Search

[Abstract] The Divide-and-conquer (D&C) pattern appears in a large number of problems and is highly suitable to exploit parallelism. This has led to much research on its easy and efficient application both in shared and distributed memory parallel systems. One of the most successful approaches explored in this area consists of expressing this pattern by means of parallel skeletons which automate and hide the complexity of the parallelization from the user while trying to provide good performance. In this paper, we tackle the development of a skeleton oriented to the efficient parallel resolution of D&C problems with a high degree of imbalance among the subproblems generated and/or a deep level of recurrence. The skeleton achieves in our experiments average speedups between 11 and 18% higher than those of other solutions, reaching a maximum speedup of 78% in some tests. Nevertheless, the new proposal requires an average of between 13 and 29% less programming effort than the usual alternatives.

[Abstract] Improving the performance of the convolution operation has become a key target for High Performance Computing (HPC) developers due to its prevalence in deep learning applied mainly to video processing. The improvement is being pushed by algorithmic and implementation innovations. Algorithmically, the convolution can be solved as it is mathematically enunciated, but other methods allow to transform it into a Fast Fourier Transform (FFT) or a GEneral Matrix Multiplication (GEMM). In this latter group, the Winograd algorithm is a state-of-the-art variant that is specially suitable for smaller convolutions. In this paper, we present openCNN, an optimized CUDA C++ implementation of the Winograd convolution algorithm. Our approach achieves speedups of up to 1.76× on Turing RTX 2080Ti and up to 1.85× on Ampere RTX 3090 with respect to Winograd convolution in cuDNN 8.2.0. OpenCNN is released as open-source software.

[Abstract]: Codes written in a naive way seldom effectively exploit the computing resources, while writing optimized codes is usually a complex task that requires certain levels of expertise. This problem is further increased in the presence of heterogeneous devices, which present more tunable parameters than regular CPUs and high sensitivity to the optimization decisions taken. Furthermore, portability is an added concern given the wide variety of accelerators available. This paper tackles this problem adding an automatic optimizer to a library that already provides an easy and portable way to program heterogeneous devices, the Heterogeneous Programming Library (HPL). Our optimizer takes as input a simple version of a code and then tunes it for the device where it is going to be executed by performing the most usual set of optimizations applicable in heterogeneous devices. These optimizations are parametrized using a set of optimization parameters that need to be tuned for the device. The HPL library has also been equipped with an autotuner that can be used to this purpose. The effectiveness of the autotuner and the optimizer has been tested on several codes and devices. The results show that the combination of the autotuner and the optimizer make the tested codes 16 times faster on average than the original codes written by the programmer.

[Abstract]: The Partitioned Global Address Space (PGAS) programming model is one of the most relevant proposals to improve the ability of developers to exploit distributed memory systems. However, despite its important advantages with respect to the traditional message-passing paradigm, PGAS has not been yet widely adopted. We think that PGAS libraries are more promising than languages because they avoid the requirement to (re)write the applications using them, with the implied uncertainties related to portability and interoperability with the vast amount of APIs and libraries that exist for widespread languages. Nevertheless, the need to embed these libraries within a host language can limit their expressiveness and very useful features can be missing. This paper contributes to the advance of PGAS by enabling the simple development of arbitrarily complex task-parallel codes following a dataflow approach on top of the PGAS UPC++ library, implemented in C++. In addition, our proposal, called UPC++ DepSpawn, relies on an optimized multithreaded runtime that provides very competitive performance, as our experimental evaluation shows.

[Abstract]: Many problems of industrial and scientific interest require the solving of tridiagonal linear systems. This paper presents several implementations for the parallel solving of large tridiagonal systems on multi-core architectures, using the OmpSs programming model. The strategy used for the parallelization is based on the combination of two different existing algorithms, PCR and Thomas. The Thomas algorithm, which cannot be parallelized, requires the fewest number of floating point operations. The PCR algorithm is the most popular parallel method, but it is more computationally expensive than Thomas. The method proposed in this paper starts applying the PCR algorithm to break down one large tridiagonal system into a set of smaller and independent ones. In a second step, these independent systems are concurrently solved using Thomas. The paper also contains an analytical study of which is the best point to switch from PCR to Thomas. Also, the paper addresses the main performance issues of combining PCR and Thomas proposing a set of alternative implementations, some of them even imply algorithmic changes. The performance evaluation shows that the best implementation achieves a peak speedup of 4 with respect to the Intel MKL counterpart routine and 2.5 with respect to a single-threaded Thomas.

[Abstract]: The existence of a wide variety of computing devices with very different properties makes essential the development of software that is not only portable among them, but which also adapts to the properties of each platform. In this paper, we present the Heterogeneous Butterfly Processing Library (HBPL), which provides optimized portable kernels for problems of small sizes that allow using orthogonal transform algorithms such as the FFT and DCT on different accelerators and regular CPUs. Our library is implemented on the OpenCL standard, which provides portability on a large number of platforms. Furthermore, high performance is achieved on a wide range of devices by exploiting run-time code generation and metaprogramming guided by a parametrization strategy. An exhaustive evaluation on different platforms shows that our proposal obtains competitive or better performance than related libraries.

Weed scientists are usually interested in the study of the distribution and density functions of the random variable that relates weed emergence with environmental indices like the hydrothermal time (HTT). However, in many situations, experimental data are presented in a grouped way and, therefore, the standard nonparametric kernel estimators cannot be computed. Kernel estimators for the density and distribution functions for interval‐grouped data, as well as bootstrap confidence bands for these functions, have been proposed and implemented in the binnednp package. Analysis with different treatments can also be performed using a bootstrap approach and a Cramér‐von Mises type distance. Several bandwidth selection procedures were also implemented. This package also allows to estimate different emergence indices that measure the shape of the data distribution. The values of these indices are useful for the selection of the soil depth at which HTT should be measured which, in turn, would maximize the predictive power of the proposed methods. This paper presents the functions of the package and provides an example using an emergence data set of Avena sterilis (wild oat). The binnednp package provides investigators with a unique set of tools allowing the weed science research community to analyze interval‐grouped data.

[Abstract] Weed scientists are usually interested in the study of the distribution and density functions of the random variable that relates weed emergence with environmental indices like the hydrothermal time (HTT). However, in many situations, experimental data are presented in a grouped way and, therefore, the standard nonparametric kernel estimators cannot be computed. Kernel estimators for the density and distribution functions for interval‐grouped data, as well as bootstrap confidence bands for these functions, have been proposed and implemented in the binnednp package. Analysis with different treatments can also be performed using a bootstrap approach and a Cramér‐von Mises type distance. Several bandwidth selection procedures were also implemented. This package also allows to estimate different emergence indices that measure the shape of the data distribution. The values of these indices are useful for the selection of the soil depth at which HTT should be measured which, in turn, would maximize the predictive power of the proposed methods. This paper presents the functions of the package and provides an example using an emergence data set of Avena sterilis (wild oat). The binnednp package provides investigators with a unique set of tools allowing the weed science research community to analyze interval‐grouped data.

[Abstract]: The rise of heterogeneous systems has given place to great challenges for users as they involve new concepts, restrictions, and frameworks. Their exploitation is further complicated in the context of distributed memory systems, which require the usage of additional different programming paradigms and tools. In this paper, we propose a novel approach to program heterogeneous clusters that is based on high-level abstractions such as tiles and hierarchical decomposition combined with the powerful APIs that data types and embedded languages can provide in languages such as C++. Rather than building our proposal from scratch, we have implemented it as a natural integration of the existing Hierarchically Tiled Arrays (HTA) and Heterogeneous Programming Library (HPL) projects, ie, the first one being focused on distributed computing and the second one on heterogeneous processing. The result, called Heterogeneous Hierarchically Tiled Arrays (H2TA), is very intuitive and easy to use thanks to the global view of the data and the single-threaded view of the execution that it provides at cluster level together with the transparency it provides with respect to the management of the heterogeneous devices. An evaluation comparing our proposal with MPI-based implementations shows its large programmability advantages and the reasonable overhead incurred.

[Abstract] Heterogeneous devices require much more work from programmers than traditional CPUs, particularly when there are several of them, as each one has its own memory space. Multidevice applications require to distribute kernel executions and, even worse, arrays portions that must be kept coherent among the di_erent device memories and the host memory. In addition, when devices with di_erent characteristics participate in a computation, optimally distributing the work among them is not trivial. In this paper we extend an existing framework for the programming of accelerators called Heterogeneous Programming Library (HPL) with three kinds of improvements that facilitate these tasks. The _rst two ones are the ability to de_ne subarrays and subkernels, which distribute kernels on di_erent devices. The last one is a convenient extension of the subkernel mechanism to distribute computations among heterogeneous devices seeking the best work balance among them. This last contribution includes two analytical models that have proved to automatically provide very good work distributions. Our experiments also show the large programmability advantages of our approach and the negligible overhead incurred.

[Abstract] Extreme Data is an incarnation of Big Data concept distinguished by the massive amounts of data that must be queried, communicated and analyzed in near real-time by using a very large number of memory or storage elements and exascale computing systems. Immediate examples are the scientific data produced at a rate of hundreds of gigabits-per-second that must be stored, filtered and analyzed, the millions of images per day that must be analyzed in parallel, the one billion of social data posts queried in real-time on an in-memory components database. Traditional disks or commercial storage nowadays cannot handle the extreme scale of such application data. Following the need of improvement of current concepts and technologies, we focus in this paper on the needs of data intensive applications running on systems composed of up to millions of computing elements (exascale systems). We propose in this paper a methodology to advance the state-of-the-art. The starting point is the definition of new programming paradigms, APIs, runtime tools and methodologies for expressing data-intensive tasks on exascale systems. This will pave the way for the exploitation of massive parallelism over a simplified model of the system architecture, thus promoting high performance and efficiency, offering powerful operations and mechanisms for processing extreme data sources at high speed and/or real time.

[Abstract] Estimating parameters of a statistical fisheries assessment model typically involves a comparison of disparate datasets to a forward simulation model through a likelihood function. In all but trivial cases the estimations of these models tend to be time-consuming due to issues related to multi-modality and non-linearity. This paper develops novel parallel implementations of popular search algorithms, applicable to expensive function calls typically encountered in fisheries stock assessment. It proposes two versions of both Simulated Annealing and Hooke & Jeeves optimization algorithms with the aim of fully utilizing the processing power of common multicore systems. The proposals have been tested on a 24-core server using three different input models. Results indicate that the parallel versions are able to take advantage of available resources without sacrificing the quality of the solution.

[Abstract] The usage of heterogeneous devices presents two main problems. One is their complex programming, a problem that grows when multiple devices are used. The second issue is that even if the codes for these devices can be portable on top of OpenCL, they lack performance portability, effectively requiring specialized implementations for each device to get good performance. In this paper we extend the Heterogeneous Programming Library (HPL), which improves the usability of heterogeneous systems on top of OpenCL, to better handle both issues. First, we provide HPL with mechanisms to support the implementation of any multi-device application that requires arbitrary patterns of communication between several devices and a host memory. In a second stage HPL is improved with an adaptive scheme to optimize communications between devices depending on the execution environment. An evaluation using benchmarks with very different nature shows that HPL reduces the SLOCs and programming effort of OpenCL applications by 27 and 43 %, respectively, while improving the performance of applications that exchange data between devices by 28 % on average.

[Abstract]: In this work, several parallel implementations of a numerical model of pollutant transport on a shallow water system are presented. These parallel implementations are developed in two phases. First, the sequential code is rewritten to exploit the stream programming model. And second, the streamed code is targeted for current multi-threaded systems, in particular, multi-core CPUs and modern GPUs. The performance is evaluated on a multi-core CPU using OpenMP, and on a GPU using the streaming-oriented programming language Brook+, as well as the standard language for heterogeneous systems, OpenCL.

[Abstract] While recognition of the advantages of heterogeneous computing is steadily growing, the issues of programmability and portability hinder its exploitation. The introduction of the OpenCL standard was a major step forward in that it provides code portability, but its interface is even more complex than that of other approaches. In this paper, we present the Heterogeneous Programming Library (HPL), which permits the development of heterogeneous applications addressing both portability and programmability while not sacrificing high performance. This is achieved by means of an embedded language and data types provided by the library with which generic computations to be run in heterogeneous devices can be expressed. A comparison in terms of programmability and performance with OpenCL shows that both approaches offer very similar performance, while outlining the programmability advantages of HPL.

[Abstract] Servet is a suite of benchmarks focused on detecting a set of parameters with high influence on the overall performance of multicore systems. These parameters can be used for autotuning codes to increase their performance on multicore clusters. Although Servet has been proved to detect accurately cache hierarchies, bandwidths and bottlenecks in memory accesses, as well as the communication overhead among cores, up to now the impact of the use of this information on application performance optimization has not been assessed. This paper presents a novel algorithm that automatically uses Servet for mapping parallel applications on multicore systems and analyzes its impact on three testbeds using three different parallel programming models: message-passing, shared memory and partitioned global address space (PGAS). Our results show that a suitable mapping policy based on the data provided by this tool can significantly improve the performance of parallel applications without source code modification.

[Abstract] MapReduce is a powerful tool for processing large data sets used by many applications running in distributed environments. However, despite the increasing number of computationally intensive problems that require low-latency communications, the adoption of MapReduce in High Performance Computing (HPC) is still emerging. Here languages based on the Partitioned Global Address Space (PGAS) programming model have shown to be a good choice for implementing parallel applications, in order to take advantage of the increasing number of cores per node and the programmability benefits achieved by their global memory view, such as the transparent access to remote data. This paper presents the first PGAS-based MapReduce implementation that uses the Unified Parallel C (UPC) language, which (1) obtains programmability benefits in parallel programming, (2) offers advanced configuration options to define a customized load distribution for different codes, and (3) overcomes performance penalties and bottlenecks that have traditionally prevented the deployment of MapReduce applications in HPC. The performance evaluation of representative applications on shared and distributed memory environments assesses the scalability of the presented MapReduce framework, confirming its suitability.