DOCSLIB.ORG

Out-Of-Order Execution Parallel Virtual Machine

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Out-Of-Order Execution Parallel Virtual Machine OVM Out-of-order Execution Parallel Virtual Machine George Bosilca, Franck Cappello Laboratoire de Recherche en Informatique Universite d’Orsay, France [email protected] Global view Evolution of parallel High performance computing architectures toward a required by a large spectre of users. multi-level (compagnies, research labs) NUMA approach Weather/Climate Clusters of SMP Physics (high energy particle) Pharmaceutics Bioinformatic Medical Images Cinema Microprocessors Simulation Etc. Programming language PC Clusters and execution models Regular Applications: Without load balancing Non-Uniform Applications: GRID, P2P With load balancing Branch and bound Dynamic generation of additional computations A central place for the execution model Applications Programming Models Ex : data parallel, concurent model, master/slave Languages Languages : Fortran, C, C++, Java, Matlab, Scilab Expression Execution Model Message Passing : MPI, OVM (RPC), Virtual Shared Memory : Scach, TreadMarks, API Global Computing Infrastructure : Netsolve, Ninf, Execution environment Computational Grids : Globus, Legion, XtremWeb GRID enabled message passing: MPICH-G, Virtual Machine Virtual shared memory for GRID: OmniOpenMP GRID, P2P Operating System Architectures Clusters of SMP Clusters Hardware Portability and virtualization Entropia, United Dev Architectures Distributed systems SETI@home Grilles locales 500 Cluster X86, Alpha, Itanium SIMD Constellation (cluster of SMP) Compaq SC Sun HPC cluster NEC SX4 / Earth 400 TOP 500 SGI O3000 / 2000 HP Hyperplex Sun HPC 10000 MPP 300 IBM SP3 Hitachi SR8000 200 CRAY T3E Fujitsu VPP 5000 100 Uni SMP HP superdome Processor NEC SX 5 0 93 94 95 96 97 98 99 00 Clusters, SMP, NUMA, Cluster de SMP,Cluster NUMA de SMP, … Cluster de SMP Cluster de Machines vectorielles SMP Clusters pseudo-vectorielles NUMA de SMPs GRID, Calcul Global, P2P Clusters de SMP Cluster de PC Proposer aux programmeurs des environnements portables permettant de pérenniser sinon leurs applications au moins leurs outils/environnements de programmation Programming Models . • RPC – Initially not really developed for // applications • OpenMP – Concurrent RPC using several threads • MPI – Data always cross the network with the RPC request • Dataflow – “global communication” like operations should be done locally + A whole range of applications could be easily expressed using RPC • Master/slave • Recursive • Some kind of branch-and-bound Programming Models .. • RPC Suitable only on shared memory architectures • OpenMP – Complex to handle irregular applications • MPI – Static environment (MPI2) • Dataflow Comp Comp Cache Penalty or Wait Charge Irregularity Unpredictable results on complex // architectures comm. comm. sync sync +Perfect for regular applications is a homogeneous environment +Have global communications Programming Models … • RPC • Most of the time the dataflow graph is statically created during the compilation • OpenMP step • MPI • Internal load-balancing • Still no “global communications” • Dataflow SMART: horizontal and vertical parallelism only for shared memory computers Programming Model • PVM – The user should do the load-balancing depending on the application – Blocking global communication – MP layer too complex and quite slow +++ Dynamic : pvm_spawn • Harness – Support multiple instance of PVM and MPI applications – More related to a execution environment Load Balancing environments Load-balancing in multi-flow environments : Athapascan (static dataflow graph + distributed resolution), lab ID, ENSIMAG PM2 (threads migration), LIFL, ENS Lyon Cilk (work stealing with a recursion heap), MIT OO with load-balancing capabilities (95, 96) : Dome (c++ et PVM), SPMD, check point and migration Charm, Charm++ (thread based) Urbana Chanpaign Load-balancing for PVM/MPI (93-94-95 ; 01) : MPVM (process migration) Oregon Graduate Institute of S & T. UPVM (threads migration) Oregon Graduate Institute of S & T. Dynamic PVM (condor checkpoint), University of Amsterdam MPI TM (task migration – core dump) Mississippi State University AMPI (threads migration using Charm++) Urbana Champaign Dynamite PVM (task migration) Choosing a programming environment • Multiple parameters equation – Computer architecture – Application type – Level of programming paradigm (performances) The ideal parallel environment ? • Dynamic : resources can join/leave the environment • Automatic load balancing (user hints) • Suitable for HPC • Suitable for different kind of applications • Same performances on all architectures OVM general principles Research goal : A virtual machine based on a dynamic macro-dataflow graph and RPC à Continuum between SPMD and master/slave approachs • Why Out-of-Order ? Broker Client Servers Dataflow User API Shared Lib. Data Application OVM RPCs Resource Management User Scheduler OVM Shared Lib. Data Based on decoupled RPC ( data and requests are separate entities) Client Simplified view of the system : Unique architecture Ø Create and initialize sets of data Ø Define affinities between data Ø Define the relationship between the data (A,B) = operation(A,B,C,D) Ø Collect the final results • Provide services using plug-ins (shared libraries) on the server side • Provide a long-time scheduling plug-in for the broker Servers • Multi-threaded approach Data – Only one server by node Services Collect. • Supporting dynamic multi-protocol communication devices • Provide generic functions through the default plug-ins • And/Or additional services through user plug-ins Broker • Correct execution • Resolve dependencies Dataflow graph construction • Resource Management – Available nodes – Available services Dependencies resolution Dataflow Flow Ready requests Resource Control Management Completed Requests Scheduler Graph destruction Different RPC modes 1) Classic RPC User Broker Servers Program RPC Load RPC Function RPC Balancing RPC RPC data Volatile Data 2) Decoupled RPC User Broker Servers Program Data RPC Function RPC RPC depende id RPCid ncies id RPC id Persistent Data RPCid Macro-dataflow Move D2 Move D1 User Program Modify D1 Servers RPC Broker D1 RPC F1 RPC id Dataflow RPC id id RPC manager Data Dependencies wait RPC RPC D2 F2 Requirement : starting a function on the servers require that all the data used by the function should be available on that server Interest : only the real data dependencies limit the parallelism SPMD Applications Purpose: execution flow as close as possible to MPI Cyclic placement of the persistant data Problem : Hide the cost of the macro-dataflow management Servers Persistent data data data data data Shared Libraries Funct 1 Funct 1 Funct 1 Funct 1 Funct 2 Funct 2 Funct 2 Funct 2 Funct 3 Funct 3 Funct 3 Funct 3 Data flow data Client + • persistent data broker • static scheduling Scheduling function • global operations Irregular Applications Purpose : limit the migration, increase the number of concurent tasks Scheduling : First Come First Serve Servers Fonct * Fonct Shared Libraries X Fonct Z Fonct Y data data data data Volatiles data Data flow data Client + • Volatile data broker Scheduling function • Dynamic scheduling • No global operations Optimizations • Classic multi-protocole RPC : • embeded data if the size < 4 Ko • or asynchronous data retrieval from the client ( >= 4 Ko ) • blocking or non-blocking mode for the client • Non-blocking global communications • each server start a new execution after his contribution to a global communication • Asynchronous global communications • the completion order of the global communications could not be the same as the starting order (independent global communications could be executed in parallel) • Task prefetch when persistent data are located on the same server – hierarchic dataflow • Task prefetch on mode RPC with volatile data, without checking for ressource dependencies to hide data transfert cost for the next task • Data cache on the servers side (for volatile data) • Both kind of RPC can be used in same time ( (a) = operation(2, b, c) ) Regular applications OVM and MPI for 3 of the kernels from the NAS NPB benchmark 2.3 : CG, FT, EP CG : frequent reduction with short messages FT : frequent all-to-all with long data EP : pure parallel kernel with no communication Cluster of PentiumPro 200 + Myrinet, Score environnement (RWCP) Optimized MPI global operations 1000 EP MPI EP OVM 100 FT MPI Time (s) FT OVM 10 CG MPI CG OVM 1 2 4 8 16 32 Non-uniform Load Application Parallel version of Aires Auger Project Simulation of very high energy particle entering the Earth atmosphere 100 Rely on the load-balancing 56,1 mechanism 26,9 Iso-energy blocs 15,4 10 7,66 3,96 36,8 Speedup 20,8 13,5 7,2 With initialization 3,72 Without initialization 1 4 8 16 32 64 Another Irregular Application Real-time raytracing with OVM and Povray Parallel algorithm •Global distribution of the scene •Dynamic work distribution •Compute each part of the final image and put them together to obtain the final image Radiosity (any light) Speedup 100 Bohn/Garmann 2000 Singh Carter 2001 Benavides OVM 10 Garmann 2000 1994 1 Master/Slave approach with 1 2 4 8 16 32 64 128 the master involved in the Number of Procs computation Conclusion • Fit well for large classes of applications • We can reach the same performances as MPI for some regular applications (even with different ratio communication/computation) • OVM allow linear speedup with irregular applications (AIRES, raytracing, radiosity) using a Master/Slave approach • Multi protocoles (PM, TCP/IP, BIP), multi architectures : OVM can be used on multiples parallel computers Fault Tolerance • How to became a fault-tolerant environment ? • The system keep trace of all requested services for the clients. • Data replication, not necessary the last version • Multiple parallel execution for the same service • Single point of failure : the broker But the most important thing is that the fault tolerance could be achieved at 100% automatically.
Load more

Recommended publications
  • A Case for High Performance Computing with Virtual Machines
    A Case for High Performance Computing with Virtual Machines Wei Huangy Jiuxing Liuz Bulent Abaliz Dhabaleswar K. Panday y Computer Science and Engineering z IBM T. J. Watson Research Center The Ohio State University 19 Skyline Drive Columbus, OH 43210 Hawthorne, NY 10532 fhuanwei, [email protected] fjl, [email protected] ABSTRACT in the 1960s [9], but are experiencing a resurgence in both Virtual machine (VM) technologies are experiencing a resur- industry and research communities. A VM environment pro- gence in both industry and research communities. VMs of- vides virtualized hardware interfaces to VMs through a Vir- fer many desirable features such as security, ease of man- tual Machine Monitor (VMM) (also called hypervisor). VM agement, OS customization, performance isolation, check- technologies allow running different guest VMs in a phys- pointing, and migration, which can be very beneficial to ical box, with each guest VM possibly running a different the performance and the manageability of high performance guest operating system. They can also provide secure and computing (HPC) applications. However, very few HPC ap- portable environments to meet the demanding requirements plications are currently running in a virtualized environment of computing resources in modern computing systems. due to the performance overhead of virtualization. Further, Recently, network interconnects such as InfiniBand [16], using VMs for HPC also introduces additional challenges Myrinet [24] and Quadrics [31] are emerging, which provide such as management and distribution of OS images. very low latency (less than 5 µs) and very high bandwidth In this paper we present a case for HPC with virtual ma- (multiple Gbps).
    [Show full text]
  • Exploring Massive Parallel Computation with GPU
    Need for parallelism Graphical Processor Units Gravitational Microlensing Modelling Exploring Massive Parallel Computation with GPU Ian Bond Massey University, Auckland, New Zealand 2011 Sagan Exoplanet Workshop Pasadena, July 25-29 2011 Ian Bond | Microlensing parallelism 1/40 Need for parallelism Graphical Processor Units Gravitational Microlensing Modelling Assumptions/Purpose You are all involved in microlensing modelling and you have (or are working on) your own code this lecture shows how to get started on getting code to run on a GPU then its over to you . Ian Bond | Microlensing parallelism 2/40 Need for parallelism Graphical Processor Units Gravitational Microlensing Modelling Outline 1 Need for parallelism 2 Graphical Processor Units 3 Gravitational Microlensing Modelling Ian Bond | Microlensing parallelism 3/40 Need for parallelism Graphical Processor Units Gravitational Microlensing Modelling Paralel Computing Parallel Computing is use of multiple computers, or computers with multiple internal processors, to solve a problem at a greater computational speed than using a single computer (Wilkinson 2002). How does one achieve parallelism? Ian Bond | Microlensing parallelism 4/40 Need for parallelism Graphical Processor Units Gravitational Microlensing Modelling Grand Challenge Problems A grand challenge problem is one that cannot be solved in a reasonable amount of time with todays computers’ Examples: – Modelling large DNA structures – Global weather forecasting – N body problem (N very large) – brain simulation Has microlensing
    [Show full text]
  • Dcuda: Hardware Supported Overlap of Computation and Communication
    dCUDA: Hardware Supported Overlap of Computation and Communication Tobias Gysi Jeremia Bar¨ Torsten Hoefler Department of Computer Science Department of Computer Science Department of Computer Science ETH Zurich ETH Zurich ETH Zurich [email protected] [email protected] [email protected] Abstract—Over the last decade, CUDA and the underlying utilization of the costly compute and network hardware. To GPU hardware architecture have continuously gained popularity mitigate this problem, application developers can implement in various high-performance computing application domains manual overlap of computation and communication [23], [27]. such as climate modeling, computational chemistry, or machine learning. Despite this popularity, we lack a single coherent In particular, there exist various approaches [13], [22] to programming model for GPU clusters. We therefore introduce overlap the communication with the computation on an inner the dCUDA programming model, which implements device- domain that has no inter-node data dependencies. However, side remote memory access with target notification. To hide these code transformations significantly increase code com- instruction pipeline latencies, CUDA programs over-decompose plexity which results in reduced real-world applicability. the problem and over-subscribe the device by running many more threads than there are hardware execution units. Whenever a High-performance system design often involves trading thread stalls, the hardware scheduler immediately proceeds with off sequential performance against parallel throughput. The the execution of another thread ready for execution. This latency architectural difference between host and device processors hiding technique is key to make best use of the available hardware perfectly showcases the two extremes of this design space.
    [Show full text]
  • On the Virtualization of CUDA Based GPU Remoting on ARM and X86 Machines in the Gvirtus Framework
    On the Virtualization of CUDA Based GPU Remoting on ARM and X86 Machines in the GVirtuS Framework Montella, R., Giunta, G., Laccetti, G., Lapegna, M., Palmieri, C., Ferraro, C., Pelliccia, V., Hong, C-H., Spence, I., & Nikolopoulos, D. (2017). On the Virtualization of CUDA Based GPU Remoting on ARM and X86 Machines in the GVirtuS Framework. International Journal of Parallel Programming, 45(5), 1142-1163. https://doi.org/10.1007/s10766-016-0462-1 Published in: International Journal of Parallel Programming Document Version: Peer reviewed version Queen's University Belfast - Research Portal: Link to publication record in Queen's University Belfast Research Portal Publisher rights © 2016 Springer Verlag. The final publication is available at Springer via http://dx.doi.org/ 10.1007/s10766-016-0462-1 General rights Copyright for the publications made accessible via the Queen's University Belfast Research Portal is retained by the author(s) and / or other copyright owners and it is a condition of accessing these publications that users recognise and abide by the legal requirements associated with these rights. Take down policy The Research Portal is Queen's institutional repository that provides access to Queen's research output. Every effort has been made to ensure that content in the Research Portal does not infringe any person's rights, or applicable UK laws. If you discover content in the Research Portal that you believe breaches copyright or violates any law, please contact [email protected]. Download date:08. Oct. 2021 Noname manuscript No. (will be inserted by the editor) On the virtualization of CUDA based GPU remoting on ARM and X86 machines in the GVirtuS framework Raffaele Montella · Giulio Giunta · Giuliano Laccetti · Marco Lapegna · Carlo Palmieri · Carmine Ferraro · Valentina Pelliccia · Cheol-Ho Hong · Ivor Spence · Dimitrios S.
    [Show full text]
  • HPVM: Heterogeneous Parallel Virtual Machine
    HPVM: Heterogeneous Parallel Virtual Machine Maria Kotsifakou∗ Prakalp Srivastava∗ Matthew D. Sinclair Department of Computer Science Department of Computer Science Department of Computer Science University of Illinois at University of Illinois at University of Illinois at Urbana-Champaign Urbana-Champaign Urbana-Champaign [email protected] [email protected] [email protected] Rakesh Komuravelli Vikram Adve Sarita Adve Qualcomm Technologies Inc. Department of Computer Science Department of Computer Science [email protected]. University of Illinois at University of Illinois at com Urbana-Champaign Urbana-Champaign [email protected] [email protected] Abstract hardware, and that runtime scheduling policies can make We propose a parallel program representation for heteroge- use of both program and runtime information to exploit the neous systems, designed to enable performance portability flexible compilation capabilities. Overall, we conclude that across a wide range of popular parallel hardware, including the HPVM representation is a promising basis for achieving GPUs, vector instruction sets, multicore CPUs and poten- performance portability and for implementing parallelizing tially FPGAs. Our representation, which we call HPVM, is a compilers for heterogeneous parallel systems. hierarchical dataflow graph with shared memory and vector CCS Concepts • Computer systems organization → instructions. HPVM supports three important capabilities for Heterogeneous (hybrid) systems; programming heterogeneous systems: a compiler interme- diate representation (IR), a virtual instruction set (ISA), and Keywords Virtual ISA, Compiler, Parallel IR, Heterogeneous a basis for runtime scheduling; previous systems focus on Systems, GPU, Vector SIMD only one of these capabilities. As a compiler IR, HPVM aims to enable effective code generation and optimization for het- 1 Introduction erogeneous systems.
    [Show full text]
  • Productive High Performance Parallel Programming with Auto-Tuned Domain-Specific Embedded Languages
    Productive High Performance Parallel Programming with Auto-tuned Domain-Specific Embedded Languages By Shoaib Ashraf Kamil A dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in Computer Science in the Graduate Division of the University of California, Berkeley Committee in charge: Professor Armando Fox, Co-Chair Professor Katherine Yelick, Co-Chair Professor James Demmel Professor Berend Smit Fall 2012 Productive High Performance Parallel Programming with Auto-tuned Domain-Specific Embedded Languages Copyright c 2012 Shoaib Kamil. Abstract Productive High Performance Parallel Programming with Auto-tuned Domain-Specific Embedded Languages by Shoaib Ashraf Kamil Doctor of Philosophy in Computer Science University of California, Berkeley Professor Armando Fox, Co-Chair Professor Katherine Yelick, Co-Chair As the complexity of machines and architectures has increased, performance tuning has become more challenging, leading to the failure of general compilers to generate the best possible optimized code. Expert performance programmers can often hand-write code that outperforms compiler- optimized low-level code by an order of magnitude. At the same time, the complexity of programs has also increased, with modern programs built on a variety of abstraction layers to manage complexity, yet these layers hinder efforts at optimization. In fact, it is common to lose one or two additional orders of magnitude in performance when going from a low-level language such as Fortran or C to a high-level language like Python, Ruby, or Matlab. General purpose compilers are limited by the inability of program analysis to determine pro- grammer intent, as well as the lack of detailed performance models that always determine the best executable code for a given computation and architecture.
    [Show full text]
  • Parallel Processing Using PVM on a Linux Cluster
    ParallelParallel ProcessingProcessing usingusing PVMPVM onon aa LinuxLinux ClusterCluster ThomasThomas K.K. GederbergGederberg CENGCENG 65326532 FallFall 20072007 WhatWhat isis PVM?PVM? PVM (Parallel Virtual Machine) is a “software system that permits a heterogeneous collection of Unix computers networked together to be viewed by a user’s program as a single parallel computer”.1 Since the distributed computers have no shared memory, the PVM system uses Message Passing for data communication among the processors. PVM and MPI (Message Passing Interface) are the two common message passing API’s for distributed computing. MPI, the newer of the two API’s, was designed for homogenous distributed computers and offers higher performance. PVM, however, sacrifices performance for flexibility. PVM was designed to run over a heterogeneous network of distributed computers. PVM also allows for fault tolerance – it allows for the detection of, and reconfiguration for, node failures. PVM includes both C and Fortran libraries. PVM was developed by Oak Ridge National Laboratory, the University of Tennessee, Emory University, and Carnegie Mellon University. 2 InstallingInstalling PVMPVM inin LinuxLinux PVM is easy to install on a machine running Linux and can be installed in the users home directory (does not need root or administrator privileges). Obtain the PVM source code (latest version is 3.4.5) from the PVM website: http://www.csm.ornl.gov/pvm/pvm_home.html Unzip and untar the package (pvm3.4.5.tgz) in your home directory – this will create a directory called pvm3 Modify your startup script (.bashrc, .cshrc, etc.) to define: $PVM_ROOT = $HOME/pvm3 $PVM_ARCH = LINUX cd to the pvm3 directory Type make The makefile will build pvm (the PVM console), pvmd3 (the pvm daemon), libpvm3.a (PVM C/C++ library), libfpvm3.a (PVM Fortran library), and libgpvm3.a (PVM group library) and places all of these files in the $PVM_ROOT/lib/LINUX directory.
    [Show full text]
  • Parallel/Distributed Computing Using Parallel Virtual Machine (PVM)
    University of Liverpool Department of Computer Science Parallel/Distributed Computing using Parallel Virtual Machine (PVM) Prof. Wojciech Rytter Dr. Aris Pagourtzis Parallel Computation breaking a large task to smaller independent tasks for all x in [1..n] in parallel do_sth(x); x :=1 x :=2 x :=3 x :=n do_sth(x) do_sth(x) do_sth(x) . do_sth(x) What is PVM? Message passing system that enables us to create a virtual parallel computer using a network of workstations. Basic book: PVM: Parallel Virtual Machine A Users' Guide and Tutorial for Networked Parallel Computing A.Geist, A. Benguelin, J. Dongarra, W. Jiang, R. Manchek and V. Sunderman Available in postscript: http://www.netlib.org/pvm3/book/pvm-book.ps and html: http://www.netlib.org/pvm3/book/pvm-book.html Advantages of PVM • Ability to establish a parallel machine using ordinary (low cost) workstations. • Ability to connect heterogeneous machines. • Use of simple C/Fortran functions. • Takes care of data conversion and low- level communication issues. • Easily installable and configurable. Disadvantages of PVM • Programmer has to explicitly implement all parallelization details. • Difficult debugging. How PVM works The programmer writes a master program and one (or more) slave program(s) in C, C++ or Fortran 77 making use of PVM library calls. master program <----> master task slave program <-----> group of slave tasks The master task is started directly by the user in the ordinary way. Slaves start concurrently by the master task by appropriate PVM call. Slaves can also be masters of new slaves. Basic operations Each task can: • Spawn tasks and associate them with some executable program.
    [Show full text]
  • Parallel Programming in Openmp About the Authors
    Parallel Programming in OpenMP About the Authors Rohit Chandra is a chief scientist at NARUS, Inc., a provider of internet business infrastructure solutions. He previously was a principal engineer in the Compiler Group at Silicon Graphics, where he helped design and implement OpenMP. Leonardo Dagum works for Silicon Graphics in the Linux Server Platform Group, where he is responsible for the I/O infrastructure in SGI’s scalable Linux server systems. He helped define the OpenMP Fortran API. His research interests include parallel algorithms and performance modeling for parallel systems. Dave Kohr is a member of the technical staff at NARUS, Inc. He previ- ously was a member of the technical staff in the Compiler Group at Silicon Graphics, where he helped define and implement the OpenMP. Dror Maydan is director of software at Tensilica, Inc., a provider of appli- cation-specific processor technology. He previously was an engineering department manager in the Compiler Group of Silicon Graphics, where he helped design and implement OpenMP. Jeff McDonald owns SolidFX, a private software development company. As the engineering department manager at Silicon Graphics, he proposed the OpenMP API effort and helped develop it into the industry standard it is today. Ramesh Menon is a staff engineer at NARUS, Inc. Prior to NARUS, Ramesh was a staff engineer at SGI, representing SGI in the OpenMP forum. He was the founding chairman of the OpenMP Architecture Review Board (ARB) and supervised the writing of the first OpenMP specifica- tions. Parallel Programming in OpenMP Rohit Chandra Leonardo Dagum Dave Kohr Dror Maydan Jeff McDonald Ramesh Menon Senior Editor Denise E.
    [Show full text]
  • The Design and Implementation of a Bytecode for Optimization On
    Trinity University Digital Commons @ Trinity Computer Science Honors Theses Computer Science Department 4-20-2012 The esiD gn and Implementation of a Bytecode for Optimization on Heterogeneous Systems Kerry A. Seitz Trinity University, [email protected] Follow this and additional works at: http://digitalcommons.trinity.edu/compsci_honors Part of the Computer Sciences Commons Recommended Citation Seitz, Kerry A., "The eD sign and Implementation of a Bytecode for Optimization on Heterogeneous Systems" (2012). Computer Science Honors Theses. 28. http://digitalcommons.trinity.edu/compsci_honors/28 This Thesis open access is brought to you for free and open access by the Computer Science Department at Digital Commons @ Trinity. It has been accepted for inclusion in Computer Science Honors Theses by an authorized administrator of Digital Commons @ Trinity. For more information, please contact [email protected]. The Design and Implementation of a Bytecode for Optimization on Heterogeneous Systems Kerry Allen Seitz, Jr. Abstract As hardware architectures shift towards more heterogeneous platforms with different vari- eties of multi- and many-core processors and graphics processing units (GPUs) by various manufacturers, programmers need a way to write simple and highly optimized code without worrying about the specifics of the underlying hardware. To meet this need, I have designed a virtual machine and bytecode around the goal of optimized execution on highly variable, heterogeneous hardware, instead of having goals such as small bytecodes as was the ob- jective of the JavaR Virtual Machine. The approach used here is to combine elements of the DalvikR virtual machine with concepts from the OpenCLR heterogeneous computing platform, along with an annotation system so that the results of complex compile time analysis can be available to the Just-In-Time compiler.
    [Show full text]
  • Hybrid MPI & Openmp Parallel Programming
    Hybrid MPI & OpenMP Parallel Programming MPI + OpenMP and other models on clusters of SMP nodes Rolf Rabenseifner1) Georg Hager2) Gabriele Jost3) [email protected] [email protected] [email protected] 1) High Performance Computing Center (HLRS), University of Stuttgart, Germany 2) Regional Computing Center (RRZE), University of Erlangen, Germany 3) Texas Advanced Computing Center, The University of Texas at Austin, USA Tutorial M02 at SC10, November 15, 2010, New Orleans, LA, USA Hybrid Parallel Programming Slide 1 Höchstleistungsrechenzentrum Stuttgart Outline slide number • Introduction / Motivation 2 • Programming models on clusters of SMP nodes 6 8:30 – 10:00 • Case Studies / pure MPI vs hybrid MPI+OpenMP 13 • Practical “How-To” on hybrid programming 48 • Mismatch Problems 97 • Opportunities: Application categories that can 126 benefit from hybrid parallelization • Thread-safety quality of MPI libraries 136 10:30 – 12:00 • Tools for debugging and profiling MPI+OpenMP 143 • Other options on clusters of SMP nodes 150 • Summary 164 • Appendix 172 • Content (detailed) 188 Hybrid Parallel Programming Slide 2 / 169 Rabenseifner, Hager, Jost Motivation • Efficient programming of clusters of SMP nodes SMP nodes SMP nodes: cores shared • Dual/multi core CPUs memory • Multi CPU shared memory Node Interconnect • Multi CPU ccNUMA • Any mixture with shared memory programming model • Hardware range • mini-cluster with dual-core CPUs Core • … CPU(socket) • large constellations with large SMP nodes SMP board … with several sockets
    [Show full text]
  • Parallel Hybrid Testing Techniques for the Dual-Programming Models-Based Programs
    S S symmetry Article Parallel Hybrid Testing Techniques for the Dual-Programming Models-Based Programs Ahmed Mohammed Alghamdi 1,* , Fathy Elbouraey Eassa 2, Maher Ali Khamakhem 2, Abdullah Saad AL-Malaise AL-Ghamdi 3 , Ahmed S. Alfakeeh 3 , Abdullah S. Alshahrani 4 and Ala A. Alarood 5 1 Department of Software Engineering, College of Computer Science and Engineering, University of Jeddah, Jeddah 21493, Saudi Arabia 2 Department of Computer Science, Faculty of Computing and Information Technology, King Abdulaziz University, Jeddah 21589, Saudi Arabia; [email protected] (F.E.E.); [email protected] (M.A.K.) 3 Department of Information Systems, Faculty of Computing and Information Technology, King Abdulaziz University, Jeddah 21589, Saudi Arabia; [email protected] (A.S.A.-M.A.-G.); [email protected] (A.S.A.) 4 Department of Computer Science and Artificial Intelligence, College of Computer Science and Engineering, University of Jeddah, Jeddah 21493, Saudi Arabia; [email protected] 5 College of Computer Science and Engineering, University of Jeddah, Jeddah 21432, Saudi Arabia; [email protected] or [email protected] * Correspondence: [email protected] Received: 4 September 2020; Accepted: 18 September 2020; Published: 20 September 2020 Abstract: The importance of high-performance computing is increasing, and Exascale systems will be feasible in a few years. These systems can be achieved by enhancing the hardware’s ability as well as the parallelism in the application by integrating more than one programming model. One of the dual-programming model combinations is Message Passing Interface (MPI) + OpenACC, which has several features including increased system parallelism, support for different platforms with more performance, better productivity, and less programming effort.
    [Show full text]