DOCSLIB.ORG

High Performance Computing Through Parallel and Distributed Processing

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
High Performance Computing Through Parallel and Distributed Processing Yadav S. et al., J. Harmoniz. Res. Eng., 2013, 1(2), 54-64 Journal Of Harmonized Research (JOHR) Journal Of Harmonized Research in Engineering 1(2), 2013, 54-64 ISSN 2347 – 7393 Original Research Article High Performance Computing through Parallel and Distributed Processing Shikha Yadav, Preeti Dhanda, Nisha Yadav Department of Computer Science and Engineering, Dronacharya College of Engineering, Khentawas, Farukhnagar, Gurgaon, India Abstract : There is a very high need of High Performance Computing (HPC) in many applications like space science to Artificial Intelligence. HPC shall be attained through Parallel and Distributed Computing. In this paper, Parallel and Distributed algorithms are discussed based on Parallel and Distributed Processors to achieve HPC. The Programming concepts like threads, fork and sockets are discussed with some simple examples for HPC. Keywords: High Performance Computing, Parallel and Distributed processing, Computer Architecture Introduction time to solve large problems like weather Computer Architecture and Programming play forecasting, Tsunami, Remote Sensing, a significant role for High Performance National calamities, Defence, Mineral computing (HPC) in large applications Space exploration, Finite-element, Cloud science to Artificial Intelligence. The Computing, and Expert Systems etc. The Algorithms are problem solving procedures Algorithms are Non-Recursive Algorithms, and later these algorithms transform in to Recursive Algorithms, Parallel Algorithms particular Programming language for HPC. and Distributed Algorithms. There is need to study algorithms for High The Algorithms must be supported the Performance Computing. These Algorithms Computer Architecture. The Computer are to be designed to computer in reasonable Architecture is characterized with Flynn’s Classification SISD, SIMD, MIMD, and For Correspondence: MISD. Most of the Computer Architectures preeti.dhanda01ATgmail.com are supported with SIMD (Single Instruction Received on: October 2013 Multiple Data Streams). The class of Accepted after revision: December 2013 Computer Architecture is VLSI Processor, Multi-processor, Vector Processor and Downloaded from: www.johronline.com Multiple Processor. www.johronline.com 54 | P a g e Yadav S. et al., J. Harmoniz. Res. Eng., 2013, 1(2), 54-64 computing is using clusters of independent processors connected in parallel. Many computing problems are suitable for parallelization, often problems can be divided in a manner so that each independent processing node can work on a portion of the problem in parallel by simply dividing the data to be processed, and then combining the final processing results for each portion. This type of parallelism is often referred to as data- High-Performance Computing (HPC) is used parallelism, and data-parallel applications are to describe computing environments which a potential solution to petabyte scale data utilize supercomputers and computer clusters processing requirements. Data-parallelism can to address complex computational be defined as a computation applied requirements, support applications with independently to each data item of a set of significant processing time requirements, or data which allows the degree of parallelism to require processing of significant amounts of be scaled with the volume of data. The most data. Supercomputers have generally been important reason for developing data-parallel associated with scientific research and applications is the potential for scalable compute-intensive types of problems, but performance in high-performance computing, more and more supercomputer technology is and may result in several orders of magnitude appropriate for both compute-intensive and performance improvement. data-intensive applications. A new trend in supercomputer design for high-performance The discussion below focuses on the case of Though, the classical PRAM model assumes multiple computers, although many of the synchronous access to the shared memory. issues are the same for concurrent processes A model that is closer to the behaviour of real- running on a single computer. world multiprocessor machines and takes into Three view points are commonly used: account the use of machine instructions, such Parallel algorithms in shared-memory model - as Compare-and-swap (CAS), is that of All computers have access to a shared asynchronous shared memory. memory. The algorithm designer chooses the Parallel algorithms in message-passing model- program executed by each computer. The algorithm designer chooses the structure One theoretical model is the parallel random of the network, as well as the program access machines (PRAM) that are used. executed by each computer. www.johronline.com 55 | P a g e Yadav S. et al., J. Harmoniz. Res. Eng., 2013, 1(2), 54-64 Models such as Boolean circuits and sorting analytics, and creation of keyed data and networks are used. A Boolean circuit can be indexes to support high-performance seen as a computer network: each gate is a structured queries and data warehouse computer that runs an extremely simple applications. The Data Refinery is also computer program. Similarly, a sorting referred to as Thor, a reference to the mythical network can be seen as a computer network, Norse god of thunder with the large hammer each comparator is a computer. symbolic of crushing large amounts of raw Distributed algorithms in message-passing data into useful information. A Thor cluster is model- similar in its function, execution environment, The algorithm designer only chooses the filesystem, and capabilities to the Google. computer program. All computers run the The figure given below shows a representation same program. The system must work of a physical Thor processing cluster which accurately regardless of the structure of the functions as a batch job execution engine for network.A commonly used model is a graph scalable data-intensive computing with one finite-state machine per node. applications. In addition to the Thor master Architecture of high performance and slave nodes, additional auxiliary and computers: common components are needed to implement The Department of High Performance a complete HPCC processing environment. Computer Architecture is concerned with the architecture, the application and the continued development of high performance computers beneficial to the natural and life sciences. Our focus is on the selection and analysis of experimental data generated by accelerator facilities such as GSI ("Society for Heavy Ion Research" in Darmstadt, Germany) and CERN (the European Centre for Nuclear Research in Geneva, Switzerland). Both of these facilities employ shared, typically massive parallel systems and clusters operating under high- Thor Processing Cluster level, real-time and dependability standards. The second of the parallel data processing Our task is the research and development of platforms is called Roxie and functions as a new computer architectures and algorithms to rapid data delivery engine. This platform is achieve better energy-efficiency. Within the designed as an online high-performance context of shared computing we implement structured query and analysis platform or data both GRID and virtual technologies as well as warehouse delivering the parallel data access cloud computing systems. processing requirements of online applications High Performance Computing Cluster through Web services interfaces supporting (HPCC) System Architecture- thousands of simultaneous queries and users The HPCC system architecture includes two with sub-second response times. Roxie utilizes distinct cluster processing environments, each a distributed indexed file system to provide of which can be optimized independently for parallel processing of queries using an its parallel data processing purpose. The first optimized execution environment and of these platforms is called a Data Refinery filesystem for high-performance online whose overall purpose is the general processing. Both Thor and Roxie clusters processing of massive volumes of raw data of utilize the ECL programming language for any type for any purpose but typically used for implementing applications, increasing data cleansing and hygiene, ETL processing of continuity and programmer productivity. the raw data, record linking and entity the figure given below shows a representation resolution, large-scale ad-hoc complex of a physical Roxie processing cluster which www.johronline.com 56 | P a g e Yadav S. et al., J. Harmoniz. Res. Eng., 2013, 1(2), 54-64 functions as an online query execution engine A computer program that runs in a distributed for high-performance query and data system is called a distributed program, and warehousing applications. A Roxie cluster distributed programming is the process of includes multiple nodes with server and writing such programs. worker processes for processing queries; an Distributed computing also refers to the use of additional auxiliary component called an ESP distributed systems to solve computational server which provides interfaces for external problems. In distributed computing, a problem client access to the cluster; and additional is divided into many tasks, each of which is common components which are shared with a solved by one or more computers which Thor cluster in an HPCC environment. communicate with each other by message Although a Thor processing cluster can be passing. implemented and used without a Roxie Properties of distributed systems - cluster, an HPCC environment which includes So
Load more

Recommended publications
  • Data-Level Parallelism
    Fall 2015 :: CSE 610 – Parallel Computer Architectures Data-Level Parallelism Nima Honarmand Fall 2015 :: CSE 610 – Parallel Computer Architectures Overview • Data Parallelism vs. Control Parallelism – Data Parallelism: parallelism arises from executing essentially the same code on a large number of objects – Control Parallelism: parallelism arises from executing different threads of control concurrently • Hypothesis: applications that use massively parallel machines will mostly exploit data parallelism – Common in the Scientific Computing domain • DLP originally linked with SIMD machines; now SIMT is more common – SIMD: Single Instruction Multiple Data – SIMT: Single Instruction Multiple Threads Fall 2015 :: CSE 610 – Parallel Computer Architectures Overview • Many incarnations of DLP architectures over decades – Old vector processors • Cray processors: Cray-1, Cray-2, …, Cray X1 – SIMD extensions • Intel SSE and AVX units • Alpha Tarantula (didn’t see light of day ) – Old massively parallel computers • Connection Machines • MasPar machines – Modern GPUs • NVIDIA, AMD, Qualcomm, … • Focus of throughput rather than latency Vector Processors 4 SCALAR VECTOR (1 operation) (N operations) r1 r2 v1 v2 + + r3 v3 vector length add r3, r1, r2 vadd.vv v3, v1, v2 Scalar processors operate on single numbers (scalars) Vector processors operate on linear sequences of numbers (vectors) 6.888 Spring 2013 - Sanchez and Emer - L14 What’s in a Vector Processor? 5 A scalar processor (e.g. a MIPS processor) Scalar register file (32 registers) Scalar functional units (arithmetic, load/store, etc) A vector register file (a 2D register array) Each register is an array of elements E.g. 32 registers with 32 64-bit elements per register MVL = maximum vector length = max # of elements per register A set of vector functional units Integer, FP, load/store, etc Some times vector and scalar units are combined (share ALUs) 6.888 Spring 2013 - Sanchez and Emer - L14 Example of Simple Vector Processor 6 6.888 Spring 2013 - Sanchez and Emer - L14 Basic Vector ISA 7 Instr.
    [Show full text]
  • Introduction to Multi-Threading and Vectorization Matti Kortelainen Larsoft Workshop 2019 25 June 2019 Outline
    Introduction to multi-threading and vectorization Matti Kortelainen LArSoft Workshop 2019 25 June 2019 Outline Broad introductory overview: • Why multithread? • What is a thread? • Some threading models – std::thread – OpenMP (fork-join) – Intel Threading Building Blocks (TBB) (tasks) • Race condition, critical region, mutual exclusion, deadlock • Vectorization (SIMD) 2 6/25/19 Matti Kortelainen | Introduction to multi-threading and vectorization Motivations for multithreading Image courtesy of K. Rupp 3 6/25/19 Matti Kortelainen | Introduction to multi-threading and vectorization Motivations for multithreading • One process on a node: speedups from parallelizing parts of the programs – Any problem can get speedup if the threads can cooperate on • same core (sharing L1 cache) • L2 cache (may be shared among small number of cores) • Fully loaded node: save memory and other resources – Threads can share objects -> N threads can use significantly less memory than N processes • If smallest chunk of data is so big that only one fits in memory at a time, is there any other option? 4 6/25/19 Matti Kortelainen | Introduction to multi-threading and vectorization What is a (software) thread? (in POSIX/Linux) • “Smallest sequence of programmed instructions that can be managed independently by a scheduler” [Wikipedia] • A thread has its own – Program counter – Registers – Stack – Thread-local memory (better to avoid in general) • Threads of a process share everything else, e.g. – Program code, constants – Heap memory – Network connections – File handles
    [Show full text]
  • Demystifying Parallel and Distributed Deep Learning: an In-Depth Concurrency Analysis
    1 Demystifying Parallel and Distributed Deep Learning: An In-Depth Concurrency Analysis TAL BEN-NUN and TORSTEN HOEFLER, ETH Zurich, Switzerland Deep Neural Networks (DNNs) are becoming an important tool in modern computing applications. Accelerating their training is a major challenge and techniques range from distributed algorithms to low-level circuit design. In this survey, we describe the problem from a theoretical perspective, followed by approaches for its parallelization. We present trends in DNN architectures and the resulting implications on parallelization strategies. We then review and model the different types of concurrency in DNNs: from the single operator, through parallelism in network inference and training, to distributed deep learning. We discuss asynchronous stochastic optimization, distributed system architectures, communication schemes, and neural architecture search. Based on those approaches, we extrapolate potential directions for parallelism in deep learning. CCS Concepts: • General and reference → Surveys and overviews; • Computing methodologies → Neural networks; Parallel computing methodologies; Distributed computing methodologies; Additional Key Words and Phrases: Deep Learning, Distributed Computing, Parallel Algorithms ACM Reference Format: Tal Ben-Nun and Torsten Hoefler. 2018. Demystifying Parallel and Distributed Deep Learning: An In-Depth Concurrency Analysis. 47 pages. 1 INTRODUCTION Machine Learning, and in particular Deep Learning [143], is rapidly taking over a variety of aspects in our daily lives.
    [Show full text]
  • Parallel Programming: Performance
    Introduction Parallel Programming: Rich space of techniques and issues • Trade off and interact with one another Performance Issues can be addressed/helped by software or hardware • Algorithmic or programming techniques • Architectural techniques Todd C. Mowry Focus here on performance issues and software techniques CS 418 • Point out some architectural implications January 20, 25 & 26, 2011 • Architectural techniques covered in rest of class –2– CS 418 Programming as Successive Refinement Outline Not all issues dealt with up front 1. Partitioning for performance Partitioning often independent of architecture, and done first 2. Relationshipp,y of communication, data locality and • View machine as a collection of communicating processors architecture – balancing the workload – reducing the amount of inherent communication 3. Orchestration for performance – reducing extra work • Tug-o-war even among these three issues For each issue: Then interactions with architecture • Techniques to address it, and tradeoffs with previous issues • Illustration using case studies • View machine as extended memory hierarchy • Application to g rid solver –extra communication due to archlhitectural interactions • Some architectural implications – cost of communication depends on how it is structured • May inspire changes in partitioning Discussion of issues is one at a time, but identifies tradeoffs 4. Components of execution time as seen by processor • Use examples, and measurements on SGI Origin2000 • What workload looks like to architecture, and relate to software issues –3– CS 418 –4– CS 418 Page 1 Partitioning for Performance Load Balance and Synch Wait Time Sequential Work 1. Balancing the workload and reducing wait time at synch Limit on speedup: Speedupproblem(p) < points Max Work on any Processor • Work includes data access and other costs 2.
    [Show full text]
  • Exploiting Parallelism in Many-Core Architectures: Architectures G Bortolotti, M Caberletti, G Crimi Et Al
    Journal of Physics: Conference Series OPEN ACCESS Related content - Computing on Knights and Kepler Exploiting parallelism in many-core architectures: Architectures G Bortolotti, M Caberletti, G Crimi et al. Lattice Boltzmann models as a test case - Implementing the lattice Boltzmann model on commodity graphics hardware Arie Kaufman, Zhe Fan and Kaloian To cite this article: F Mantovani et al 2013 J. Phys.: Conf. Ser. 454 012015 Petkov - Many-core experience with HEP software at CERN openlab Sverre Jarp, Alfio Lazzaro, Julien Leduc et al. View the article online for updates and enhancements. Recent citations - Lattice Boltzmann benchmark kernels as a testbed for performance analysis M. Wittmann et al - Enrico Calore et al - Computing on Knights and Kepler Architectures G Bortolotti et al This content was downloaded from IP address 170.106.202.58 on 23/09/2021 at 15:57 24th IUPAP Conference on Computational Physics (IUPAP-CCP 2012) IOP Publishing Journal of Physics: Conference Series 454 (2013) 012015 doi:10.1088/1742-6596/454/1/012015 Exploiting parallelism in many-core architectures: Lattice Boltzmann models as a test case F Mantovani1, M Pivanti2, S F Schifano3 and R Tripiccione4 1 Department of Physics, Univesit¨atRegensburg, Germany 2 Department of Physics, Universit`adi Roma La Sapienza, Italy 3 Department of Mathematics and Informatics, Universit`adi Ferrara and INFN, Italy 4 Department of Physics and CMCS, Universit`adi Ferrara and INFN, Italy E-mail: [email protected], [email protected], [email protected], [email protected] Abstract. In this paper we address the problem of identifying and exploiting techniques that optimize the performance of large scale scientific codes on many-core processors.
    [Show full text]
  • An Introduction to Gpus, CUDA and Opencl
    An Introduction to GPUs, CUDA and OpenCL Bryan Catanzaro, NVIDIA Research Overview ¡ Heterogeneous parallel computing ¡ The CUDA and OpenCL programming models ¡ Writing efficient CUDA code ¡ Thrust: making CUDA C++ productive 2/54 Heterogeneous Parallel Computing Latency-Optimized Throughput- CPU Optimized GPU Fast Serial Scalable Parallel Processing Processing 3/54 Why do we need heterogeneity? ¡ Why not just use latency optimized processors? § Once you decide to go parallel, why not go all the way § And reap more benefits ¡ For many applications, throughput optimized processors are more efficient: faster and use less power § Advantages can be fairly significant 4/54 Why Heterogeneity? ¡ Different goals produce different designs § Throughput optimized: assume work load is highly parallel § Latency optimized: assume work load is mostly sequential ¡ To minimize latency eXperienced by 1 thread: § lots of big on-chip caches § sophisticated control ¡ To maXimize throughput of all threads: § multithreading can hide latency … so skip the big caches § simpler control, cost amortized over ALUs via SIMD 5/54 Latency vs. Throughput Specificaons Westmere-EP Fermi (Tesla C2050) 6 cores, 2 issue, 14 SMs, 2 issue, 16 Processing Elements 4 way SIMD way SIMD @3.46 GHz @1.15 GHz 6 cores, 2 threads, 4 14 SMs, 48 SIMD Resident Strands/ way SIMD: vectors, 32 way Westmere-EP (32nm) Threads (max) SIMD: 48 strands 21504 threads SP GFLOP/s 166 1030 Memory Bandwidth 32 GB/s 144 GB/s Register File ~6 kB 1.75 MB Local Store/L1 Cache 192 kB 896 kB L2 Cache 1.5 MB 0.75 MB
    [Show full text]
  • Parallel Computing a Key to Performance
    Parallel Computing A Key to Performance Dheeraj Bhardwaj Department of Computer Science & Engineering Indian Institute of Technology, Delhi –110 016 India http://www.cse.iitd.ac.in/~dheerajb Dheeraj Bhardwaj <[email protected]> August, 2002 1 Introduction • Traditional Science • Observation • Theory • Experiment -- Most expensive • Experiment can be replaced with Computers Simulation - Third Pillar of Science Dheeraj Bhardwaj <[email protected]> August, 2002 2 1 Introduction • If your Applications need more computing power than a sequential computer can provide ! ! ! ❃ Desire and prospect for greater performance • You might suggest to improve the operating speed of processors and other components. • We do not disagree with your suggestion BUT how long you can go ? Can you go beyond the speed of light, thermodynamic laws and high financial costs ? Dheeraj Bhardwaj <[email protected]> August, 2002 3 Performance Three ways to improve the performance • Work harder - Using faster hardware • Work smarter - - doing things more efficiently (algorithms and computational techniques) • Get help - Using multiple computers to solve a particular task. Dheeraj Bhardwaj <[email protected]> August, 2002 4 2 Parallel Computer Definition : A parallel computer is a “Collection of processing elements that communicate and co-operate to solve large problems fast”. Driving Forces and Enabling Factors Desire and prospect for greater performance Users have even bigger problems and designers have even more gates Dheeraj Bhardwaj <[email protected]>
    [Show full text]
  • Parallel System Performance: Evaluation & Scalability
    ParallelParallel SystemSystem Performance:Performance: EvaluationEvaluation && ScalabilityScalability • Factors affecting parallel system performance: – Algorithm-related, parallel program related, architecture/hardware-related. • Workload-Driven Quantitative Architectural Evaluation: – Select applications or suite of benchmarks to evaluate architecture either on real or simulated machine. – From measured performance results compute performance metrics: • Speedup, System Efficiency, Redundancy, Utilization, Quality of Parallelism. – Resource-oriented Workload scaling models: How the speedup of a parallel computation is affected subject to specific constraints: 1 • Problem constrained (PC): Fixed-load Model. 2 • Time constrained (TC): Fixed-time Model. 3 • Memory constrained (MC): Fixed-Memory Model. • Parallel Performance Scalability: For a given parallel system and a given parallel computation/problem/algorithm – Definition. Informally: – Conditions of scalability. The ability of parallel system performance to increase – Factors affecting scalability. with increased problem size and system size. Parallel Computer Architecture, Chapter 4 EECC756 - Shaaban Parallel Programming, Chapter 1, handout #1 lec # 9 Spring2013 4-23-2013 Parallel Program Performance • Parallel processing goal is to maximize speedup: Time(1) Sequential Work Speedup = < Time(p) Max (Work + Synch Wait Time + Comm Cost + Extra Work) Fixed Problem Size Speedup Max for any processor Parallelizing Overheads • By: 1 – Balancing computations/overheads (workload) on processors
    [Show full text]
  • Scalable Task Parallel Programming in the Partitioned Global Address Space
    Scalable Task Parallel Programming in the Partitioned Global Address Space DISSERTATION Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University By James Scott Dinan, M.S. Graduate Program in Computer Science and Engineering The Ohio State University 2010 Dissertation Committee: P. Sadayappan, Advisor Atanas Rountev Paolo Sivilotti c Copyright by James Scott Dinan 2010 ABSTRACT Applications that exhibit irregular, dynamic, and unbalanced parallelism are grow- ing in number and importance in the computational science and engineering commu- nities. These applications span many domains including computational chemistry, physics, biology, and data mining. In such applications, the units of computation are often irregular in size and the availability of work may be depend on the dynamic, often recursive, behavior of the program. Because of these properties, it is challenging for these programs to achieve high levels of performance and scalability on modern high performance clusters. A new family of programming models, called the Partitioned Global Address Space (PGAS) family, provides the programmer with a global view of shared data and allows for asynchronous, one-sided access to data regardless of where it is physically stored. In this model, the global address space is distributed across the memories of multiple nodes and, for any given node, is partitioned into local patches that have high affinity and low access cost and remote patches that have a high access cost due to communication. The PGAS data model relaxes conventional two-sided communication semantics and allows the programmer to access remote data without the cooperation of the remote processor.
    [Show full text]
  • Computer Architecture: Parallel Processing Basics
    Computer Architecture: Parallel Processing Basics Onur Mutlu & Seth Copen Goldstein Carnegie Mellon University 9/9/13 Today What is Parallel Processing? Why? Kinds of Parallel Processing Multiprocessing and Multithreading Measuring success Speedup Amdhal’s Law Bottlenecks to parallelism 2 Concurrent Systems Embedded-Physical Distributed Sensor Claytronics Networks Concurrent Systems Embedded-Physical Distributed Sensor Claytronics Networks Geographically Distributed Power Internet Grid Concurrent Systems Embedded-Physical Distributed Sensor Claytronics Networks Geographically Distributed Power Internet Grid Cloud Computing EC2 Tashi PDL'09 © 2007-9 Goldstein5 Concurrent Systems Embedded-Physical Distributed Sensor Claytronics Networks Geographically Distributed Power Internet Grid Cloud Computing EC2 Tashi Parallel PDL'09 © 2007-9 Goldstein6 Concurrent Systems Physical Geographical Cloud Parallel Geophysical +++ ++ --- --- location Relative +++ +++ + - location Faults ++++ +++ ++++ -- Number of +++ +++ + - Processors + Network varies varies fixed fixed structure Network --- --- + + connectivity 7 Concurrent System Challenge: Programming The old joke: How long does it take to write a parallel program? One Graduate Student Year 8 Parallel Programming Again?? Increased demand (multicore) Increased scale (cloud) Improved compute/communicate Change in Application focus Irregular Recursive data structures PDL'09 © 2007-9 Goldstein9 Why Parallel Computers? Parallelism: Doing multiple things at a time Things: instructions,
    [Show full text]
  • 1 Parallelism
    Lecture 23: Parallelism • Administration – Take QUIZ 17 over P&H 7.1-5, before 11:59pm today – Project: Cache Simulator, Due April 29, 2010 • Last Time – On chip communication – How I/O works • Today – Where do architectures exploit parallelism? – What are the implications for programming models? – What are the implications for communication? – … for caching? UTCS 352, Lecture 23 1 Parallelism • What type of parallelism do applications have? • How can computer architectures exploit this parallelism? UTCS 352, Lecture 23 2 1 Granularity of Parallelism • Fine grain instruction level parallelism • Fine grain data parallelism • Coarse grain (data center) parallelism • Multicore parallelism UTCS 352, Lecture 23 3 Fine grain instruction level parallelism • Fine grain instruction level parallelism – Pipelining – Multi-issue (dynamic scheduling & issue) – VLIW (static issue) – Very Long Instruction Word • Each VLIW instruction includes up to N RISC-like operations • Compiler groups up to N independent operations • Architecture issues fixed size VLIW instructions UTCS 352, Lecture 23 4 2 VLIW Example Theory Let N = 4 Theoretically 4 ops/cycle • 8 VLIW = 32 operations in practice In practice, • Compiler cannot fill slots • Memory stalls - load stall - UTCS 352, Lecture 23 5 Multithreading • Performing multiple threads together – Replicate registers, PC, etc. – Fast switching between threads • Fine-grain multithreading – Switch threads after each cycle – Interleave instruction execution – If one thread stalls, others are executed • Medium-grain
    [Show full text]
  • Exploiting Parallelism Opportunities with Deep Learning Frameworks
    Exploiting Parallelism Opportunities with Deep Learning Frameworks YU EMMA WANG, Harvard University CAROLE-JEAN WU, Facebook XIAODONG WANG, Facebook KIM HAZELWOOD, Facebook DAVID BROOKS, Harvard University State-of-the-art machine learning frameworks support a wide variety of design features to enable a flexible machine learning programming interface and to ease the programmability burden on machine learning developers. Identifying and using a performance-optimal setting in feature-rich frameworks, however, involves a non-trivial amount of performance profiling efforts and often relies on domain-specific knowledge. This paper takesa deep dive into analyzing the performance impact of key design features in a machine learning framework and quantifies the role of parallelism. The observations and insights distill into a simple set of guidelines thatone can use to achieve much higher training and inference speedup. Across a diverse set of real-world deep learning models, the evaluation results show that the proposed performance tuning guidelines outperform the Intel and TensorFlow recommended settings by 1.29× and 1.34×, respectively. ACM Reference Format: Yu Emma Wang, Carole-Jean Wu, Xiaodong Wang, Kim Hazelwood, and David Brooks. 2020. Exploiting Parallelism Opportunities with Deep Learning Frameworks. 1, 1 (July 2020), 21 pages. https://doi.org/10.1145/ nnnnnnn.nnnnnnn 1 INTRODUCTION The wide adoption of deep learning (DL) tasks has spawned a plethora of domain-specific frameworks to help improve development efficiency and to enable a flexible path from research to production deployment for machine learning (ML). Notable ML frameworks include Caffe/Caffe2 [23], Py- Torch [24], TensorFlow [2], and MXNet [8]. These frameworks provide high-level APIs as the building blocks of DL models.
    [Show full text]