption to 90 MB with respect to the original code, by adding less than ten OpenMP directives., This work is partially supported by the Spanish Government through Programa Severo Ochoa (SEV2015-0493), by the Spanish Ministry of Science and Technology (project TIN2015-65316-P), by the Generalitat de Catalunya (contract 2017-SGR-1414) and by the BSC-IBM Deep Learning Research Agreement, under JSA “Application porting, analysis and optimization for POWER and POWER AI”. This work was partially supported by the Scientific Discovery through Advanced Computing (SciDAC) program funded by U.S. Department of Energy, Office of Science, Advanced Scientific Computing Research and Basic Energy Sciences, Division of Materials Sciences and Engineering. This research used resources of the Oak Ridge Leadership Computing Facility at the Oak Ridge National Laboratory, which is supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC05-00OR22725., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Enhancing resource management through prediction-based policies

  • Navarro Muñoz, Antoni
  • Lorenzon, Arthur F.
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
  • Beltran Querol, Vicenç
Task-based programming models are emerging as a promising alternative to make the most of multi-/many-core systems. These programming models rely on runtime systems, and their goal is to improve application performance by properly scheduling application tasks to cores. Additionally, these runtime systems offer policies to cope with application phases that lack in parallelism to fill all cores. However, these policies are usually static and favor either performance or energy efficiency. In this paper, we have extended a task-based runtime system with a lightweight monitoring and prediction infrastructure that dynamically predicts the optimal number of cores required for each application phase, thus improving both performance and energy efficiency. Through the execution of several benchmarks in multi-/many-core systems, we show that our prediction-based policies have competitive performance while improving energy efficiency when compared to state of the art policies., This project is supported by the European Union’s Horizon 2021 research and innovation programme under the grant agreement No 754304 (DEEP-EST), the Ministry of Economy of Spain through the Severo Ochoa Center of Excellence Program (SEV-2015-0493), by the Spanish Ministry of Science and Innovation (contract TIN2015-65316-P) and by the Generalitat de Catalunya (2017-SGR1481). This work was also supported by Project HPC-EUROPA3 (INFRAIA2016-1-730897), with the support of the EC Research Innovation Action under the H2020 Programme., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

A Linux kernel scheduler extension for multi-core systems

  • Roca Nonell, Aleix|||0000-0002-6715-3605
  • Rodríguez, Samuel|||0000-0002-2406-6499
  • Segura, Albert
  • Marquet, Kevin
  • Beltran, Vicenç
©2020 IEEE. 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 Linux kernel is mostly designed for multi-programed environments, but high-performance applications have other requirements. Such applications are run standalone, and usually rely on runtime systems to distribute the application's workload on worker threads, one per core. However, due to current OSes limitations, it is not feasible to track whether workers are actually running or blocked due to, for instance, a requested resource. For I/O intensive applications, this leads to a significant performance degradation given that the core of a blocked thread becomes idle until it is able to run again. In this paper, we present the proof-of-concept of a Linux kernel extension denoted User-Monitored Threads (UMT) which tackles this problem. Our extension allows a user-space process to be notified of when the selected threads become blocked or unblocked, making it possible for a runtime to schedule additional work on the idle core. We implemented the extension on the Linux Kernel 5.1 and adapted the Nanos6 runtime of the OmpSs-2 programming model to take advantage of it. The whole prototype was tested on two applications which, on the tested hardware and the appropriate conditions, reported speedups of almost 2x., This project is supported by the European Union’s Horizon 2021 research and innovation programme under the grant agreement No 754304 (DEEP-EST), the Ministry of Economy of Spain through the Severo Ochoa Center of Excellence Program (SEV-2015-0493), by the Spanish Ministry of Science and Innovation (contract TIN2015-65316-P) and by the Generalitat de Catalunya (2017-SGR-1481)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Node architecture implications for in-memory data analytics on scale-in clusters

  • Awan, Ashan Javed
  • Vlassov, Vladimir
  • Brorsson, Mats
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
While cluster computing frameworks are continuously evolving to provide real-time data analysis capabilities, Apache Spark has managed to be at the forefront of big data analytics. Recent studies propose scale-in clusters with in-storage processing devices to process big data analytics with Spark However the proposal is based solely on the memory bandwidth characterization of in-memory data analytics and also does not shed light on the specification of host CPU and memory. Through empirical evaluation of in-memory data analytics with Apache Spark on an Ivy Bridge dual socket server, we have found that (i) simultaneous multithreading is effective up to 6 cores (ii) data locality on NUMA nodes can improve the performance by 10% on average, (iii) disabling next-line L1-D prefetchers can reduce the execution time by up to 14%, (iv) DDR3 operating at 1333 MT/s is suffcient and (v) multiple small executors can provide up to 36% speedup over single large executor., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Random forest as a tumour genetic marker extractor

  • Pérez Arnal, Raquel Leandra|||0000-0002-0041-8146
  • Garcia Gasulla, Dario|||0000-0001-6732-5641
  • Torrents Rodas, David
  • Pares, Ferran
  • Cortés García, Claudio Ulises|||0000-0003-0192-3096
  • Labarta Mancho, Jesús José|||0000-0002-7489-4727
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
Finding tumour genetic markers is essential to biomedicine due to their relevance for cancer detection and therapy development. In this paper, we explore a recently released dataset of chromosome rearrangements in 2,586 cancer patients, where different sorts of alterations have been detected. Using a Random Forest classifier, we evaluate the relevance of several features (some directly available in the original data, some engineered by us) related to chromosome rearrangements. This evaluation results in a set of potential tumour genetic markers, some of which are validated in the bibliography, while others are potentially novel., This work is partially supported by the Joint Study Agreement no. W156463 under the IBM/BSC Deep Learning Center agreement, by the Spanish Government through Programa Severo Ochoa (SEV-2015-0493), by the Spanish Ministry of Science and Technology through TIN2015-65316-P project, and by the Generalitat de Catalunya (contracts 2017-SGR-1414)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

The OTree: multidimensional indexing with efficient data sampling for HPC

  • Cugnasco, Cesare
  • Calmet, Hadrien|||0000-0001-5443-761X
  • Santamaria Mateu, Pol
  • Sirvent Pardell, Raül|||0000-0003-0606-2512
  • Eguzkitza, Ane Beatriz
  • Houzeaux, Guillaume|||0000-0002-2592-1426
  • Becerra Fontal, Yolanda|||0000-0003-2357-7796
  • Torres Viñals, Jordi|||0000-0003-1963-7418
  • Labarta Mancho, Jesús José|||0000-0002-7489-4727
Spatial big data is considered an essential trend in future scientific and business applications. Indeed, research instruments, medical devices, and social networks generate hundreds of petabytes of spatial data per year. However, many authors have pointed out that the lack of specialized frameworks for multidimensional Big Data is limiting possible applications and precluding many scientific breakthroughs. Paramount in achieving High-Performance Data Analytics is to optimize and reduce the I/O operations required to analyze large data sets. To do so, we need to organize and index the data according to its multidimensional attributes. At the same time, to enable fast and interactive exploratory analysis, it is vital to generate approximate representations of large datasets efficiently. In this paper, we propose the Outlook Tree (or OTree), a novel Multidimensional Indexing with efficient data Sampling (MIS) algorithm. The OTree enables exploratory analysis of large multidimensional datasets with arbitrary precision, a vital missing feature in current distributed data management solutions. Our algorithm reduces the indexing overhead and achieves high performance even for write-intensive HPC applications. Indeed, we use the OTree to store the scientific results of a study on the efficiency of drug inhalers. Then we compare the OTree implementation on Apache Cassandra, named Qbeast, with PostgreSQL and plain storage. Lastly, we demonstrate that our proposal delivers better performance and scalability., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Improving the interoperability between MPI and task-based programming models

  • Sala Penadés, Kevin|||0000-0001-8233-1185
  • Bellón, Jorge
  • Farré, Pau|||0000-0002-5705-8620
  • Teruel, Xavier
  • Pérez, Josep M.
  • Peña, Antonio J.
  • Holmes, Daniel
  • Beltran Querol, Vicenç
  • Labarta Mancho, Jesús José|||0000-0002-7489-4727
In this paper we propose an API to pause and resume task execution depending on external events. We leverage this generic API to improve the interoperability between MPI synchronous communication primitives and tasks. When an MPI operation blocks, the task running is paused so that the runtime system can schedule a new task on the core that became idle. Once the MPI operation is completed, the paused task is put again on the runtime system's ready queue. We expose our proposal through a new MPI threading level which we implement through two approaches. The first approach is an MPI wrapper library that works with any MPI implementation by intercepting MPI synchronous calls, implementing them on top of their asynchronous counterparts. In this case, the task-based runtime system is also extended to periodically check for pending MPI operations and resume the corresponding tasks once MPI operations complete. The second approach consists in directly modifying the MPICH runtime system, a well-known implementation of MPI, to directly call the pause/resume API when a synchronous MPI operation blocks and completes, respectively. Our experiments reveal that this proposal not only simplifies the development of hybrid MPI+OpenMP applications that naturally overlap computation and communication phases; it also improves application performance and scalability by removing artificial dependencies across communication tasks., This work has been developed with the support of the European Union Horizon 2020 Programme through both the INTERTWinE project (agreement No. 671602) and the Marie Sk lodowska-Curie grant (agreement No. 749516); the Spanish Government through the Severo Ochoa Program (SEV-2015-0493); the Spanish Ministry of Science and Innovation (TIN2015-65316-P) and the Generalitat de Catalunya (2017-SGR-1414)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

DRAM errors in the field: a statistical approach

  • Živanovič, Darko
  • Esmaili Dokht, Pouya|||0000-0001-8799-5773
  • Moré, Sergi
  • Bartolomé, Javier
  • Carpenter, Paul Matthew
  • Radojković, Petar
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
This paper summarizes our two-year study of corrected and uncor-rected errors on the MareNostrum 3 supercomputer, covering 2000 billion MB-hours of DRAM in the field. The study analyzes 4.5 million corrected and 71 uncorrected DRAM errors and it compares the reliability of DIMMs from all three major memory manufacturers, built in three different technologies. Our work has two sets of contributions. First, we illustrate the complexity of in-field DRAM error analysis and demonstrate the limitations of various widely-used methods and metrics. For example, we show that average error rates, errors per MB-hour and mean time between failures can provide volatile and unreliable results even after long periods of error logging, leading to incorrect conclusions about DRAM reliability. Second, we present formal statistical methods that overcome many of the limitations of the current approaches. The methods that we present are simple to understand and implement, reliable and widely accepted in the statistical community. Overall, our study alerts the community about the need to, firstly, question the current practice in quantifying DRAM reliability and, secondly, to select a proper analysis approach for future studies. Our strong recommendations are to focus on metrics with a practical value that could be easily related to system reliability, and to select methods that provide stable results, ideally supported with statistical significance., This work was supported by the Collaboration Agreement between Samsung Electronics Co., Ltd. and BSC, Spanish Government through Severo Ochoa programme (SEV-2015-0493), by the Spanish Ministry of Science and Technology through TIN2015-65316-P project and by the Generalitat de Catalunya (contracts 2014-SGR1051 and 2014-SGR-1272). This work has also received funding from the European Union’s Horizon 2020 research and innovation programme under EuroEXA project (grant agreement No 754337). Darko Zivanovic holds the Severo Ochoa grant (SVP-2014-068501) of the Ministry of Economy and Competitiveness of Spain.
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Reducing cache coherence traffic with hierarchical directory cache and NUMA-aware runtime scheduling

  • Caheny, Paul
  • Casas, Marc|||0000-0003-4564-2093
  • Moretó Planas, Miquel|||0000-0002-9848-8758
  • Gloaguen, Hervé
  • Saintes, Maxime
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
  • Labarta Mancho, Jesús José|||0000-0002-7489-4727
  • Valero Cortés, Mateo|||0000-0003-2917-2482
Cache Coherent NUMA (ccNUMA) architectures are a widespread paradigm due to the benefits they provide for scaling core count and memory capacity. Also, the flat memory address space they offer considerably improves programmability. However, ccNUMA architectures require sophisticated and expensive cache coherence protocols to enforce correctness during parallel executions, which trigger a significant amount of on- and off-chip traffic in the system. This paper analyses how coherence traffic may be best constrained in a large, real ccNUMA platform through the use of a joint hardware/software approach. For several benchmarks, we study coherence traffic in detail under the influence of an added hierarchical cache layer in the directory protocol combined with runtime managed NUMA-aware scheduling and data allocation techniques to make most efficient use of the added hardware. The effectiveness of this joint approach is demonstrated by speedups of 1.23x to 2.54x and coherence traffic reductions between 44% and 77% in comparison to NUMA-oblivious scheduling and data allocation. Furthermore, we show that the NUMA-aware techniques we employ at the runtime level are crucial to ensure the added hierarchical layer in the directory coherence protocol does not introduce significant coherence traffic to the system., This work has been supported by the Spanish Government (Severo Ochoa grants SEV2015-0493), by the Spanish Ministry of Science and Innovation (contracts TIN2015-65316-P), by the Generalitat de Catalunya (contracts 2014-SGR-1051 and 2014-SGR-1272), by the RoMoL ERC Advanced Grant (GA 321253) and the European HiPEAC Network of Excellence. The Mont-Blanc project receives funding from the EU's Seventh Framework Programme (FP7/2007-2013) under grant agreement no 610402 and from the EU's H2020 Framework Programme (H2020/2014-2020) under grant agreement no 671697. M. Moretó has been partially supported by the Ministry of Economy and Competitiveness under Juan de la Cierva postdoctoral fellowship number JCI-2012- 15047. M. Casas is supported by the Secretary for Universities and Research of the Ministry of Economy and Knowledge of the Government of Catalonia and the Cofund programme of the Marie Curie Actions of the 7th R&D Framework Programme of the European Union (Contract 2013 BP B 00243)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

A toolchain to verify the parallelization of OmpSs-2 applications

  • Economo, Simone
  • Royuela Alcázar, Sara|||0000-0002-7644-0868
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
  • Beltran Querol, Vicenç
Programming models for task-based parallelization based on compile-time directives are very effective at uncovering the parallelism available in HPC applications. Despite that, the process of correctly annotating complex applications is error-prone and may hinder the general adoption of these models. In this paper, we target the OmpSs-2 programming model and present a novel toolchain able to detect parallelization errors coming from non-compliant OmpSs-2 applications. Our toolchain verifies the compliance with the OmpSs-2 programming model using local task analysis to deal with each task separately, and structural induction to extend the analysis to the whole program. To improve the effectiveness of our tools, we also introduce some ad-hoc verification annotations, which can be used manually or automatically to disable the analysis of specific code regions. Experiments run on a sample of representative kernels and applications show that our toolchain can be successfully used to verify the parallelization of complex real-world applications., This project is supported by the European Union’s Horizon 2021 research and innovation programme under grant agreement No 754304 (DEEP-EST), by the European Union’s Horizon 2020 research and innovation programme under grant agreement No 871669 (AMPERE) and the Project HPCEUROPA3 (INFRAIA-2016-1-730897), by the Ministry of Economy of Spain through the Severo Ochoa Center of Excellence Program (SEV-2015-0493), by the Spanish Ministry of Science and Innovation (contract TIN2015-65316-P), and by the Generalitat de Catalunya (2017-SGR-1481)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Evaluating worksharing tasks on distributed environments

  • Maroñas Bravo, Marcos
  • Teruel García, Xavier
  • Bull, J. Mark
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
  • Beltran Querol, Vicenç
©2020 IEEE. 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., Hybrid programming is a promising approach to exploit clusters of multicore systems. Our focus is on the combination of MPI and tasking. This hybrid approach combines the low-latency and high throughput of MPI with the flexibility of tasking models and their inherent ability to handle load imbalance. However, combining tasking with standard MPI implementations can be a challenge. The Task-Aware MPI library (TAMPI) eases the development of applications combining tasking with MPI. TAMPI enables developers to overlap computation and communication phases by relying on the tasking data-flow execution model. Using this approach, the original computation that was distributed in many different MPI ranks is grouped together in fewer MPI ranks, and split into several tasks per rank. Nevertheless, programmers must be careful with task granularity. Too fine-grained tasks introduce too much overhead, while too coarse-grained tasks lead to lack of parallelism. An adequate granularity may not always exist, especially in distributed environments where the same amount of work is distributed among many more cores. Worksharing tasks are a special kind of tasks, recently proposed, that internally leverage worksharing techniques. By doing so, a single worksharing task may run in several cores concurrently. Nonetheless, the task management costs remain the same than a regular task. In this work, we study the combination of worksharing tasks and TAMPI on distributed environments using two well known mini-apps: HPCCG and LULESH. Our results show significant improvements using worksharing tasks compared to regular tasks, and to other state-of-the-art alternatives such as OpenMP worksharing., This project is supported by the European Union’s Horizon 2020 Research and Innovation programme under grant agreement No.s 754304 (DEEP-EST) and 823767 (PRACE), the Ministry of Economy of Spain through the Severo Ochoa Center of Excellence Program (SEV-2015-0493), by the Spanish Ministry of Science and Innovation (contract PID2019-107255GB) and by the Generalitat de Catalunya (2017-SGR1481). The work has been performed under the Project HPCEUROPA3 (INFRAIA-2016-1-730897), with the support of the EC Research Innovation Action under the H2020 Programme; in particular, the author gratefully acknowledges the support of Dr Mark Bull (EPCC) and the computer resources and technical support provided by EPCC., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Feature discriminativity estimation in CNNs for transfer learning

  • Giménez Ábalos, Víctor
  • Vilalta Arias, Armand
  • Garcia Gasulla, Dario|||0000-0001-6732-5641
  • Labarta Mancho, Jesús José|||0000-0002-7489-4727
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
The purpose of feature extraction on convolutional neural networks is to reuse deep representations learnt for a pre-trained model to solve a new, potentially unrelated problem. However, raw feature extraction from all layers is unfeasible given the massive size of these networks. Recently, a supervised method using complexity reduction was proposed, resulting in significant improvements in performance for transfer learning tasks. This approach first computes the discriminative power of features, and then discretises them using thresholds computed for the task. In this paper, we analyse the behaviour of these thresholds, with the purpose of finding a methodology for their estimation. After a comprehensive study, we find a very strong correlation between problem size and threshold value, with coefficient of determination above 90%. These results allow us to propose a unified model for threshold estimation, with potential application to transfer learning tasks., This work is partially supported by BSC-IBM Deep Learning Center agreement, the Spanish Government through Programa Severo Ochoa (SEV-2015-0493), the Spanish Ministry of Science and Technology through TIN2015-65316-P project and the Generalitat de Catalunya (contract 2017-SGR-1414)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Echo state hoeffding tree learning

  • Marrón Vida, Diego
  • Read, Jesse
  • Bifet Figuerol, Albert Carles
  • Abdessalem, Talel
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
  • Herrero Zaragoza, José Ramón|||0000-0002-4060-367X
Nowadays, real-time classi cation of Big Data streams is becoming essential in a variety of application domains. While decision trees are powerful and easy{to{deploy approaches for accurate and fast learning from data streams, they are unable to capture the strong temporal dependences typically present in the input data. Recurrent Neural Networks are an alternative solution that include an internal memory to capture these temporal dependences; however their training is computationally very expensive, with slow convergence and not easy{to{deploy (large number of hyper-parameters). Reservoir Computing was proposed to reduce the computation requirements of the training phase but still include a feed-forward layer which requires a large number of parameters to tune. In this work we propose a novel architecture for real-time classification based on the combination of a Reservoir and a decision tree. This combination makes classification fast, reduces the number of hyper-parameters and keeps the good temporal properties of recurrent neural networks. The capabilities of the proposed architecture to learn some typical string-based functions with strong temporal dependences are evaluated in the paper. The paper shows how the new architecture is able to incrementally learn these functions in real-time with fast adaptation to unknown sequences and analyzes the influence of the reduced number of hyper-parameters in the behaviour of the proposed solution., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Contention-aware application performance prediction for disaggregated memory systems

  • Vieira Zacarias, Felippe
  • Nishtala, Rajiv
  • Carpenter, Paul Matthew
Disaggregated memory has recently been proposed as a way to allow flexible and fine-grained allocation of memory capacity to compute jobs. This paper makes an important step towards effective resource allocation on disaggregated memory systems. Specifically, we propose a generic approach to predict the performance degradation due to sharing of disaggregated memory. In contrast to prior work, cache capacity is not shared among multiple applications, which removes a major contributor to application performance. For this reason, our analysis is driven by the demand for memory bandwidth, which has been shown to have an important effect on application performance. We show that profiling the application slowdown often involves significant experimental error and noise, and to this end, we improve the accuracy by linear smoothing of the sensitivity curves. We also show that contention is sensitive to the ratio between read and write memory accesses, and we address this sensitivity by building a family of sensitivity curves according to the read/write ratios. Our results show that the methodology predicts the slowdown in application performance subject to memory contention with an average error of 1.19% and max error of 14.6%. Compared with state-of-the-art, the relative improvements are almost 24% on average and 33% for the worst case., This work is part of a project that has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 754337 (EuroEXA); it has been supported by the Spanish Ministry of Science and Innovation (project TIN2015-65316-P and Ramon y Cajal fellowship RYC2018-025628-I), Generalitat de Catalunya (contracts 2014-SGR-1051 and 2014-SGR1272), and the Severo Ochoa Programme (SEV-2015-0493)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

A vulnerability factor for ECC-protected memory

  • Jaulmes, Luc|||0000-0002-5815-2504
  • Moretó Planas, Miquel|||0000-0002-9848-8758
  • Valero Cortés, Mateo|||0000-0003-2917-2482
  • Casas, Marc|||0000-0003-4564-2093
Fault injection studies and vulnerability analyses have been used to estimate the reliability of data structures in memory. We survey these metrics and look at their adequacy to describe the data stored in ECC-protected memory. We also introduce FEA, a new metric improving on the memory derating factor by ignoring a class of false errors. We measure all metrics using simulations and compare them to the outcomes of injecting errors in real runs. This in-depth study reveals that FEA provides more accurate results than any state-of-the-art vulnerability metric. Furthermore, FEA gives an upper bound on the failure probability due to an error in memory, making this metric a tool of choice to quantify memory vulnerability. Finally, we show that ignoring these false errors reduces the failure rate on average by 12.75% and up to over 45%., This work has been supported by the RoMoL ERC Advanced Grant (GA 321253), by the European HiPEAC Network of Excellence, by the Spanish Ministry of Economy and Competitiveness (contract TIN2015-65316- P), by the Generalitat de Catalunya (contracts 2017-SGR-1414 and 2017- SGR-1328), by the Spanish Government (Severo Ochoa grant SEV-2015- 0493) and by the European Union’s Horizon 2020 research and innovation programme (grant agreements 671697 and 779877). L. Jaulmes has been partially supported by the Spanish Ministry of Education, Culture and Sports under grant FPU2013/06982. M. Moreto and M. Casas have been partially supported by the Spanish Ministry of Economy, Industry and Competitiveness under Ramon y Cajal fellowships RYC-2016-21104 and RYC-2017-23269., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Constant-time approximate sliding window framework with error control

  • Villalba Navarro, Álvaro
  • Carrera Pérez, David|||0000-0003-4898-3424
Stream Processing is a crucial element for the Edge Computing paradigm, in which large amount of devices generate data at the edge of the network. This data needs to be aggregated and processed on-the-move across different layers before reaching the Cloud. Therefore, defining Stream Processing services that adapt to different levels of resource availability is of paramount importance. In this context, Stream Processing frameworks need to combine efficient algorithms with low computational complexity to manage sliding windows, with the ability to adjust resource demands for different deployment scenarios, from very low capacity edge devices to virtually unlimited Cloud platforms. The Approximate Computing paradigm provides improved performance and adaptive resource demands in data analytics, at the price of introducing some level of inaccuracy that can be calculated. In this paper we present the Approximate and Amortized Monoid Tree Aggregator (A 2 MTA). It is, to our knowledge, the first general purpose sliding window programable framework that combines constant-time aggregations with error bounded approximate computing techniques. It is very suitable for adverse stream processing environments, such as resource scarce multi-tenant edge computing. The framework can compute aggregations over multiple data dimensions, setting error bounds on any of them, and has been designed to support decoupling computation and data storage through the use of distributed Key-Value Stores to keep window elements and partial aggregations., This project is partially supported by the European Research Council (ERC), Spain under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 639595). It is also partially supported by the Ministry of Economy of Spain under contract TIN2015-65316-P and Generalitat de Catalunya, Spain under contract 2014SGR1051, by the ICREA Academia program, and by the BSC-CNS Severo Ochoa program (SEV-2015-0493)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Benchmarking of state-of-the-art HPC clusters with a production CFD code

  • Banchelli Gracia, Fabio
  • Garcia Gasulla, Marta|||0000-0003-3682-9905
  • Houzeaux, Guillaume|||0000-0002-2592-1426
  • Mantovani, Filippo|||0000-0003-3559-4825
Computing technologies populating high-performance computing (HPC) clusters are getting more and more diverse, offering a wide range of architectural features. As a consequence, efficient programming of such platforms becomes a complex task. In this paper we provide a micro-benchmarking of three HPC clusters based on different CPU architectures, predominant in the Top500 ranking: x86, Armv8 and IBM Power9. On these platforms we study a production fluid-dynamics application leveraging different compiler technologies and micro-architectural features. We finally provide a scalability study on state-of-the-art HPC clusters. The two most relevant conclusions of our study are: i) Compiler development is critical for squeezing performance out of most recent technologies; ii) Micro-architectural features such as Single Instruction Multiple Data (SIMD) units and Simultaneous Multi-Threading (SMT) can impact the overall performance. However, a closer look shows that while SIMD is improving the performance of compute bound regions, SMT does not show a clear benefit on HPC workloads., The authors thank the support team of Dibona operating at ATOS/Bull. This work is partially supported by the Spanish Government through Programa Severo Ochoa (SEV-2015-0493), the Spanish Ministry of Science and Technology (project TIN2015-65316-P), the Generalitat de Catalunya (2017-SGR-1414), the European project Mont-Blanc 3 (GA n. 671697), and the European centre of excellence on HPC POP2 (GA n. 824080)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Workflow environments for advanced cyberinfrastructure platforms

  • Badia Sala, Rosa Maria|||0000-0003-2941-5499
  • Ejarque Artigas, Jorge
  • Lordan Gomis, Francesc|||0000-0002-9845-8890
  • Lezzi, Daniele|||0000-0001-5081-7244
  • Conejero Bañón, Javier
  • Álvarez Cid-Fuentes, Javier
  • Becerra Fontal, Yolanda|||0000-0003-2357-7796
  • Queralt Calafat, Anna|||0000-0003-2782-2955
Progress in science is deeply bound to the effective use of high-performance computing infrastructures and to the efficient extraction of knowledge from vast amounts of data. Such data comes from different sources that follow a cycle composed of pre-processing steps for data curation and preparation for subsequent computing steps, and later analysis and analytics steps applied to the results. However, scientific workflows are currently fragmented in multiple components, with different processes for computing and data management, and with gaps in the viewpoints of the user profiles involved. Our vision is that future workflow environments and tools for the development of scientific workflows should follow a holistic approach, where both data and computing are integrated in a single flow built on simple, high-level interfaces. The topics of research that we propose involve novel ways to express the workflows that integrate the different data and compute processes, dynamic runtimes to support the execution of the workflows in complex and heterogeneous computing infrastructures in an efficient way, both in terms of performance and energy. These infrastructures include highly distributed resources, from sensors and instruments, and devices in the edge, to High-Performance Computing and Cloud computing resources. This paper presents our vision to develop these workflow environments and also the steps we are currently following to achieve it., This work has been supported by the Spanish Government (SEV2015-0493), by the Spanish Ministry of Science and Innovation (contract TIN2015-65316-P), by Generalitat de Catalunya (contract 2014-SGR-1051). Javier Conejero postdoctoral contract is co-financed by the Ministry of Economy and Competitiveness under Juan de la Cierva Formacion´ postdoctoral fellowship number FJCI-2015-24651. This work is supported by the H2020 mF2C project (730929) and the CLASS project (780622). The participation of Rosa M Badia in the BDEC2 meetings is supported by the EXDCI project (800957). The dislib library developments are partially funded under the project agreement between BSC and FUJITSU., Peer Reviewed
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

The OTree: multidimensional indexing with efficient data sampling for HPC

  • Cugnasco, Cesare
  • Calmet, Hadrien
  • Santamaria Mateu, Pol
  • Sirvent Pardell, Raül
  • Eguzkitza, Ane Beatriz
  • Houzeaux, Guillaume
  • Becerra Fontal, Yolanda
  • Torres Viñals, Jordi
  • Labarta Mancho, Jesús José
Spatial big data is considered an essential trend in future scientific and business applications. Indeed, research instruments, medical devices, and social networks generate hundreds of petabytes of spatial data per year. However, many authors have pointed out that the lack of specialized frameworks for multidimensional Big Data is limiting possible applications and precluding many scientific breakthroughs. Paramount in achieving High-Performance Data Analytics is to optimize and reduce the I/O operations required to analyze large data sets. To do so, we need to organize and index the data according to its multidimensional attributes. At the same time, to enable fast and interactive exploratory analysis, it is vital to generate approximate representations of large datasets efficiently. In this paper, we propose the Outlook Tree (or OTree), a novel Multidimensional Indexing with efficient data Sampling (MIS) algorithm. The OTree enables exploratory analysis of large multidimensional datasets with arbitrary precision, a vital missing feature in current distributed data management solutions. Our algorithm reduces the indexing overhead and achieves high performance even for write-intensive HPC applications. Indeed, we use the OTree to store the scientific results of a study on the efficiency of drug inhalers. Then we compare the OTree implementation on Apache Cassandra, named Qbeast, with PostgreSQL and plain storage. Lastly, we demonstrate that our proposal delivers better performance and scalability., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

A toolchain to verify the parallelization of OmpSs-2 applications

  • Economo, Simone
  • Royuela Alcázar, Sara
  • Ayguadé Parra, Eduard
  • Beltran Querol, Vicenç
Programming models for task-based parallelization based on compile-time directives are very effective at uncovering the parallelism available in HPC applications. Despite that, the process of correctly annotating complex applications is error-prone and may hinder the general adoption of these models. In this paper, we target the OmpSs-2 programming model and present a novel toolchain able to detect parallelization errors coming from non-compliant OmpSs-2 applications. Our toolchain verifies the compliance with the OmpSs-2 programming model using local task analysis to deal with each task separately, and structural induction to extend the analysis to the whole program. To improve the effectiveness of our tools, we also introduce some ad-hoc verification annotations, which can be used manually or automatically to disable the analysis of specific code regions. Experiments run on a sample of representative kernels and applications show that our toolchain can be successfully used to verify the parallelization of complex real-world applications., This project is supported by the European Union’s Horizon 2021 research and innovation programme under grant agreement No 754304 (DEEP-EST), by the European Union’s Horizon 2020 research and innovation programme under grant agreement No 871669 (AMPERE) and the Project HPCEUROPA3 (INFRAIA-2016-1-730897), by the Ministry of Economy of Spain through the Severo Ochoa Center of Excellence Program (SEV-2015-0493), by the Spanish Ministry of Science and Innovation (contract TIN2015-65316-P), and by the Generalitat de Catalunya (2017-SGR-1481)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Accelerating K-mer Frequency Counting with GPU and Non-Volatile Memory

  • Cadenelli, Nicola
  • Polo Cantero, José
  • Carrera Pérez, David
The emergence of Next Generation Sequencing (NGS) platforms has increased the throughput of genomic sequencing and in turn the amount of data that needs to be processed, requiring highly efficient computation for its analysis. In this context, modern architectures including accelerators and non-volatile memory are essential to enable the mass exploitation of these bioinformatics workloads. This paper presents a redesign of the main component of a state-of-the-art reference-free method for variant calling, SMUFIN, which has been adapted to make the most of GPUs and NVM devices. SMUFIN relies on counting the frequency of k-mers (substrings of length k) in DNA sequences, which also constitutes a well-known problem for many bioinformatics workloads, such as genome assembly. We propose techniques to improve the efficiency of k-mer counting and to scale-up workloads like SMUFIN that used to require 16 nodes of Marenostrum 3 to a single machine with a GPU and NVM drives. Results show that although the single machine is not able to improve the time to solution of 16 nodes, its CPU time is 7.5x shorter than the aggregate CPU time of the 16 nodes, with a reduction in energy consumption of 5.5x. © 2017 IEEE., Peer Reviewed, Preprint
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

MUSA: a multi-level simulation approach for next-generation HPC machines

  • Grass, Thomas
  • Allande, César
  • Armejach, Adrià
  • Rico, Alejandro
  • Ayguadé Parra, Eduard
  • Labarta Mancho, Jesús José
  • Valero Cortés, Mateo
  • Casas, Marc
  • Moretó Planas, Miquel
The complexity of High Performance Computing (HPC) systems is increasing in the number of components and their heterogeneity. Interactions between software and hardware involve many different aspects which are typically not transparent to scientific programmers and system architects. Therefore, predicting the behavior of current scientific applications on future HPC infrastructures is a challenging task. In this paper we present MUSA, an end-to-end methodology that employs a multi-level simulation infrastructure. By combining different levels of abstraction, MUSA is able to model the communication network, microarchitectural details and system software interactions, providing different trade-offs in terms of simulation cost and accuracy. We compare detailed MUSA simulations with native executions of up to 2,048 cores and find relative errors that are within 10% in the common case. In addition, we use MUSA to simulate up to 16,384 cores and successfully identify scalability bottlenecks due to different factors, e.g. memory contention or load imbalance. We also compare different system configurations, showing how MUSA can help system designers to assess the usefulness of future technologies in next-generation HPC machines., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Feature discriminativity estimation in CNNs for transfer learning

  • Giménez Ábalos, Víctor
  • Vilalta Arias, Armand
  • Garcia Gasulla, Dario
  • Labarta Mancho, Jesús José
  • Ayguadé Parra, Eduard
The purpose of feature extraction on convolutional neural networks is to reuse deep representations learnt for a pre-trained model to solve a new, potentially unrelated problem. However, raw feature extraction from all layers is unfeasible given the massive size of these networks. Recently, a supervised method using complexity reduction was proposed, resulting in significant improvements in performance for transfer learning tasks. This approach first computes the discriminative power of features, and then discretises them using thresholds computed for the task. In this paper, we analyse the behaviour of these thresholds, with the purpose of finding a methodology for their estimation. After a comprehensive study, we find a very strong correlation between problem size and threshold value, with coefficient of determination above 90%. These results allow us to propose a unified model for threshold estimation, with potential application to transfer learning tasks., This work is partially supported by BSC-IBM Deep Learning Center agreement, the Spanish Government through Programa Severo Ochoa (SEV-2015-0493), the Spanish Ministry of Science and Technology through TIN2015-65316-P project and the Generalitat de Catalunya (contract 2017-SGR-1414)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

TaskPoint: sampled simulation of task-based programs

  • Grass, Thomas Dieter
  • Rico, Alejandro
  • Casas, Marc
  • Moretó Planas, Miquel
  • Ayguadé Parra, Eduard
Sampled simulation is a mature technique for reducing simulation time of single-threaded programs, but it is not directly applicable to simulation of multi-threaded architectures. Recent multi-threaded sampling techniques assume that the workload assigned to each thread does not change across multiple executions of a program. This assumption does not hold for dynamically scheduled task-based programming models. Task-based programming models allow the programmer to specify program segments as tasks which are instantiated many times and scheduled dynamically to available threads. Due to system noise and variation in scheduling decisions, two consecutive executions on the same machine typically result in different instruction streams processed by each thread. In this paper, we propose TaskPoint, a sampled simulation technique for dynamically scheduled task-based programs. We leverage task instances as sampling units and simulate only a fraction of all task instances in detail. Between detailed simulation intervals we employ a novel fast-forward mechanism for dynamically scheduled programs. We evaluate the proposed technique on a set of 19 task-based parallel benchmarks and two different architectures. Compared to detailed simulation, TaskPoint accelerates architectural simulation with 64 simulated threads by an average factor of 19.1 at an average error of 1.8% and a maximum error of 15.0%., This work has been supported by the Spanish Government (Severo Ochoa grants SEV2015-0493, SEV-2011-00067), the Spanish Ministry of Science and Innovation (contract TIN2015-65316-P), Generalitat de Catalunya (contracts 2014-SGR-1051 and 2014-SGR-1272), the RoMoL ERC Advanced Grant (GA 321253), the European HiPEAC Network of Excellence and the Mont-Blanc project (EU-FP7-610402 and EU-H2020-671697). M. Moreto has been partially supported by the Ministry of Economy and Competitiveness under Juan de la Cierva postdoctoral fellowship JCI-2012-15047. M. Casas is supported by the Ministry of Economy and Knowledge of the Government of Catalonia and the Cofund programme of the Marie Curie Actions of the EUFP7 (contract 2013BP B 00243). T.Grass has been partially supported by the AGAUR of the Generalitat de Catalunya (grant 2013FI B 0058)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Evaluating worksharing tasks on distributed environments

  • Maroñas Bravo, Marcos
  • Teruel García, Xavier
  • Bull, J. Mark
  • Ayguadé Parra, Eduard
  • Beltran Querol, Vicenç
©2020 IEEE. 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., Hybrid programming is a promising approach to exploit clusters of multicore systems. Our focus is on the combination of MPI and tasking. This hybrid approach combines the low-latency and high throughput of MPI with the flexibility of tasking models and their inherent ability to handle load imbalance. However, combining tasking with standard MPI implementations can be a challenge. The Task-Aware MPI library (TAMPI) eases the development of applications combining tasking with MPI. TAMPI enables developers to overlap computation and communication phases by relying on the tasking data-flow execution model. Using this approach, the original computation that was distributed in many different MPI ranks is grouped together in fewer MPI ranks, and split into several tasks per rank. Nevertheless, programmers must be careful with task granularity. Too fine-grained tasks introduce too much overhead, while too coarse-grained tasks lead to lack of parallelism. An adequate granularity may not always exist, especially in distributed environments where the same amount of work is distributed among many more cores. Worksharing tasks are a special kind of tasks, recently proposed, that internally leverage worksharing techniques. By doing so, a single worksharing task may run in several cores concurrently. Nonetheless, the task management costs remain the same than a regular task. In this work, we study the combination of worksharing tasks and TAMPI on distributed environments using two well known mini-apps: HPCCG and LULESH. Our results show significant improvements using worksharing tasks compared to regular tasks, and to other state-of-the-art alternatives such as OpenMP worksharing., This project is supported by the European Union’s Horizon 2020 Research and Innovation programme under grant agreement No.s 754304 (DEEP-EST) and 823767 (PRACE), the Ministry of Economy of Spain through the Severo Ochoa Center of Excellence Program (SEV-2015-0493), by the Spanish Ministry of Science and Innovation (contract PID2019-107255GB) and by the Generalitat de Catalunya (2017-SGR1481). The work has been performed under the Project HPCEUROPA3 (INFRAIA-2016-1-730897), with the support of the EC Research Innovation Action under the H2020 Programme; in particular, the author gratefully acknowledges the support of Dr Mark Bull (EPCC) and the computer resources and technical support provided by EPCC., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

A Linux kernel scheduler extension for multi-core systems

  • Roca Nonell, Aleix
  • Rodríguez, Samuel
  • Segura, Albert
  • Marquet, Kevin
  • Beltran, Vicenç
©2020 IEEE. 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 Linux kernel is mostly designed for multi-programed environments, but high-performance applications have other requirements. Such applications are run standalone, and usually rely on runtime systems to distribute the application's workload on worker threads, one per core. However, due to current OSes limitations, it is not feasible to track whether workers are actually running or blocked due to, for instance, a requested resource. For I/O intensive applications, this leads to a significant performance degradation given that the core of a blocked thread becomes idle until it is able to run again. In this paper, we present the proof-of-concept of a Linux kernel extension denoted User-Monitored Threads (UMT) which tackles this problem. Our extension allows a user-space process to be notified of when the selected threads become blocked or unblocked, making it possible for a runtime to schedule additional work on the idle core. We implemented the extension on the Linux Kernel 5.1 and adapted the Nanos6 runtime of the OmpSs-2 programming model to take advantage of it. The whole prototype was tested on two applications which, on the tested hardware and the appropriate conditions, reported speedups of almost 2x., This project is supported by the European Union’s Horizon 2021 research and innovation programme under the grant agreement No 754304 (DEEP-EST), the Ministry of Economy of Spain through the Severo Ochoa Center of Excellence Program (SEV-2015-0493), by the Spanish Ministry of Science and Innovation (contract TIN2015-65316-P) and by the Generalitat de Catalunya (2017-SGR-1481)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Constant-time approximate sliding window framework with error control

  • Villalba Navarro, Álvaro
  • Carrera Pérez, David
Stream Processing is a crucial element for the Edge Computing paradigm, in which large amount of devices generate data at the edge of the network. This data needs to be aggregated and processed on-the-move across different layers before reaching the Cloud. Therefore, defining Stream Processing services that adapt to different levels of resource availability is of paramount importance. In this context, Stream Processing frameworks need to combine efficient algorithms with low computational complexity to manage sliding windows, with the ability to adjust resource demands for different deployment scenarios, from very low capacity edge devices to virtually unlimited Cloud platforms. The Approximate Computing paradigm provides improved performance and adaptive resource demands in data analytics, at the price of introducing some level of inaccuracy that can be calculated. In this paper we present the Approximate and Amortized Monoid Tree Aggregator (A 2 MTA). It is, to our knowledge, the first general purpose sliding window programable framework that combines constant-time aggregations with error bounded approximate computing techniques. It is very suitable for adverse stream processing environments, such as resource scarce multi-tenant edge computing. The framework can compute aggregations over multiple data dimensions, setting error bounds on any of them, and has been designed to support decoupling computation and data storage through the use of distributed Key-Value Stores to keep window elements and partial aggregations., This project is partially supported by the European Research Council (ERC), Spain under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 639595). It is also partially supported by the Ministry of Economy of Spain under contract TIN2015-65316-P and Generalitat de Catalunya, Spain under contract 2014SGR1051, by the ICREA Academia program, and by the BSC-CNS Severo Ochoa program (SEV-2015-0493)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Reducing cache coherence traffic with hierarchical directory cache and NUMA-aware runtime scheduling

  • Caheny, Paul
  • Casas, Marc
  • Moretó Planas, Miquel
  • Gloaguen, Hervé
  • Saintes, Maxime
  • Ayguadé Parra, Eduard
  • Labarta Mancho, Jesús José
  • Valero Cortés, Mateo
Cache Coherent NUMA (ccNUMA) architectures are a widespread paradigm due to the benefits they provide for scaling core count and memory capacity. Also, the flat memory address space they offer considerably improves programmability. However, ccNUMA architectures require sophisticated and expensive cache coherence protocols to enforce correctness during parallel executions, which trigger a significant amount of on- and off-chip traffic in the system. This paper analyses how coherence traffic may be best constrained in a large, real ccNUMA platform through the use of a joint hardware/software approach. For several benchmarks, we study coherence traffic in detail under the influence of an added hierarchical cache layer in the directory protocol combined with runtime managed NUMA-aware scheduling and data allocation techniques to make most efficient use of the added hardware. The effectiveness of this joint approach is demonstrated by speedups of 1.23x to 2.54x and coherence traffic reductions between 44% and 77% in comparison to NUMA-oblivious scheduling and data allocation. Furthermore, we show that the NUMA-aware techniques we employ at the runtime level are crucial to ensure the added hierarchical layer in the directory coherence protocol does not introduce significant coherence traffic to the system., This work has been supported by the Spanish Government (Severo Ochoa grants SEV2015-0493), by the Spanish Ministry of Science and Innovation (contracts TIN2015-65316-P), by the Generalitat de Catalunya (contracts 2014-SGR-1051 and 2014-SGR-1272), by the RoMoL ERC Advanced Grant (GA 321253) and the European HiPEAC Network of Excellence. The Mont-Blanc project receives funding from the EU's Seventh Framework Programme (FP7/2007-2013) under grant agreement no 610402 and from the EU's H2020 Framework Programme (H2020/2014-2020) under grant agreement no 671697. M. Moretó has been partially supported by the Ministry of Economy and Competitiveness under Juan de la Cierva postdoctoral fellowship number JCI-2012- 15047. M. Casas is supported by the Secretary for Universities and Research of the Ministry of Economy and Knowledge of the Government of Catalonia and the Cofund programme of the Marie Curie Actions of the 7th R&D Framework Programme of the European Union (Contract 2013 BP B 00243)., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Improving the interoperability between MPI and task-based programming models

  • Sala Penadés, Kevin
  • Bellón, Jorge
  • Farré, Pau
  • Teruel, Xavier
  • Pérez, Josep M.
  • Peña, Antonio J.
  • Holmes, Daniel
  • Beltran Querol, Vicenç
  • Labarta Mancho, Jesús José
In this paper we propose an API to pause and resume task execution depending on external events. We leverage this generic API to improve the interoperability between MPI synchronous communication primitives and tasks. When an MPI operation blocks, the task running is paused so that the runtime system can schedule a new task on the core that became idle. Once the MPI operation is completed, the paused task is put again on the runtime system's ready queue. We expose our proposal through a new MPI threading level which we implement through two approaches. The first approach is an MPI wrapper library that works with any MPI implementation by intercepting MPI synchronous calls, implementing them on top of their asynchronous counterparts. In this case, the task-based runtime system is also extended to periodically check for pending MPI operations and resume the corresponding tasks once MPI operations complete. The second approach consists in directly modifying the MPICH runtime system, a well-known implementation of MPI, to directly call the pause/resume API when a synchronous MPI operation blocks and completes, respectively. Our experiments reveal that this proposal not only simplifies the development of hybrid MPI+OpenMP applications that naturally overlap computation and communication phases; it also improves application performance and scalability by removing artificial dependencies across communication tasks., This work has been developed with the support of the European Union Horizon 2020 Programme through both the INTERTWinE project (agreement No. 671602) and the Marie Sk lodowska-Curie grant (agreement No. 749516); the Spanish Government through the Severo Ochoa Program (SEV-2015-0493); the Spanish Ministry of Science and Innovation (TIN2015-65316-P) and the Generalitat de Catalunya (2017-SGR-1414)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Enhancing resource management through prediction-based policies

  • Navarro Muñoz, Antoni
  • Lorenzon, Arthur F.
  • Ayguadé Parra, Eduard
  • Beltran Querol, Vicenç
Task-based programming models are emerging as a promising alternative to make the most of multi-/many-core systems. These programming models rely on runtime systems, and their goal is to improve application performance by properly scheduling application tasks to cores. Additionally, these runtime systems offer policies to cope with application phases that lack in parallelism to fill all cores. However, these policies are usually static and favor either performance or energy efficiency. In this paper, we have extended a task-based runtime system with a lightweight monitoring and prediction infrastructure that dynamically predicts the optimal number of cores required for each application phase, thus improving both performance and energy efficiency. Through the execution of several benchmarks in multi-/many-core systems, we show that our prediction-based policies have competitive performance while improving energy efficiency when compared to state of the art policies., This project is supported by the European Union’s Horizon 2021 research and innovation programme under the grant agreement No 754304 (DEEP-EST), the Ministry of Economy of Spain through the Severo Ochoa Center of Excellence Program (SEV-2015-0493), by the Spanish Ministry of Science and Innovation (contract TIN2015-65316-P) and by the Generalitat de Catalunya (2017-SGR1481). This work was also supported by Project HPC-EUROPA3 (INFRAIA2016-1-730897), with the support of the EC Research Innovation Action under the H2020 Programme., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Node architecture implications for in-memory data analytics on scale-in clusters

  • Awan, Ashan Javed
  • Vlassov, Vladimir
  • Brorsson, Mats
  • Ayguadé Parra, Eduard
While cluster computing frameworks are continuously evolving to provide real-time data analysis capabilities, Apache Spark has managed to be at the forefront of big data analytics. Recent studies propose scale-in clusters with in-storage processing devices to process big data analytics with Spark However the proposal is based solely on the memory bandwidth characterization of in-memory data analytics and also does not shed light on the specification of host CPU and memory. Through empirical evaluation of in-memory data analytics with Apache Spark on an Ivy Bridge dual socket server, we have found that (i) simultaneous multithreading is effective up to 6 cores (ii) data locality on NUMA nodes can improve the performance by 10% on average, (iii) disabling next-line L1-D prefetchers can reduce the execution time by up to 14%, (iv) DDR3 operating at 1333 MT/s is suffcient and (v) multiple small executors can provide up to 36% speedup over single large executor., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Improving accuracy and speeding up document image classification through parallel systems

  • Ferrando Monsonís, Javier
  • Domínguez, Juan Luis
  • Torres Viñals, Jordi
  • García Fuentes, Raul
  • García Doménech, David
  • Garrido Miñambres, Daniel
  • Cortada, Jordi
  • Valero Cortés, Mateo
This paper presents a study showing the benefits of the EfficientNet models compared with heavier Convolutional Neural Networks (CNNs) in the Document Classification task, essential problem in the digitalization process of institutions. We show in the RVL-CDIP dataset that we can improve previous results with a much lighter model and present its transfer learning capabilities on a smaller in-domain dataset such as Tobacco3482. Moreover, we present an ensemble pipeline which is able to boost solely image input by combining image model predictions with the ones generated by BERT model on extracted text by OCR. We also show that the batch size can be effectively increased without hindering its accuracy so that the training process can be sped up by parallelizing throughout multiple GPUs, decreasing the computational time needed. Lastly, we expose the training performance differences between PyTorch and Tensorflow Deep Learning frameworks., This work was partially supported by the Spanish Ministry of Science and Innovation and the European Regional Development Fund under contract TIN2015- 65316-P, by the BSC-CNS Severo Ochoa program SEV-2015-0493, and grant 2017-SGR-1414 by Generalitat de Catalunya and by the research agreement CaixaBank-BSC 2016-2021., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Benchmarking of state-of-the-art HPC clusters with a production CFD code

  • Banchelli Gracia, Fabio
  • Garcia Gasulla, Marta
  • Houzeaux, Guillaume
  • Mantovani, Filippo
Computing technologies populating high-performance computing (HPC) clusters are getting more and more diverse, offering a wide range of architectural features. As a consequence, efficient programming of such platforms becomes a complex task. In this paper we provide a micro-benchmarking of three HPC clusters based on different CPU architectures, predominant in the Top500 ranking: x86, Armv8 and IBM Power9. On these platforms we study a production fluid-dynamics application leveraging different compiler technologies and micro-architectural features. We finally provide a scalability study on state-of-the-art HPC clusters. The two most relevant conclusions of our study are: i) Compiler development is critical for squeezing performance out of most recent technologies; ii) Micro-architectural features such as Single Instruction Multiple Data (SIMD) units and Simultaneous Multi-Threading (SMT) can impact the overall performance. However, a closer look shows that while SIMD is improving the performance of compute bound regions, SMT does not show a clear benefit on HPC workloads., The authors thank the support team of Dibona operating at ATOS/Bull. This work is partially supported by the Spanish Government through Programa Severo Ochoa (SEV-2015-0493), the Spanish Ministry of Science and Technology (project TIN2015-65316-P), the Generalitat de Catalunya (2017-SGR-1414), the European project Mont-Blanc 3 (GA n. 671697), and the European centre of excellence on HPC POP2 (GA n. 824080)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Echo state hoeffding tree learning

  • Marrón Vida, Diego
  • Read, Jesse
  • Bifet Figuerol, Albert Carles
  • Abdessalem, Talel
  • Ayguadé Parra, Eduard
  • Herrero Zaragoza, José Ramón
Nowadays, real-time classi cation of Big Data streams is becoming essential in a variety of application domains. While decision trees are powerful and easy{to{deploy approaches for accurate and fast learning from data streams, they are unable to capture the strong temporal dependences typically present in the input data. Recurrent Neural Networks are an alternative solution that include an internal memory to capture these temporal dependences; however their training is computationally very expensive, with slow convergence and not easy{to{deploy (large number of hyper-parameters). Reservoir Computing was proposed to reduce the computation requirements of the training phase but still include a feed-forward layer which requires a large number of parameters to tune. In this work we propose a novel architecture for real-time classification based on the combination of a Reservoir and a decision tree. This combination makes classification fast, reduces the number of hyper-parameters and keeps the good temporal properties of recurrent neural networks. The capabilities of the proposed architecture to learn some typical string-based functions with strong temporal dependences are evaluated in the paper. The paper shows how the new architecture is able to incrementally learn these functions in real-time with fast adaptation to unknown sequences and analyzes the influence of the reduced number of hyper-parameters in the behaviour of the proposed solution., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Improving HPC system throughput and response time using memory disaggregation

  • Vieira Zacarias, Felippe
  • Carpenter, Paul Matthew
  • Petrucci, Vinicius
© 2021 IEEE. 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., HPC clusters are cost-effective, well understood, and scalable, but the rigid boundaries between compute nodes may lead to poor utilization of compute and memory resources. HPC jobs may vary, by orders of magnitude, in memory consumption per core. Thus, even when the system is provisioned to accommodate normal and large capacity nodes, a mismatch between the system and the memory demands of the scheduled jobs can lead to inefficient usage of both memory and compute resources. Disaggregated memory has recently been proposed as a way to mitigate this problem by flexibly allocating memory capacity across cluster nodes. This paper presents a simulation approach for at-scale evaluation of job schedulers with disaggregated memories and it introduces a new disaggregated-aware job allocation policy for the Slurm resource manager. Our results show that using disaggregated memories, depending on the imbalance between the system and the submitted jobs, a similar throughput and job response time can be achieved on a system with up to 33% less total memory provisioning., This work is part of a project that has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 754337 (EuroEXA); it has been supported by the Spanish Ministry of Science and Innovation (project TIN2015-65316-P and Ramon y Cajal fellowship RYC2018-025628-I), Generalitat de Catalunya (contracts 2014-SGR-1051 and 2014-SGR-1272), and the Severo Ochoa Programme (SEV-2015-0493)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Random forest as a tumour genetic marker extractor

  • Pérez Arnal, Raquel Leandra
  • Garcia Gasulla, Dario
  • Torrents Rodas, David
  • Pares, Ferran
  • Cortés García, Claudio Ulises
  • Labarta Mancho, Jesús José
  • Ayguadé Parra, Eduard
Finding tumour genetic markers is essential to biomedicine due to their relevance for cancer detection and therapy development. In this paper, we explore a recently released dataset of chromosome rearrangements in 2,586 cancer patients, where different sorts of alterations have been detected. Using a Random Forest classifier, we evaluate the relevance of several features (some directly available in the original data, some engineered by us) related to chromosome rearrangements. This evaluation results in a set of potential tumour genetic markers, some of which are validated in the bibliography, while others are potentially novel., This work is partially supported by the Joint Study Agreement no. W156463 under the IBM/BSC Deep Learning Center agreement, by the Spanish Government through Programa Severo Ochoa (SEV-2015-0493), by the Spanish Ministry of Science and Technology through TIN2015-65316-P project, and by the Generalitat de Catalunya (contracts 2017-SGR-1414)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

STT-MRAM for real-time embedded systems: performance and WCET implications

  • Asifuzzaman, Kazi
  • Fernández, Mikel
  • Radojković, Petar
  • Abella Ferrer, Jaume
  • Cazorla Almeida, Francisco Javier
STT-MRAM is an emerging non-volatile memory quickly approaching DRAM in terms of capacity, frequency and device size. Intensified efforts in STT-MRAM research by the memory manufacturers may indicate a revolution with STT-MRAM memory technology is imminent, and therefore it is essential to perform system level research to explore use-cases and identify computing domains that could benefit from this technology. Special STT-MRAM features such as intrinsic radiation hardness, non-volatility, zero stand-by power and capability to function in extreme temperatures makes it particularly suitable for aerospace, avionics and automotive applications. Such applications often have real-time requirements --- that is, certain tasks must complete within a strict deadline. Analyzing whether this deadline is met requires Worst Case Execution Time (WCET) Analysis, which is a fundamental part of evaluating any real-time system. In this study, we investigate the feasibility of using STT-MRAM in real-time embedded systems by analyzing average system performance impact and WCET implications., This work was supported by BSC, Spanish Government through Programa Severo Ochoa (SEV-2015-0493), by the Spanish Ministry of Science and Technology through TIN2015-65316-P project and by the Generalitat de Catalunya (contracts 2014-SGR-1051 and 2014-SGR-1272). This work has also received funding from the European Union’s Horizon 2020 research and innovation programme under ExaNoDe project (grant agreement No 671578). Jaume Abella was partially supported by the Ministry of Economy and Competitive-ness under Ramon y Cajal postdoctoral fellowship RYC-2013-14717., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Contention-aware application performance prediction for disaggregated memory systems

  • Vieira Zacarias, Felippe
  • Nishtala, Rajiv
  • Carpenter, Paul Matthew
Disaggregated memory has recently been proposed as a way to allow flexible and fine-grained allocation of memory capacity to compute jobs. This paper makes an important step towards effective resource allocation on disaggregated memory systems. Specifically, we propose a generic approach to predict the performance degradation due to sharing of disaggregated memory. In contrast to prior work, cache capacity is not shared among multiple applications, which removes a major contributor to application performance. For this reason, our analysis is driven by the demand for memory bandwidth, which has been shown to have an important effect on application performance. We show that profiling the application slowdown often involves significant experimental error and noise, and to this end, we improve the accuracy by linear smoothing of the sensitivity curves. We also show that contention is sensitive to the ratio between read and write memory accesses, and we address this sensitivity by building a family of sensitivity curves according to the read/write ratios. Our results show that the methodology predicts the slowdown in application performance subject to memory contention with an average error of 1.19% and max error of 14.6%. Compared with state-of-the-art, the relative improvements are almost 24% on average and 33% for the worst case., This work is part of a project that has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 754337 (EuroEXA); it has been supported by the Spanish Ministry of Science and Innovation (project TIN2015-65316-P and Ramon y Cajal fellowship RYC2018-025628-I), Generalitat de Catalunya (contracts 2014-SGR-1051 and 2014-SGR1272), and the Severo Ochoa Programme (SEV-2015-0493)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Optimization of condensed matter physics application with OpenMP tasking model

  • Criado, Joel
  • Garcia Gasulla, Marta
  • Labarta Mancho, Jesús José
  • Chatterjee, Arghya
  • Hernández, Óscar
  • Sirvent, Raül
  • Álvarez, Gonzalo
The Density Matrix Renormalization Group (DMRG++) is a condensed matter physics application used to study superconductivity properties of materials. It’s main computations consist of calculating hamiltonian matrix which requires sparse matrix-vector multiplications. This paper presents task-based parallelization and optimization strategies of the Hamiltonian algorithm. The algorithm is implemented as a mini-application in C++ and parallelized with OpenMP. The optimization leverages tasking features, such as dependencies or priorities included in the OpenMP standard 4.5. The code refactoring targets performance as much as programmability. The optimized version achieves a speedup of 8.0 × with 8 threads and 20.5 × with 40 threads on a Power9 computing node while reducing the memory consumption to 90 MB with respect to the original code, by adding less than ten OpenMP directives., This work is partially supported by the Spanish Government through Programa Severo Ochoa (SEV2015-0493), by the Spanish Ministry of Science and Technology (project TIN2015-65316-P), by the Generalitat de Catalunya (contract 2017-SGR-1414) and by the BSC-IBM Deep Learning Research Agreement, under JSA “Application porting, analysis and optimization for POWER and POWER AI”. This work was partially supported by the Scientific Discovery through Advanced Computing (SciDAC) program funded by U.S. Department of Energy, Office of Science, Advanced Scientific Computing Research and Basic Energy Sciences, Division of Materials Sciences and Engineering. This research used resources of the Oak Ridge Leadership Computing Facility at the Oak Ridge National Laboratory, which is supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC05-00OR22725., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

A vulnerability factor for ECC-protected memory

  • Jaulmes, Luc
  • Moretó Planas, Miquel
  • Valero Cortés, Mateo
  • Casas, Marc
Fault injection studies and vulnerability analyses have been used to estimate the reliability of data structures in memory. We survey these metrics and look at their adequacy to describe the data stored in ECC-protected memory. We also introduce FEA, a new metric improving on the memory derating factor by ignoring a class of false errors. We measure all metrics using simulations and compare them to the outcomes of injecting errors in real runs. This in-depth study reveals that FEA provides more accurate results than any state-of-the-art vulnerability metric. Furthermore, FEA gives an upper bound on the failure probability due to an error in memory, making this metric a tool of choice to quantify memory vulnerability. Finally, we show that ignoring these false errors reduces the failure rate on average by 12.75% and up to over 45%., This work has been supported by the RoMoL ERC Advanced Grant (GA 321253), by the European HiPEAC Network of Excellence, by the Spanish Ministry of Economy and Competitiveness (contract TIN2015-65316- P), by the Generalitat de Catalunya (contracts 2017-SGR-1414 and 2017- SGR-1328), by the Spanish Government (Severo Ochoa grant SEV-2015- 0493) and by the European Union’s Horizon 2020 research and innovation programme (grant agreements 671697 and 779877). L. Jaulmes has been partially supported by the Spanish Ministry of Education, Culture and Sports under grant FPU2013/06982. M. Moreto and M. Casas have been partially supported by the Spanish Ministry of Economy, Industry and Competitiveness under Ramon y Cajal fellowships RYC-2016-21104 and RYC-2017-23269., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

DRAM errors in the field: a statistical approach

  • Živanovič, Darko
  • Esmaili Dokht, Pouya
  • Moré, Sergi
  • Bartolomé, Javier
  • Carpenter, Paul Matthew
  • Radojković, Petar
  • Ayguadé Parra, Eduard
This paper summarizes our two-year study of corrected and uncor-rected errors on the MareNostrum 3 supercomputer, covering 2000 billion MB-hours of DRAM in the field. The study analyzes 4.5 million corrected and 71 uncorrected DRAM errors and it compares the reliability of DIMMs from all three major memory manufacturers, built in three different technologies. Our work has two sets of contributions. First, we illustrate the complexity of in-field DRAM error analysis and demonstrate the limitations of various widely-used methods and metrics. For example, we show that average error rates, errors per MB-hour and mean time between failures can provide volatile and unreliable results even after long periods of error logging, leading to incorrect conclusions about DRAM reliability. Second, we present formal statistical methods that overcome many of the limitations of the current approaches. The methods that we present are simple to understand and implement, reliable and widely accepted in the statistical community. Overall, our study alerts the community about the need to, firstly, question the current practice in quantifying DRAM reliability and, secondly, to select a proper analysis approach for future studies. Our strong recommendations are to focus on metrics with a practical value that could be easily related to system reliability, and to select methods that provide stable results, ideally supported with statistical significance., This work was supported by the Collaboration Agreement between Samsung Electronics Co., Ltd. and BSC, Spanish Government through Severo Ochoa programme (SEV-2015-0493), by the Spanish Ministry of Science and Technology through TIN2015-65316-P project and by the Generalitat de Catalunya (contracts 2014-SGR1051 and 2014-SGR-1272). This work has also received funding from the European Union’s Horizon 2020 research and innovation programme under EuroEXA project (grant agreement No 754337). Darko Zivanovic holds the Severo Ochoa grant (SVP-2014-068501) of the Ministry of Economy and Competitiveness of Spain., Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Large-memory nodes for energy efficient high-performance computing

  • Živanovič, Darko
  • Radulović, Milan
  • Llort, German
  • Zaragoza, David
  • Strassburg, Janko
  • Carpenter, Paul Matthew
  • Radojković, Petar
  • Ayguadé Parra, Eduard
Energy consumption is by far the most important contributor to HPC cluster operational costs, and it accounts for a significant share of the total cost of ownership. Advanced energy-saving techniques in HPC components have received significant research and development effort, but a simple measure that can dramatically reduce energy consumption is often overlooked. We show that, in capacity computing, where many small to medium-sized jobs have to be solved at the lowest cost, a practical energy-saving approach is to scale-in the application on large-memory nodes. We evaluate scaling-in; i.e. decreasing the number of application processes and compute nodes (servers) to solve a fixed-sized problem, using a set of HPC applications running in a production system. Using standard-memory nodes, we obtain average energy savings of 36%, already a huge figure. We show that the main source of these energy savings is a decrease in the node-hours (node_hours = #nodes x exe_time), which is a consequence of the more efficient use of hardware resources. Scaling-in is limited by the per-node memory capacity. We therefore consider using large-memory nodes to enable a greater degree of scaling-in. We show that the additional energy savings, of up to 52%, mean that in many cases the investment in upgrading the hardware would be recovered in a typical system lifetime of less than five years., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Workflow environments for advanced cyberinfrastructure platforms

  • Badia Sala, Rosa Maria
  • Ejarque Artigas, Jorge
  • Lordan Gomis, Francesc
  • Lezzi, Daniele
  • Conejero Bañón, Javier
  • Álvarez Cid-Fuentes, Javier
  • Becerra Fontal, Yolanda
  • Queralt Calafat, Anna
Progress in science is deeply bound to the effective use of high-performance computing infrastructures and to the efficient extraction of knowledge from vast amounts of data. Such data comes from different sources that follow a cycle composed of pre-processing steps for data curation and preparation for subsequent computing steps, and later analysis and analytics steps applied to the results. However, scientific workflows are currently fragmented in multiple components, with different processes for computing and data management, and with gaps in the viewpoints of the user profiles involved. Our vision is that future workflow environments and tools for the development of scientific workflows should follow a holistic approach, where both data and computing are integrated in a single flow built on simple, high-level interfaces. The topics of research that we propose involve novel ways to express the workflows that integrate the different data and compute processes, dynamic runtimes to support the execution of the workflows in complex and heterogeneous computing infrastructures in an efficient way, both in terms of performance and energy. These infrastructures include highly distributed resources, from sensors and instruments, and devices in the edge, to High-Performance Computing and Cloud computing resources. This paper presents our vision to develop these workflow environments and also the steps we are currently following to achieve it., This work has been supported by the Spanish Government (SEV2015-0493), by the Spanish Ministry of Science and Innovation (contract TIN2015-65316-P), by Generalitat de Catalunya (contract 2014-SGR-1051). Javier Conejero postdoctoral contract is co-financed by the Ministry of Economy and Competitiveness under Juan de la Cierva Formacion´ postdoctoral fellowship number FJCI-2015-24651. This work is supported by the H2020 mF2C project (730929) and the CLASS project (780622). The participation of Rosa M Badia in the BDEC2 meetings is supported by the EXDCI project (800957). The dislib library developments are partially funded under the project agreement between BSC and FUJITSU., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Continuous-action reinforcement learning for memory allocation in virtualized servers

  • Garrido Platero, Luis Ángel
  • Nishtala, Rajiv
  • Carpenter, Paul Matthew
In a virtualized computing server (node) with multiple Virtual Machines (VMs), it is necessary to dynamically allocate memory among the VMs. In many cases, this is done only considering the memory demand of each VM without having a node-wide view. There are many solutions for the dynamic memory allocation problem, some of which use machine learning in some form. This paper introduces CAVMem (Continuous-Action Algorithm for Virtualized Memory Management), a proof-of-concept mechanism for a decentralized dynamic memory allocation solution in virtualized nodes that applies a continuous-action reinforcement learning (RL) algorithm called Deep Deterministic Policy Gradient (DDPG). CAVMem with DDPG is compared with other RL algorithms such as Q-Learning (QL) and Deep Q-Learning (DQL) in an environment that models a virtualized node. In order to obtain linear scaling and be able to dynamically add and remove VMs, CAVMem has one agent per VM connected via a lightweight coordination mechanism. The agents learn how much memory to bid for or return, in a given state, so that each VM obtains a fair level of performance subject to the available memory resources. Our results show that CAVMem with DDPG performs better than QL and a static allocation case, but it is competitive with DQL. However, CAVMem incurs significant less training overheads than DQL, making the continuous-action approach a more cost-effective solution., This research is part of a project that has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 754337 (EuroEXA) and the European Union’s 7th Framework Programme under grant agreement number 610456 (Euroserver). It also received funding from the Spanish Ministry of Science and Technology (project TIN2015-65316-P), Generalitat de Catalunya (contract 2014-SGR-1272), and the Severo Ochoa Programme (SEV-2015-0493) of the Spanish Government., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Micro-architectural characterization of Apache Spark on batch and stream processing workloads

  • Awan, Ahsan
  • Brorsson, Mats
  • Vlassov, Vladimir
  • Ayguadé Parra, Eduard
While cluster computing frameworks are continuously evolving to provide real-time data analysis capabilities, Apache Spark has managed to be at the forefront of big data analytics for being a unified framework for both, batch and stream data processing. However, recent studies on micro-architectural characterization of in-memory data analytics are limited to only batch processing workloads. We compare the micro-architectural performance of batch processing and stream processing workloads in Apache Spark using hardware performance counters on a dual socket server. In our evaluation experiments, we have found that batch processing and stream processing has same micro-architectural behavior in Spark if the difference between two implementations is of micro-batching only. If the input data rates are small, stream processing workloads are front-end bound. However, the front end bound stalls are reduced at larger input data rates and instruction retirement is improved. Moreover, Spark workloads using DataFrames have improved instruction retirement over workloads using RDDs., Peer Reviewed, Postprint (published version)