Beau Johnston, Brian Lee, Luke Angove, Alistair P. Rendell
{"title":"Embedded Accelerators for Scientific High-Performance Computing: An Energy Study of OpenCL Gaussian Elimination Workloads","authors":"Beau Johnston, Brian Lee, Luke Angove, Alistair P. Rendell","doi":"10.1109/ICPPW.2017.22","DOIUrl":"https://doi.org/10.1109/ICPPW.2017.22","url":null,"abstract":"Energy efficient High-Performance Computing (HPC) is becoming increasingly important. Recent ventures into this space have introduced an unlikely candidate to achieve exascale scientific computing hardware with a small energy footprint. ARM processors and embedded GPU accelerators originally developed for energy efficiency in mobile devices, where battery life is critical, are being repurposed and deployed in the next generation of supercomputers. Unfortunately, the performance of executing scientific workloads on many of these devices is largely unknown, yet the bulk of computation required in high-performance supercomputers is scientific. We present an analysis of one such scientific code, in the form of Gaussian Elimination, and evaluate both execution time and energy used on a range of embedded accelerator SoCs. These include three ARM CPUs and two mobile GPUs. Understanding how these low power devices perform on scientific workloads will be critical in the selection of appropriate hardware for these supercomputers, for how can we estimate the performance of tens of thousands of these chips if the performance of one is largely unknown?","PeriodicalId":353142,"journal":{"name":"2017 46th International Conference on Parallel Processing Workshops (ICPPW)","volume":"27 2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123552659","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Efficient Broadcasting Algorithm in Harary-like Networks","authors":"Puspal Bhabak, Hovhannes A. Harutyunyan, P. Kropf","doi":"10.1109/ICPPW.2017.34","DOIUrl":"https://doi.org/10.1109/ICPPW.2017.34","url":null,"abstract":"In this paper, we analyze the properties of Harary graphs and some derivatives with respect to the achievable performance of communication within network structures based on these graphs. In particular we defined Cordal-Haray graphs on n nodes which can be constructed for any even n for any odd degree between 3 and 2[log n] – 1. We also present a simple algorithm for fast message broadcasting in this network. Our analysis show that when nodes of a Cordal-Harary Graph have logarithmic degree then the broadcasting time will be as small as [log n] which is the minimum possible value for a network on n nodes. All this properties show that Cordal-Harary is a very good network architecture for parallel processing.","PeriodicalId":353142,"journal":{"name":"2017 46th International Conference on Parallel Processing Workshops (ICPPW)","volume":"40 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123437627","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Communication-Computation Overlapping with Dynamic Loop Scheduling for Preconditioned Parallel Iterative Solvers on Multicore and Manycore Clusters","authors":"K. Nakajima, T. Hanawa","doi":"10.1109/ICPPW.2017.39","DOIUrl":"https://doi.org/10.1109/ICPPW.2017.39","url":null,"abstract":"Preconditioned parallel solvers based on the Krylov iterative method are widely used in scientific and engineering applications. Communication overhead is a critical issue when executing these solvers on large-scale massively parallel supercomputers. In this work, we introduced communication-computation (CC) overlapping with dynamic loop scheduling of OpenMP to the sparse matrix-vector multiplication (SpMV) process of a parallel iterative solver. We then used the solver to evaluate the performance of a parallel finite element application (GeoFEM/Cube) on multicore and manycore clusters. The dynamic loop scheduling of OpenMP improved the efficiency of CC overlapping in halo exchanges, and the developed method attained a significant performance improvement of 40-50% for parallel iterative solvers in strong scaling using up to 16,384 cores of a Fujitsu PRIMEHPC FX10 supercomputer and an Intel Xeon Phi (KNL) cluster. Finally, the developed method was applied to GeoFEM/Cube using a parallel BiCGSTAB solver with sparse approximate inverse (SAI) preconditioning, and a 15-20% performance improvement was obtained on 12,288 cores of the Fujitsu FX10 and the KNL cluster.","PeriodicalId":353142,"journal":{"name":"2017 46th International Conference on Parallel Processing Workshops (ICPPW)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125292436","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yuko Hirabe, Manato Fujimoto, Yutaka Arakawa, H. Suwa, K. Yasumoto
{"title":"Effect on Group Detection Based on Human Proximity for Human Relationship Extraction in Daily Life","authors":"Yuko Hirabe, Manato Fujimoto, Yutaka Arakawa, H. Suwa, K. Yasumoto","doi":"10.1109/ICPPW.2017.14","DOIUrl":"https://doi.org/10.1109/ICPPW.2017.14","url":null,"abstract":"In this paper, we investigate an effect on group detection based on human proximity depending on RSSI values, measured by BLE signals. To realize the investigation, we present a novel method for detecting groups. In the method, we developed a smartphone application for measurement of RSSI values of BLE, and determine a threshold of RSSI values which selected experimentally by making the propagation characteristics of BLE signal. To show the effectiveness of proposed method, we evaluated the method through a series of experiments and investigated the effect on group detection by the difference of human proximity. As the results, we found that the group detection was affected by the difference of human proximity.","PeriodicalId":353142,"journal":{"name":"2017 46th International Conference on Parallel Processing Workshops (ICPPW)","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126172185","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"An Improved Abstract GPU Model with Data Transfer","authors":"T. Carroll, Prudence W. H. Wong","doi":"10.1109/ICPPW.2017.28","DOIUrl":"https://doi.org/10.1109/ICPPW.2017.28","url":null,"abstract":"GPUs are commonly used as coprocessors to accelerate a compute-intensive task, thanks to their massively parallel architecture. There is study into different abstract parallel models, which allow researchers to design and analyse parallel algorithms. However, most work on analysing GPU algorithms has been software based tools for profiling a GPU algorithm. Recently, some abstract GPU models have been proposed, yet they do not capture all elements of a GPU. In particular, they miss the data transfer between CPU and GPU, which in practice can cause a bottleneck and reduce performance dramatically. We propose a comprehensive model called Abstract Transferring GPU which to our knowledge is the first abstract GPU model to capture data transfer between CPU and GPU. We show via experiments, that existing abstract GPU models cannot sufficiently capture all of the actual running of a GPU algorithm time in all cases, as they do not capture data transfer. We show that by capturing data transfer with our model, we are able to obtain more accurate predictions of the GPU algorithm actual running time. It is expected that our model helps improve design and analysis of heterogeneous systems consisting of CPU and GPU, and will allow researchers to make better informed implementation decisions, as they will be aware how data transfer will affect their programs.","PeriodicalId":353142,"journal":{"name":"2017 46th International Conference on Parallel Processing Workshops (ICPPW)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131878499","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Heterogeneous Hardware Support in BEAGLE, a High-Performance Computing Library for Statistical Phylogenetics","authors":"Daniel L. Ayres, M. P. Cummings","doi":"10.1109/ICPPW.2017.17","DOIUrl":"https://doi.org/10.1109/ICPPW.2017.17","url":null,"abstract":"We describe our approach to extend the BEAGLE library for high-performance statistical phylogenetic inference (maximum likelihood estimation and Bayesian analysis) in order to support a wider range of modern accelerators and multicore CPUs, and present the corresponding performance results from these platforms. Our solution includes a shared code design providing a uniform interface for a variety of compute platforms available under CUDA and OpenCL parallel computing frameworks. We have also implemented CPU threading in BEAGLE, and in sum these improvements allow the library to exploit a wide-range of hardware parallelism including CPU and Xeon Phi, vectorization intrinsics (e.g., SSE, AVX), and GPUs. Although code reuse and maintainability are features of our design, our approach also includes hardware specific optimizations for performance critical portions of the code. Extending the BEAGLE library in this manner allows a greater variety of users to exploit the hardware resources available to them. As an example of increase in performance, using BEAGLE on a system with two Intel Xeon E5-2680v4 CPUs we observe a 39-fold speedup for a MrBayes 3.2.6 codon-model analysis, compared to the native MrBayes MPI-SSE implementation. The general design features of the library also provide a model for software development using parallel computing frameworks that is applicable to other domains.","PeriodicalId":353142,"journal":{"name":"2017 46th International Conference on Parallel Processing Workshops (ICPPW)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130501313","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Manuel Alejandro Betancourt-Odio, Angelina Lazaro-Alquezar, A. Colino
{"title":"Scheduling Optimization in Ophthalmology Using Multi-Objective Integer Models","authors":"Manuel Alejandro Betancourt-Odio, Angelina Lazaro-Alquezar, A. Colino","doi":"10.1109/ICPPW.2017.48","DOIUrl":"https://doi.org/10.1109/ICPPW.2017.48","url":null,"abstract":"This paper studies a scheduling problem in which patients request appointments at specific future days within a specialty definite time window. This research is inspired by a study of Ophthalmology scheduling practices at the Clinic Hospital at Santiago de Cuba (Cuba). In this hospital, patients do not know in advance neither the proximate time at which they will be seen by the doctor nor the total amount of time to be spent at the Hospital. In this article, an optimal distribution of patients during the designing scheduling process has been performed. The optimality condition obeys to two different goals: to give the appointments as soon as possible and to minimize the time that a patient would spend at the hospital to complete a protocol. We formulate this problem as a Multi-Objective Integer Problem (MOIP) and compare the performance of the resulting MOIP policies with traditional practices decision rules for the diagnosis and treatment of ophthalmology diseases. We show that this method outperforms traditional methods by far. Specifically, it reduces the total diagnosis time for cataract, cornea and glaucoma diseases by 70% on average or, in other words, by roughly two hours, relative to the standard approach. Arguably, this translates into improved patient satisfaction and efficiency in the use of resources in health services.","PeriodicalId":353142,"journal":{"name":"2017 46th International Conference on Parallel Processing Workshops (ICPPW)","volume":"37 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114713339","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"OpenCL 2.0 Compiler Adaptation on LLVM for PTX Simulators","authors":"Chun-Chieh Yang, Shao-Chung Wang, Min-Yih Hsu, Yuan-Ming Chang, Yuan-Shin Hwang, Jenq-Kuen Lee","doi":"10.1109/ICPPW.2017.21","DOIUrl":"https://doi.org/10.1109/ICPPW.2017.21","url":null,"abstract":"OpenCL continues to gather momentum on both desktop and mobile devices. The new features of OpenCL 2.0 provides developers better expressive power in programming heterogeneous computing environments. Currently in the experimental simulation environment, gem5-gpu only supports CUDA, but GPGPU-Sim can support OpenCL by compiling OpenCL kernel code to PTX using real GPU driver. However, this driver compilation in GPGPU-Sim only can support up to OpenCL 1.2. To support OpenCL 2.0, it is necessary to extend the compiler to enable the compilation of OpenCL 2.0 kernel code to PTX. In this paper, our experience in enabling the compiler flow is reported. In OpenCL 2.0, it provides new features such as dynamic parallelism, work-group built-in functions, extend atomic builtin functions, and so on. The proposed compiler that is modified from Low Level Virtual Machine (LLVM) extends such features for enhancing the emulator to support OpenCL 2.0. After the compiler is modified, it can support dynamic parallelism, workgroup built-in functions and extend atomic built-in functions. Using existing dynamic parallelism APIs in CUDA to implement OpenCL 2.0 enqueue kernel and revise compilation scheme in clang. Furthermore, the proposed compiler also creates local buffers for each work group to use for work-group builtin functions, and adds atomic built-in functions with memory order and memory scope for OpenCL 2.0 in NVPTX. From benchmarks, the proposed compiler can support the claim target.","PeriodicalId":353142,"journal":{"name":"2017 46th International Conference on Parallel Processing Workshops (ICPPW)","volume":"56 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115735834","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Klaus Kofler, J. Durillo, P. Gschwandtner, T. Fahringer
{"title":"A Region-Aware Multi-Objective Auto-Tuner for Parallel Programs","authors":"Klaus Kofler, J. Durillo, P. Gschwandtner, T. Fahringer","doi":"10.1109/ICPPW.2017.37","DOIUrl":"https://doi.org/10.1109/ICPPW.2017.37","url":null,"abstract":"Auto-tuning has become increasingly popular for optimizing non-functional parameters of parallel programs. The typically large search space requires sophisticated techniques to find well performing parameter values in a reasonable amount of time. Different parts of a program often perform best with different parameter values. We therefore subdivide programs into several regions, and try to optimize the parameter values for each of those regions separately as opposed to setting the parameter values globally for the entire program. As this enlarges the search space even further, we have to extend existing auto-tuning techniques in order to obtain good results. In this paper we introduce a novelenhancement to the RS-GDE3 algorithm which is used to explore the search space for auto-tuning programs with multiple regions regarding several objectives. We have implemented our auto-tuner using the Insieme compiler and runtime system. In comparison to a non-optimized parallel version of the tested programs, our novel approach achieves up to 7.6, 10.5, and 61.6 fold improvements for three tuned objectives wall time, energy consumption, and resource usage, respectively.","PeriodicalId":353142,"journal":{"name":"2017 46th International Conference on Parallel Processing Workshops (ICPPW)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126010081","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Deriving Highly Efficient Implementations of Parallel PageRank","authors":"B. V. Strien, K. Rietveld, H. Wijshoff","doi":"10.1109/ICPPW.2017.26","DOIUrl":"https://doi.org/10.1109/ICPPW.2017.26","url":null,"abstract":"In 1999, Page et al. described their algorithm — PageRank — for scoring pages in their web search engine, Google. Already from the start it was clear that due to the large number of pages to be ranked, the PageRank algorithm's efficiency and optimal performance was and is a critical feature. One method of increasing the efficiency of calculations is based on MapReduce, also originally published by Google. In this paper, we use the forelem framework to go from an initial specification of PageRank to a much more performant implementation.","PeriodicalId":353142,"journal":{"name":"2017 46th International Conference on Parallel Processing Workshops (ICPPW)","volume":"85 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133354276","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}