Pasar al contenido principal

TECNICAS ALGORITMICAS PARA COMPUTACION DE ALTO RENDIMIENTO CONSCIENTE DEL CONSUMO ENERGETICO Y RESISTENTE A ERRORES

TIN2017-82972-R

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal de I+D+i Orientado a los Retos de la Sociedad
Convocatoria Proyectos I+D+i Retos de Investigación
Año convocatoria 2017
Plan Estatal 2017-2020 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSITAT JAUME I DE CASTELLO

Resultados relacionados

Resultados totales (Incluyendo duplicados): 29
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Improving the management efficiency of GPU workloads in data centers through GPU virtualization

  • Iserte, Sergio
  • Reaño González, Carlos
  • Prades, Javier|||0000-0003-3349-2200
  • Silla, Federico|||0000-0002-6435-1200
[EN] Graphics processing units (GPUs) are currently used in data centers to reduce the execution time of compute-intensive applications. However, the use of GPUs presents several side effects, such as increased acquisition costs and larger space requirements. Furthermore, GPUs require a nonnegligible amount of energy even while idle. Additionally, GPU utilization is usually low for most applications. In a similar way to the use of virtual machines, using virtual GPUs may address the concerns associated with the use of these devices. In this regard, the remote GPU virtualization mechanism could be leveraged to share the GPUs present in the computing facility among the nodes of the cluster. This would increase overall GPU utilization, thus reducing the negative impact of the increased costs mentioned before. Reducing the amount of GPUs installed in the cluster could also be possible. However, in the same way as job schedulers map GPU resources to applications, virtual GPUs should also be scheduled before job execution. Nevertheless, current job schedulers are not able to deal with virtual GPUs. In this paper, we analyze the performance attained by a cluster using the remote Compute Unified Device Architecture middleware and a modified version of the Slurm scheduler, which is now able to assign remote GPUs to jobs. Results show that cluster throughput, measured as jobs completed per time unit, is doubled at the same time that the total energy consumption is reduced up to 40%. GPU utilization is also increased., Generalitat Valenciana, Grant/Award Number: PROMETEO/2017/077; MINECO and FEDER, Grant/Award Number: TIN2014-53495-R, TIN2015-65316-P and TIN2017-82972-R
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Selecting optimal SpMV realizations for GPUs via machine learning

  • Dufrechou, Ernesto
  • Ezzatti, Pablo
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] More than 10 years of research related to the development of efficient GPU routines for the sparse matrix-vector product (SpMV) have led to several realizations, each with its own strengths and weaknesses. In this work, we review some of the most relevant efforts on the subject, evaluate a few prominent routines that are publicly available using more than 3000 matrices from different applications, and apply machine learning techniques to anticipate which SpMV realization will perform best for each sparse matrix on a given parallel platform. Our numerical experiments confirm the methods offer such varied behaviors depending on the matrix structure that the identification of general rules to select the optimal method for a given matrix becomes extremely difficult, though some useful strategies (heuristics) can be defined. Using a machine learning approach, we show that it is possible to obtain unexpensive classifiers that predict the best method for a given sparse matrix with over 80% accuracy, demonstrating that this approach can deliver important reductions in both execution time and energy consumption, The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: ES Quintana-Ort was supported by project TIN2017-82972-R of the MINECO and FEDER
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Ginkgo: A Modern Linear Operator Algebra Framework for High Performance Computing

  • Anzt, Hartwig
  • Cojean, Terry
  • Flegar, Goran
  • Göbel, Fritz
  • Grützmacher, Thomas
  • Nayak, Pratik
  • Ribizel, Tobias
  • Tsai, Yuhsiang Mike
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] In this article, we present GINKGO, a modern C++ math library for scientific high performance computing. While classical linear algebra libraries act on matrix and vector objects, Gnswo's design principle abstracts all functionality as linear operators," motivating the notation of a "linear operator algebra library" GINKGO'S current focus is oriented toward providing sparse linear algebra functionality for high performance graphics processing unit (GPU) architectures, but given the library design, this focus can be easily extended to accommodate other algorithms and hardware architectures. We introduce this sophisticated software architecture that separates core algorithms from architecture-specific backends and provide details on extensibility and sustainability measures. We also demonstrate GINKGO'S usability by providing examples on how to use its functionality inside the MFEM and deal.ii finite element ecosystems. Finally, we offer a practical demonstration of GINKGO'S high performance on state-of-the-art GPU architectures., This work was supported by the "Impuls und Vernetzungsfond of the Helmholtz Association" under grant VH-NG-1241. G. Flegar and E. S. Quintana-Orti were supported by project TIN2017-82972-R of the MINECO and FEDER and the H2020 EU FETHPC Project 732631 "OPRECOMP". This researchwas also supported by the Exascale Computing Project (17-SC-20-SC), a collaborative effort of the U.S. Department of Energy Office of Science and the National Nuclear Security Administration. The experiments on the NVIDIA A100 GPU were performed on the HAICORE@KIT partition, funded by the "Impuls und Vernetzungsfond" of the Helmholtz Association. The experiments on the AMD MI100 GPU were performed on Tulip, an early-access platform hosted by HPE.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Programming parallel dense matrix factorizations with look-ahead and OpenMP

  • Catalán, Sandra
  • Igual, Francisco D.
  • Rodríguez-Sánchez, Rafael
  • Castelló, Adrián|||0000-0002-8576-8451
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] We investigate a parallelization strategy for dense matrix factorization (DMF) algorithms, using OpenMP, that departs from the legacy (or conventional) solution, which simply extracts concurrency from a multi-threaded version of basic linear algebra subroutines (BLAS). The proposed approach is also different from the more sophisticated runtime-based implementations, which decompose the operation into tasks and identify dependencies via directives and runtime support. Instead, our strategy attains high performance by explicitly embedding a static look-ahead technique into the DMF code, in order to overcome the performance bottleneck of the panel factorization, and realizing the trailing update via a cache-aware multi-threaded implementation of the BLAS. Although the parallel algorithms are specified with a high level of abstraction, the actual implementation can be easily derived from them, paving the road to deriving a high performance implementation of a considerable fraction of linear algebra package (LAPACK) functionality on any multicore platform with an OpenMP-like runtime., The researchers from Universidad Jaume I were supported by the CICYT Projects TIN2014-53495-R and TIN2017-82972-R of the MINECO and FEDER, and the H2020 EU FETHPC Project 671602 "INTERTWinE". The researchers from Universidad Complutense de Madrid were supported by the CICYT Project TIN2015-65277-R of the MINECO and FEDER. Sandra Catalan was supported during part of this time by the FPU program of the Ministerio de Educacion, Cultura y Deporte. Adrian Castello was supported by the ValI+D 2015 FPI program of the Generalitat Valenciana.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Performance modeling of the sparse matrix-vector product via convolutional neural networks

  • Barreda, María
  • Dolz, Manuel F.
  • CASTAÑO ALVAREZ, MARIA ASUNCION
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] Modeling the execution time of the sparse matrix-vector multiplication (SpMV) on a current CPU architecture is especially complex due to (i) irregular memory accesses; (ii) indirect memory referencing; and (iii) low arithmetic intensity. While analytical models may yield accurate estimates for the total number of cache hits/misses, they often fail to predict accurately the total execution time. In this paper, we depart from the analytic approach to instead leverage convolutional neural networks (CNNs) in order to provide an effective estimation of the performance of the SpMV operation. For this purpose, we present a high-level abstraction of the sparsity pattern of the problem matrix and propose a blockwise strategy to feed the CNN models by blocks of nonzero elements. The experimental evaluation on a representative subset of the matrices from the SuiteSparse Matrix collection demonstrates the robustness of the CNN models for predicting the SpMV performance on an Intel Haswell core. Furthermore, we show how to generalize the network models to other target architectures to estimate the performance of SpMV on an ARM A57 core, This work was supported by project TIN2017-82972-R from the MINECO, Spain. Manuel F. Dolz was also supported by the Plan GenT project CDEIGENT/2018/014 from the Generalitat Valenciana, Spain. Maria Barreda was also supported by the POSDOC-A/2017/11 project from the Universitat Jaume I
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

High performance and energy efficient inference for deep learning on multicore ARM processors using general optimization techniques and BLIS

  • Castelló, Adrián|||0000-0002-8576-8451
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
  • Tomás Domínguez, Andrés Enrique
  • SERGIO BARRACHINA
  • DOLZ ZARAGOZÁ, MANUEL FRANCISCO
  • San Juan-Sebastian, Pablo
[EN] We evolve PyDTNN, a framework for distributed parallel training of Deep Neural Networks (DNNs), into an efficient inference tool for convolutional neural networks. Our optimization process on multicore ARM processors involves several high-level transformations of the original framework, such as the development and integration of Cython routines to exploit thread-level parallelism; the design and development of micro-kernels for the matrix multiplication, vectorized with ARM's NEON intrinsics, that can accommodate layer fusion; and the appropriate selection of several cache configuration parameters tailored to the memory hierarchy of the target ARM processors.Our experiments evaluate both inference throughput (measured in processed images/s) and inference latency (i.e., time-to-response) as well as energy consumption per image when varying the level of thread parallelism and the processor power modes. The experiments with the new inference engine are reported for the ResNet50 v1.5 model on the ImageNet dataset from the MLPerf suite using the ARM v8.2 cores in the NVIDIA Jetson AGX Xavier board. These results show superior performance compared with the well-spread TFLite from Google and slightly inferior results when compared with ArmNN, the native library from ARM for DNN inference., This research was partially sponsored by projects TIN2017-82972-R of Ministerio de Ciencia, Innovacion y Universidades, Spain and Prometeo/2019/109 of the Generalitat Valenciana, Spain. Adrian Castello was supported by the Juan de la Cierva-Formacion project FJC2019-039222-I of the Ministerio de Ciencia, Innovacion y Universidades, Spain. Manuel F. Dolz was also supported by the Plan GenT project CDEIGENT/2018/014 of the Generalitat Valenciana, Spain.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

FloatX: A C++ Library for Customized Floating-Point Arithmetic

  • Flegar, Goran
  • Scheidegger, Florian
  • Novakovic, Vedran
  • Mariani, Giovani
  • Malossi, Cristiano
  • Tomás Domínguez, Andrés Enrique
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] We present FloatX (Float eXtended), a C++ framework to investigate the effect of leveraging customized floating-point formats in numerical applications. FloatX formats are based on binary IEEE 754 with smaller significand and exponent bit counts specified by the user. Among other properties, FloatX facilitates an incremental transformation of the code, relies on hardware-supported floating-point types as back-end to preserve efficiency, and incurs no storage overhead. The article discusses in detail the design principles, programming interface, and datatype casting rules behind FloatX. Furthermore, it demonstrates FloatX's usage and benefits via several case studies from well-known numerical dense linear algebra libraries, such as BLAS and LAPACK; the Ginkgo library for sparse linear systems; and two neural network applications related with image processing and text recognition., This work was supported by the CICYT projects TIN2014-53495-R and TIN2017-82972-R of the MINECO and FEDER, and the EU H2020 project 732631 "OPRECOMP. Open Transprecision Computing."
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Tall-and-skinny QR factorization with approximate Householder reflectors on graphics processors

  • Tomás Domínguez, Andrés Enrique
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] We present a novel method for the QR factorization of large tall-and-skinny matrices that introduces an approximation technique for computing the Householder vectors. This approach is very competitive on a hybrid platform equipped with a graphics processor, with a performance advantage over the conventional factorization due to the reduced amount of data transfers between the graphics accelerator and the main memory of the host. Our experiments show that, for tall¿skinny matrices, the new approach outperforms the code in MAGMA by a large margin, while it is very competitive for square matrices when the memory transfers and CPU computations are the bottleneck of the Householder QR factorization, This research was supported by the Project TIN2017-82972-R from the MINECO (Spain) and the EU H2020 Project 732631 "OPRECOMP. Open Transprecision Computing".
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

On the performance of a GPU-based SoC in a distributed spatial audio system

  • Belloch, José A.
  • Badía, José M.
  • Larios, Diego F.
  • Personal, Enrique
  • Lupoiu, Mihaita
  • León, Carlos
  • Ferrer Contreras, Miguel|||0000-0002-8743-1887
  • Fuster Criado, Laura|||0000-0003-2633-0449
  • Gonzalez, Alberto|||0000-0002-6984-3212
  • Vidal Maciá, Antonio Manuel
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] Many current system-on-chip (SoC) devices are composed of low-power multicore processors combined with a small graphics accelerator (or GPU) offering a trade-off between computational capacity and low-power consumption. In this context, spatial audio methods such as wave field synthesis (WFS) can benefit from a distributed system composed of several SoCs that collaborate to tackle the high computational cost of rendering virtual sound sources. This paper aims at evaluating important aspects dealing with a distributed WFS implementation that runs over a network of Jetson Nano boards composed of embedded GPU-based SoCs: computational performance, energy efficiency, and synchronization issues. Our results show that the maximum efficiency is obtained when the WFS system operates the GPU frequency at 691.2 MHz, achieving 11 sources-per-Watt. Synchronization experiments using the NTP protocol show that the maximum initial delay of 10 ms between nodes does not prevent us from achieving high spatial sound quality., This work has been supported by the Spanish Government through TIN2017-82972-R, ESP2015-68245-C4-1-P, the Valencian Regional Government through PROMETEO/2019/109 and the Universitat Jaume I through UJI-B2019-36.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Low precision matrix multiplication for efficient deep learning in NVIDIA Carmel processors

  • San Juan-Sebastian, Pablo
  • Rodríguez-Sánchez, Rafael
  • Igual, Francisco D.
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] We introduce a high performance, multi-threaded realization of the gemm kernel for the ARMv8.2 architecture that operates with 16-bit (half precision)/queryKindly check and confirm whether the corresponding author is correctly identified. floating point operands. Our code is especially designed for efficient machine learning inference (and to a certain extent, also training) with deep neural networks. The results on the NVIDIA Carmel multicore processor, which implements the ARMv8.2 architecture, show considerable performance gains for the gemm kernel, close to the theoretical peak acceleration that could be expected when moving from 32-bit arithmetic/data to 16-bit. Combined with the type of convolution operator arising in convolutional neural networks, the speed-ups are more modest though still relevant., This work was supported by projects TIN2017-82972-R and RTI2018-093684-B-I00 from the Ministerio de Ciencia, Innovacion y Universidades, project S2018/TCS-4423 of the Comunidad de Madrid, project PR65/19-22445 of the UCM, and project Prometeo/2019/109 of the Generalitat Valenciana.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

An Efficient Implementation of Parallel Parametric HRTF Models for Binaural Sound Synthesis in Mobile Multimedia

  • Belloch, José A.
  • Badia, Jose M.
  • Cobos, Máximo
  • Ramos Peinado, Germán|||0000-0001-7152-0044
[EN] The extended use of mobile multimedia devices in applications like gaming, 3D video and audio reproduction, immersive teleconferencing, or virtual and augmented reality, is demanding efficient algorithms and methodologies. All these applications require real-time spatial audio engines with the capability of dealing with intensive signal processing operations while facing a number of constraints related to computational cost, latency and energy consumption. Most mobile multimedia devices include a Graphics Processing Unit (GPU) that is primarily used to accelerate video processing tasks, providing high computational capabilities due to its inherent parallel architecture. This paper describes a scalable parallel implementation of a real-time binaural audio engine for GPU-equipped mobile devices. The engine is based on a set of head-related transfer functions (HRTFs) modelled with a parametric parallel structure, allowing efficient synthesis and interpolation while reducing the size required for HRTF data storage. Several strategies to optimize the GPU implementation are evaluated over a well-known kind of processor present in a wide range of mobile devices. In this context, we analyze both the energy consumption and real-time capabilities of the system by exploring different GPU and CPU configuration alternatives. Moreover, the implementation has been conducted using the OpenCL framework, guarantying the portability of the code., This work was supported in part by the Spanish Ministry of Economy and Competitiveness under Grant RTI2018-097045-B-C21, Grant RTI2018-097045-B-C22, Grant TIN2017-82972-R, and Grant ESP2015-68245-C4-1-P, and in part by the Universitat Jaume I Project UJI-B2019-36.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

DMRlib: Easy-coding and efficient resource management for job malleability

  • Iserte, Sergio
  • Mayo, Rafael
  • Peña Monferrer, Antonio José
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] Process malleability has proved to have a highly positive impact on the resource utilization and global productivity in data centers compared with the conventional static resource allocation policy. However, the non-negligible additional development effort this solution imposes has constrained its adoption by the scientific programming community. In this work, we present DMRlib, a library designed to offer the global advantages of process malleability while providing a minimalist MPI-like syntax. The library includes a series of predefined communication patterns that greatly ease the development of malleable applications. In addition, we deploy several scenarios to demonstrate the positive impact of process malleability featuring different scalability patterns. Concretely, we study two job submission modes (rigid and moldable) in order to identify the best-case scenarios for malleability using metrics such as resource allocation rate, completed jobs per second, and energy consumption. The experiments prove that our elastic approach may improve global throughput by a factor higher than 3x compared to the traditional workloads of non-malleable jobs, This work was supported in part by Projects TIN2014-53495-R, TIN2015-65316-P, and TIN2017-82972-R from MINECO and FEDER. This project has received funding fromthe European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska Curie Grant 749516. Thework of Sergio Isertewas supported by a postdoctoral fellowship from Generalitat Valenciana and European Social Fund APOSTD/2020/026. Finally, the authors would like to thank the anonymous reviewers whose suggestions significantly improved the quality of this article
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

On the use of many-core machines for the acceleration of a mesh truncation technique for FEM

  • Belloch Rodríguez, José Antonio
  • Amor-Martin, Adrian
  • García-Donoro, Daniel
  • Garcia-Castillo, Luis E.
  • Martínez Zaldívar, Francisco José|||0000-0002-7845-6273
[EN] Finite element method (FEM) has been used for years for radiation problems in the field of electromagnetism. To tackle problems of this kind, mesh truncation techniques are required, which may lead to the use of high computational resources. In fact, electrically large radiation problems can only be tackled using massively parallel computational resources. Different types of multi-core machines are commonly employed in diverse fields of science for accelerating a number of applications. However, properly managing their computational resources becomes a very challenging task. On the one hand, we present a hybrid message passing interface + OpenMP-based acceleration of a mesh truncation technique included in a FEM code for electromagnetism in a high-performance computing cluster equipped with 140 compute nodes. Results show that we obtain about 85% of the theoretical maximum speedup of the machine. On the other hand, a graphics processing unit has been used to accelerate one of the parts that presents high fine-grain parallelism., This work has been financially supported by TEC2016-80386-P, TIN2017-82972-R, CAM S2013/ICE-3004 projects and "Ayudas para contratos predoctorales de Formacion del Profesorado Universitario FPU".
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Accelerating the task/data-parallel version of ILUPACK¿s BiCG in multi-CPU/GPU configurations

  • Aliaga, Jose I.
  • Dufrechou, Ernesto
  • Ezzatti, Pablo
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] ILUPACK is a valuable tool for the solution of sparse linear systems via iterative Krylov subspace-based methods. Its relevance for the solution of real problems has motivated several efforts to enhance its performance on parallel machines. In this work we focus on exploiting the task-level parallelism derived from the structure of the BiCG method, in addition to the data-level parallelism of the internal matrix computations, with the goal of boosting the performance of a GPU (graphics processing unit) implementation of this solver. First, we revisit the use of dual-GPU systems to execute independent stages of the BiCG concurrently on both accelerators, while leveraging the extra memory space to improve the data access patterns. In addition, we extend our ideas to compute the BiCG method efficiently in multicore platforms with a single GPU. In this line, we study the possibilities offered by hybrid CPU-GPU computations, as well as a novel synchronization-free sparse triangular linear solver. The experimental results with the new solvers show important acceleration factors with respect to the previous data-parallel CPU and GPU versions. (C) 2019 Elsevier B.V. All rights reserved., J. I. Aliaga and E. S. Quintana-Orti were supported by project TIN2017-82972-R of the MINECO and FEDER. E. Dufrechou and P. Ezzatti were supported by Programa de Desarrollo de las Ciencias Basicas (PEDECIBA), Uruguay.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Exploiting nested task-parallelism in the H-LU factorization

  • Carratalá-Sáez, Rocío
  • Christophersen, Sven
  • Aliaga, José I.
  • Beltrán, Vicenç
  • Börm, Steffen
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] We address the parallelization of the LU factorization of hierarchical matrices (H-matrices) arising from boundary element methods. Our approach exploits task-parallelism via the OmpSs programming model and runtime, which discovers the data-flow parallelism intrinsic to the operation at execution time, via the analysis of data dependencies based on the memory addresses of the tasks' operands. This is especially challenging for H-matrices, as the structures containing the data vary in dimension during the execution. We tackle this issue by decoupling the data structure from that used to detect dependencies. Furthermore, we leverage the support for weak operands and early release of dependencies, recently introduced in OmpSs-2, to accelerate the execution of parallel codes with nested task-parallelism and fine-grain tasks. As a result, we obtain a significant improvement in the parallel performance with respect to our previous work., The researchers from Universidad Jaume I (UJI) were supported by projects CICYT TIN2014-53495-R and TIN2017-82972-R of MINECO and FEDER; project UJI-B2017-46 of UJI; and the FPU program of MECD.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Accessible C-programming course from scratch using a MOOC platform without limitations

  • Castelló, Adrián
  • Iserte, Sergio
  • Belloch, Jose Antonio
[EN] The C language has been used for ages in the application development in multidisciplinary environments. However, in the academia, this language is being replaced by other higher-level languages due to they are easier to understand, learn and apply. Moreover, the necessity of professionals with a good knowledge in those high-level languages is constantly increasing because of the boosting of mobile devices. This scenario generates a lack of low-level language programmers, required in other less trendy fields, but equal or more important, such as science, engineering or research.  In order to revive the interest in low-level languages and provide those minority fields with well-prepared staff, we present in this work a MOCC C-programming course that is addressed to any kind of people with or without IT background. A feature that differentiates this course from others programming online-based courses is that we mainly focus on the C language syntax providing, via a self-tuned virtual machine, an encapsulated environment that hides any interaction with the command-line of the underlying operating system. A secondary target of this work is to foster the computer science degree students to enrol the computer architecture specialization at the Universitat Jaume I (Spain). For this purpose, the High Performance Computing and Architectures research group of that University has decided to use this C course as a tool for fulfill the gap of the current syllabus. The results show that half of the participants that completed the first session of the course have satisfactorily finished the course, and the number of computer science degree students that chose the computer architecture specialization the following academic course was increment by 3x., This research has been partly funded by TIN2017-82972-R. Adrián Castelló was supported by the ValI+D 2015 FPI program of the Generalitat Valenciana.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Integration and exploitation of intra-routine malleability in BLIS

  • Rodríguez-Sánchez, Rafael
  • Igual, Francisco D.
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] Malleability is a property of certain applications (or tasks) that, given an external request or autonomously, can accommodate a dynamic modification of the degree of parallelism being exploited at runtime. Malleability improves resource usage (core occupation) on modern multicore architectures for applications that exhibit irregular and divergent execution paths and heavily depend on the underlying library performance to attain high performance. The integration of malleability within high-performance instances of the Basic Linear Algebra Subprograms (BLAS) is nonexistent, and, in addition, it is difficult to attain given the rigidity of current application programming interfaces (APIs). In this paper, we overcome these issues presenting the integration of a malleability mechanism within BLIS, a high-performance and portable framework to implement BLAS-like operations. For this purpose, we leverage low-level (yet simple) APIs to integrate on-demand malleability across all Level-3 BLAS routines, and we demonstrate the performance benefits of this approach by means of a higher-level dense matrix operation: the LU factorization with partial pivoting and look-ahead, The researchers from Universidad Complutense de Madrid were supported by the EU (FEDER) and Spanish MINECO (TIN2015-65277-R, RTI2018-093684-B-I00), and by Spanish CM (S2018/TCS-4423). The researcher from Universitat Poliecnica de Valencia was supported by the Spanish MINECO (TIN2017-82972-R)
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Static scheduling of the LU factorization with look-ahead on asymmetric multicore processors

  • Catalán, Sandra
  • Herrero, José R.
  • Rodríguez-Sánchez, Rafael
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] We analyze the benefits of look-ahead in the parallel execution of the LU factorization with partial pivoting (LUpp) in two distinct "asymmetric" multicore scenarios. The first one corresponds to an actual hardware-asymmetric architecture such as the Samsung Exynos 5422 system-on-chip (SoC), equipped with an ARM big.LITTLE processor consisting of a quad core Cortex-A15 cluster plus a quad-core Cortex-A7 cluster. For this scenario, we propose a careful mapping of the different types of tasks appearing in LUpp to the computational resources, in order to produce an efficient architecture-aware exploitation of the computational resources integrated in this SoC. The second asymmetric configuration appears in a hardware-symmetric multicore architecture where the cores can individually operate at a different frequency levels. In this scenario, we show how to employ the frequency slack to accelerate the tasks in the critical path of LUpp in order to produce a faster global execution as well as a lower energy consumption. (C) 2018 Elsevier B.V. All rights reserved., The researchers from Universidad Jaume I were supported by projects TIN2014-53495-R and TIN2017-82972-R of MINECO and FEDER, and the FPU program of MECD. The researcher from Universitat Politecnica de Catalunya was supported by projects TIN2015-65316-P of MINECO and FEDER and 2017-SGR-1414 from the Generalitat de Catalunya.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Acceleration of PageRank with customized precision based on mantissa segmentation

  • Gruetzmacher, Thomas
  • Cojean, Terry
  • Flegar, Goran
  • Anzt, Hartwig
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] We describe the application of a communication-reduction technique for the PageRank algorithm that dynamically adapts the precision of the data access to the numerical requirements of the algorithm as the iteration converges. Our variable-precision strategy, using a customized precision format based on mantissa segmentation (CPMS), abandons the IEEE 754 single- and double-precision number representation formats employed in the standard implementation of PageRank, and instead handles the data in memory using a customized floating-point format. The customized format enables fast data access in different accuracy, prevents overflow/underflow by preserving the IEEE 754 double-precision exponent, and efficiently avoids data duplication, since all bits of the original IEEE 754 double-precision mantissa are preserved in memory, but re-organized for efficient reduced precision access. With this approach, the truncated values (omitting significand bits), as well as the original IEEE double-precision values, can be retrieved without duplicating the data in different formats. Our numerical experiments on an NVIDIA V100 GPU (Volta architecture) and a server equipped with two Intel Xeon Platinum 8168 CPUs (48 cores in total) expose that, compared with a standard ieee double-precision implementation, the CPMS-based PageRank completes about 10% faster if high-accuracy output is needed, and about 30% faster if reduced output accuracy is acceptable., H. Anzt was supported by the "Impuls und Vernetzungsfond" of the Helmholtz Association under grant VH-NG-1241. G. Flegar and E. S. Quintana-Orti were supported by project TIN2017-82972-R of the MINECO and FEDER. This work was also supported by the EU H2020 project 732631 "OPRECOMP. Open Transprecision Computing,' and the US Department of Energy Office of Science, Office of Advanced Scientific Computing Research, Applied Mathematics program under Award Numbers DE-SC0016513 and DE-SC-0010042
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Efficient update of determinants of rmany-electron wave function overlaps

  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
  • Davidovic, Davor
  • Sapunar, Marin
  • Herrero, Jose R.
[EN] The calculation of overlaps between many-electron wave functions at different nuclear geometries during nonadiabatic dynamics simulations requires the evaluation of a large number of determinants of matrices that differ only in a few rows/columns. While this calculation is fast for small systems, its cost grows faster than the alternative electronic structure calculation used to obtain the wave functions. For wave functions that can be written as a CIS expansion, all determinants can be computed using the set of level-2 minors of the reference matrix. However, this is still a costly computation for large systems. In this paper, we provide an algorithm for efficiently calculating all level-2 minors of a matrix by re-utilizing and updating the LU factorization for the determinants of the minors. This approach results in a parallel version of the algorithm that is up to an order of magnitude faster then the current best parallel implementation. The algorithm thus allows the computation of exact wave function overlaps for relatively large systems, with a high density of states, at virtually no cost compared with the electronic structure calculations. Furthermore, the new algorithm opens the path to further investigations in efficient computing of the exact wave function overlaps for complex wave functions such as MR-CIS and MR-CISD. Program summary Program Title: CIS Overlap Licensing provisions: MIT Programming language: FORTRAN 2008, C Nature of problem: Calculation of overlaps between CIS type wave functions at different nuclear geometries during nonadiabatic dynamics simulation requires calculating a large number of connected determinants and scales with the seventh power of the size of the system being studied. Without additional approximations, for large systems this computation becomes more costly than the electronic structure calculation used to obtain the wave functions. Solution method: All of the determinants required for a CIS wave function overlap calculation can be derived from the set of all level-2 minors of the matrix of overlaps between the reference Slater determinants. We developed an algorithm for efficiently computing all level-2 minors of a matrix. This part of the computational process was the bottleneck in previous solutions to the problem and is now an order of magnitude faster resulting in significantly faster overall calculation of the wave function overlaps., Enrique S. Quintana-Orti was supported by project TIN2017-82972-R, Pedro Alonso-Jorda by project TIN2017-89314-P-AR, and Jose R. Herrero by project TIN2015-65316-P, all of the MINECO and FEDER. The latter was also funded by the Generalitat de Catalunya (contract 2017-SGR-1414). Marin Sapunar and Davor Davidovic were supported by Croatian Science Foundation under grant HRZZ IP-2016-06-1142.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Compression and load balancing for efficient sparse matrix-vector product on multicore processors and graphics processing units

  • Aliaga, José I.
  • Anzt, Hartwig
  • Grützmacher, Thomas
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
  • Tomás Domínguez, Andrés Enrique
[EN] We contribute to the optimization of the sparse matrix-vector product by introducing a variant of the coordinate sparse matrix format that balances the workload distribution and compresses both the indexing arrays and the numerical information. Our approach is multi-platform, in the sense that the realizations for (general-purpose) multicore processors as well as graphics accelerators (GPUs) are built upon common principles, but differ in the implementation details, which are adapted to avoid thread divergence in the GPU case or maximize compression element-wise (i.e., for each matrix entry) for multicore architectures. Our evaluation on the two last generations of NVIDIA GPUs as well as Intel and AMD processors demonstrate the benefits of the new kernels when compared with the optimized implementations of the sparse matrix-vector product in NVIDIA's cuSPARSE and Intel's MKL, respectively., J. I. Aliaga, E. S. Quintana-Ortí, and A. E. Tomás were supported by TIN2017-82972-R of the Spanish MINECO. H. Anzt and T. Grützmacher were supported by the Impuls und Vernetzungsfond of the Helmholtz Association under grant VH-NG-1241 and by the Exascale Computing Project (17-SC-20-SC), a collaborative effort of the U.S. Department of Energy Office of Science and the National Nuclear Security Administration. The authors would like to thank the Steinbuch Centre for Computing (SCC) of the Karlsruhe Institute of Technology for providing access to an NVIDIA A100 GPU.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Analyzing the impact of the MPI allreduce in distributed training of convolutional neural networks

  • Castelló, Adrián|||0000-0002-8576-8451
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
  • Catalán, Mar
  • Dolz, Manuel F.
  • Duato, José
[EN] For many distributed applications, data communication poses an important bottleneck from the points of view of performance and energy consumption. As more cores are integrated per node, in general the global performance of the system increases yet eventually becomes limited by the interconnection network. This is the case for distributed data-parallel training of convolutional neural networks (CNNs), which usually proceeds on a cluster with a small to moderate number of nodes. In this paper, we analyze the performance of the Allreduce collective communication primitive, a key to the efficient data-parallel distributed training of CNNs. Our study targets the distinct realizations of this primitive in three high performance instances of Message Passing Interface (MPI), namely MPICH, OpenMPI, and IntelMPI, and employs a cluster equipped with state-of-the-art processor and network technologies. In addition, we apply the insights gained from the experimental analysis to the optimization of the TensorFlow framework when running on top of Horovod. Our study reveals that a careful selection of the most convenient MPI library and Allreduce (ARD) realization accelerates the training throughput by a factor of 1.2x compared with the default algorithm in the same MPI library, and up to 2.8x when comparing distinct MPI libraries in a number of relevant combinations of CNN model+dataset., Project TIN2017-82972-R of the Spanish Ministerio de Ciencia, Innovacion y Universidades. Agencia Valenciana de la Innovacion. This research was partially sponsored by projects TIN2017-82972-R of Ministerio de Ciencia, Innovación y Universidades and PROMETEO/2019/109 of the Generalitat Valenciana. Adrián Castelló was supported by the Juan de la Cierva-Formación project FJC2019-039222-I of the Ministerio de Ciencia, Innovación y Universidades. Manuel F. Dolz was also supported by the Plan GenT project CDEIGENT/2018/014 of the Generalitat Valenciana.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Adaptive Precision Block-Jacobi for High Performance Preconditioning in the Ginkgo Linear Algebra Software

  • Flegar, Goran
  • Anzt, Hartwig
  • Cojean, Terry
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] The use of mixed precision in numerical algorithms is a promising strategy for accelerating scientific applications. In particular, the adoption of specialized hardware and data formats for low-precision arithmetic in high-end GPUs (graphics processing units) has motivated numerous efforts aiming at carefully reducing the working precision in order to speed up the computations. For algorithms whose performance is bound by the memory bandwidth, the idea of compressing its data before (and after) memory accesses has received considerable attention. One idea is to store an approximate operator-like a preconditioner-in lower than working precision hopefully without impacting the algorithm output. We realize the first high-performance implementation of an adaptive precision block-Jacobi preconditioner which selects the precision format used to store the preconditioner data on-the-fly, taking into account the numerical properties of the individual preconditioner blocks. We implement the adaptive block-Jacobi preconditioner as production-ready functionality in the Ginkgo linear algebra library, considering not only the precision formats that are part of the IEEE standard, but also customized formats which optimize the length of the exponent and significand to the characteristics of the preconditioner blocks. Experiments run on a state-of-the-art GPU accelerator show that our implementation offers attractive runtime savings., H. Anzt and T. Cojean were supported by the "Impuls und Vernetzungsfond of the Helmholtz Association" under grant VH-NG-1241. G. Flegar and E. S. Quintana-Orti were supported by project TIN2017-82972-R of the MINECO and FEDER and the H2020 EU FETHPC Project 732631 "OPRECOMP". This research was supported by the Exascale Computing Project (17-SC-20-SC), a collaborative effort of the U.S. Department of Energy Office of Science and the National Nuclear Security Administration. The authors want to acknowledge the access to the Piz Daint supercomputer at the Swiss National Supercomputing Centre (CSCS) granted under the project #d100 and the Summit supercomputer at the Oak Ridge National Lab (ORNL).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

A case for malleable thread-level linear algebra libraries: The LU factorization with partial pivoting

  • Catalán, Sandra
  • Herrero, José R.
  • Rodríguez-Sánchez, Rafael
  • van de Geijn, Robert
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] We propose two novel techniques for overcoming load-imbalance encountered when implementing so-called look-ahead mechanisms in relevant dense matrix factorizations for the solution of linear systems. Both techniques target the scenario where two thread teams are created/activated during the factorization, with each team in charge of performing an independent task/branch of execution. The first technique promotes worker sharing (WS) between the two tasks, allowing the threads of the task that completes first to be reallocated for use by the costlier task. The second technique allows a fast task to alert the slower task of completion, enforcing the early termination (ET) of the second task, and a smooth transition of the factorization procedure into the next iteration. The two mechanisms are instantiated via a new malleable thread-level implementation of the basic linear algebra subprograms, and their benefits are illustrated via an implementation of the LU factorization with partial pivoting enhanced with look-ahead. Concretely, our experimental results on an Intel-Xeon system with 12 cores show the benefits of combining WS+ET, reporting competitive performance in comparison with a task-parallel runtime-based solution., This work was supported in part by the Spanish Ministerio de Economia y Competitividad under Project TIN2014-53495-R, Project TIN2015-65316-P, and Project TIN2017-82972-R, in part by the H2020 EU FETHPC "INTERTWinE" under Project 671602, in part by the Generalitat de Catalunya under Project 2017-SGR-1414, and in part by the NSF under Grant ACI-1550493.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Reproducibility of parallel preconditioned conjugate gradient in hybrid programming environments

  • Iakymchuk, Roman
  • Barreda Vayá, Maria
  • Graillat, Stef
  • Aliaga, José I.
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] The Preconditioned Conjugate Gradient method is often employed for the solution of linear systems of equations arising in numerical simulations of physical phenomena. While being widely used, the solver is also known for its lack of accuracy while computing the residual. In this article, we propose two algorithmic solutions that originate from the ExBLAS project to enhance the accuracy of the solver as well as to ensure its reproducibility in a hybrid MPI + OpenMP tasks programming environment. One is based on ExBLAS and preserves every bit of information until the final rounding, while the other relies upon floating-point expansions and, hence, expands the intermediate precision. Instead of converting the entire solver into its ExBLAS-related implementation, we identify those parts that violate reproducibility/non-associativity, secure them, and combine this with the sequential executions. These algorithmic strategies are reinforced with programmability suggestions to assure deterministic executions. Finally, we verify these approaches on two modern HPC systems: both versions deliver reproducible number of iterations, residuals, direct errors, and vector-solutions for the overhead of less than 37.7% on 768 cores., The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was partially supported by the European Union's Horizon 2020 research, innovation program under the Marie Sklodowska-Curie grant agreement via the Robust project No. 842528 as well as the Project HPC-EUROPA3 (INFRAIA-2016-1-730897), with the support of the H2020 EC RIA Programme; in particular, the author gratefully acknowledges the support of Vicenc comma Beltran and the computer resources and technical support provided by BSC. The researchers from Universitat Jaume I (UJI) and Universitat Polit ' ecnica de Valencia (UPV) were supported by MINECO project TIN2017-82972-R. Maria Barreda was also supported by the POSDOC-A/2017/11 project from the Universitat Jaume I.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

FaST-LMM for two-way Epistasis tests on high performance clusters

  • Martínez, Héctor
  • Barrachina, Sergio
  • Castillo, Maribel
  • Rambla de Argila, Jordi
  • Farre, Xavier
  • Navarro, Arcadi
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] We introduce a version of the epistasis test in FaST-LMM for clusters of multithreaded processors. This new software maintains the sensitivity of the original FaST-LMM while delivering acceleration that is close to linear on 12-16 nodes of two recent platforms, with respect to improved implementation of FaST-LMM presented in an earlier work. This efficiency is attained through several enhancements on the original single-node version of FaST-LMM, together with the development of a message passing interface (MPI)-based version that ensures a balanced distribution of the workload as well as a multigraphics processing unit (GPU) module that can exploit the presence of multiple GPUs per node., The researchers from the Universitat Jaume I were supported by projects TIN2014-53495-R and TIN2017-82972-R of the MINECO and FEDER.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Accelerating distributed deep neural network training with pipelined MPI allreduce

  • Castelló, Adrián|||0000-0002-8576-8451
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
  • Duato Marín, José Francisco
[EN] TensorFlow (TF) is usually combined with the Horovod (HVD) workload distribution package to obtain a parallel tool to train deep neural network on clusters of computers. HVD in turn utilizes a blocking Allreduce primitive to share information among processes, combined with a communication thread to overlap communication with computation. In this work, we perform a thorough experimental analysis to expose (1) the importance of selecting the best algorithm in MPI libraries to realize the Allreduce operation; and (2) the performance acceleration that can be attained when replacing a blocking Allreduce with its non-blocking counterpart (while maintaining the blocking behaviour via the appropriate synchronization mechanism). Furthermore, (3) we explore the benefits of applying pipelining to the communication exchange, demonstrating that these improvements carry over to distributed training via TF+HVD. Finally, (4) we show that pipelining can also boost performance for applications that make heavy use of other collectives, such as Broadcast and Reduce-Scatter., Project TIN2017-82972-R of the Spanish Ministerio de Ciencia, Innovacion y Universidades. Agencia Valenciana de la Innovacion. Juan de la Cierva-Formacion project FJC2019-039222-I of the Ministerio de Ciencia, Innovacion y Universidades PRACE Preparatory Access project #2010PA5531.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Reproducibility strategies for parallel preconditioned Conjugate Gradient

  • Iakymchuk, Roman
  • Barreda, María
  • Wiesenberger, Matthias
  • Aliaga, José I.
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] The Preconditioned Conjugate Gradient method is often used in numerical simulations. While being widely used, the solver is also known for its lack of accuracy while computing the residual. In this article, we aim at a twofold goal: enhance the accuracy of the solver but also ensure its reproducibility in a message-passing implementation. We design and employ various strategies starting from the ExBLAS approach (through preserving every bit of information until final rounding) to its more lightweight performance-oriented variant (through expanding the intermediate precision). These algorithmic strategies are reinforced with programmability suggestions to assure deterministic executions. Finally, we verify these strategies on modern HPC systems: both versions deliver reproducible number of iterations, residuals, direct errors, and vector-solutions for the overhead of only 29% (ExBLAS) and 4% (lightweight) on 768 processes., To begin with, we would like to thank the reviewers for their thorough reading of the article as well as their valuable comments and suggestions. This research was partially supported by the European Union's Horizon 2020 research, innovation programme under the Marie Sklodowska-Curie grant agreement via the Robust project No. 842528 as well as the Project HPC-EUROPA3 (INFRAIA-2016-1-730897), with the support of the H2020 EC RIA Programme; in particular, the author gratefully acknowledges the support of Vicenc Beltran and the computer resources and technical support provided by BSC. The researchers from Universitat Jaume I (UJI) and Universidad Politecnica de Valencia (UPV) were supported by MINECO, Spain project TIN2017-82972-R. Maria Barreda was also supported by the POSDOC-A/2017/11 project from the Universitat Jaume I, Spain.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Toward a modular precision ecosystem for high performance computing

  • Anzt, Hartwig
  • Flegar, Goran
  • Gruetzmacher, Thomas
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] With the memory bandwidth of current computer architectures being significantly slower than the (floating point) arithmetic performance, many scientific computations only leverage a fraction of the computational power in today's high-performance architectures. At the same time, memory operations are the primary energy consumer of modern architectures, heavily impacting the resource cost of large-scale applications and the battery life of mobile devices. This article tackles this mismatch between floating point arithmetic throughput and memory bandwidth by advocating a disruptive paradigm change with respect to how data are stored and processed in scientific applications. Concretely, the goal is to radically decouple the data storage format from the processing format and, ultimately, design a "modular precision ecosystem" that allows for more flexibility in terms of customized data access. For memory-bounded scientific applications, dynamically adapting the memory precision to the numerical requirements allows for attractive resource savings. In this article, we demonstrate the potential of employing a modular precision ecosystem for the block-Jacobi preconditioner and the PageRank algorithm-two applications that are popular in the communities and at the same characteristic representatives for the field of numerical linear algebra and data analytics, respectively., The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Impuls und Vernetzungsfond of the Helmholtz Association under grant VH-NG-1241. G Flegar and ES Quintana-Ortí were supported by project TIN2017-82972-R of the MINECO and FEDER and the H2020 EU FETHPC Project 732631 OPRECOMP .