Parallelization of the Telemedicine Benchmark for the Xbox ...

Parallelization of the Telemedicine Benchmark for the Xbox 360 Architecture Howard Wong, SURF-IT Fellow Professor Jean-Luc Gaudiot, EECS August 29, 2008 PASCAL: PASCAL PArallel Systems and Computer Architecture Lab. University of California, Irvine Outline Background (Benchmark, Platform) Current Work Methodology (Compiler, Data Set) Results Conclusions Future Work PASCAL: PArallel Systems & Computer Architecture Lab. Background Work

Why Parallel Programming? Advent of everyday multicomputers Ultimate goal: Auto-parallelization Basic concepts ? Problems Programming primitives Telemedicine Benchmark Platform Xbox 360 3 Cores Graphics Engine

Vector Processing Core 1 Core 2 PASCAL: PArallel Systems & Computer Architecture Lab. Core n Current Work Goal: Identify the parallelization process Efficiency measured in performance Performance in relation to load POSIX threads (pthreads) and OpenMP Sorting Routines 'fallbackSort'

Making search 'brackets' 'mainSort' Dependencies between loop iterations PASCAL: PArallel Systems & Computer Architecture Lab. Methodology Compilation Data Sets

gcc or g++ version 4.2 Monkey brain image in PPM format Derived data via netpbm Test Platform Xbox 360 with Ubuntu Linux Images courtesy of Neuroscience Center, UC Davis, and Joerg Meyer, Center of GRAVITY, Calit2, UC Irvine. PASCAL: PArallel Systems & Computer Architecture Lab. Initial Results Speedup versus Number of Threads Compression of brains.ppm; Compared to bzip2 3.500 3.000 Speedup 2.500

2.000 bzip2mod Linear Linear 1.500 1.000 0.500 0.000 0 1 2 No. of Threads 3 4 PASCAL: PArallel Systems & Computer Architecture Lab.

Analysis Possible thread contention Thread Creation 'bitmap' of data as former optimization Optimized for long runs of 0's or 1's Extra mutex locks required Sorting algorithm called at least 300 times for the large image Thread creation efficiency Thread management structures PASCAL: PArallel Systems & Computer Architecture Lab.

Results (Contd) Speedup versus Load (pbzip2 - 3 Threads) Speedup versus Load (bzip2mod - 2 Threads) Compared to bzip2; 1/4, 1/2, whole image Compared to bzip2; 1/4, 1/2, whole image 3.050 0.690 0.680 3.000 0.670 Speedup Speedup 2.950 0.660

2.900 0.650 2.850 2.800 0.000 0.640 0.250 0.500 Processed 0.750 Fraction of Image 1.000 0.630 0.000 0.250 0.500 Processed 0.750 Fraction of Image PASCAL: PArallel Systems & Computer Architecture Lab.

1.000 Conclusions & Discussion Speedup dependent on the load size Possible improvements Use a 'threadpool' Create other important compression functions Examine alternative algorithms with a parallel mindset End result Thread creation Thread management overhead Heavy contention

PASCAL: PArallel Systems & Computer Architecture Lab. Questions for Future Work What is the impact of thread creation? Do the other TMB programs have the same features? Can vector instructions improve program performance? Are new, more efficient parallel programming primitives needed for our application? PASCAL: PArallel Systems & Computer Architecture Lab. Acknowledgments Professor Jean-Luc Gaudiot and the PASCAL group UC Davis Neuroscience Center Professor Joerg Meyer, Center of GRAVITY, Calit2 Calit2 UROP PASCAL: PArallel Systems & Computer Architecture Lab.

Recently Viewed Presentations

  • LIFE IN THE 1950S By Erika Cox LIFE

    LIFE IN THE 1950S By Erika Cox LIFE

    Popular celebrities were very influential, kids wanted to be just like Elvis Presley, Marilyn Monroe, James Dean, and Ricky Nelson to name a few. One thing to note is that the popular size for women back in the 1950's is...
  • ABO Blood Group System - الصفحات الشخصية

    ABO Blood Group System - الصفحات الشخصية

    Arial Wingdings Times New Roman Capsules 1_Capsules ABO Blood Group System Importance of ABO system ABO antigens ABO and H Antigen Genetics Type I and Type II Precursors RBC Precursor Structure Formation of the H antigen H antigen Formation of...
  • Junior Parent Night - Madison Public Schools

    Junior Parent Night - Madison Public Schools

    Junior Parent Night 2/12/2014 Brett Levine, Director of Guidance * Tonight's Program 60-75 minutes PPT on Web site Information rather than advice Financial Aid not covered Scholarships Covered at Sr. Parent night Informal Q/A up front at the end.
  • CO324 - Computer Systems

    CO324 - Computer Systems

    the project was called GNU - a recursive acronym standing for "GNU's Not UNIX" History of UNIX - Part II. Stallman. The GNU logo. Stallman is still a majorcampaigner. The slide shows a photograph of a very hairy (long hair,...
  • Introduction to Geometry: Points, Lines, and Planes PRE-ALGEBRA

    Introduction to Geometry: Points, Lines, and Planes PRE-ALGEBRA

    Use the figure to name each of the following. a. four points Name a point with a capital letter. ... and Planes Angle Relationships and Parallel Lines Angle Relationships and Parallel Lines Angle Relationships and Parallel Lines Angle Relationships and...
  • WIND - Wellington High School

    WIND - Wellington High School

    Wind is from Hughes' very first published collection of poems, Hawk in the Wind (1957).Like so many of his poems, Wind is related to nature, particularly as it appeared in his childhood setting of West Yorkshire.His parents lived in a...
  • B O N N E V I L

    B O N N E V I L

    Whold Facility (CDA, Blowers, etc) Yes, 10/1. United Electric Coop. High Desert Milk. ... Yes, Charlie Weber did approve this, but I don't have a record of the date. 12/24 Comments. Added 5 potential EPMs based on ESIP feedback. Moved...
  • Molecular Geometry and Bonding Theories

    Molecular Geometry and Bonding Theories

    Molecular Geometry and Bonding Theories. Chapter 9. Molecular ShapesSection 9.1. Lewis structures only provide a 2-D representation of a molecule ... Chemical bonds are simply represented with lines in a Lewis structure; however covalent bonds are formed from the overlap...