BigSim - PowerPoint PPT Presentation

Loading...

PPT – BigSim PowerPoint presentation | free to download - id: 14016c-ZTcwN



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

BigSim

Description:

Performance prediction for a large varieties of extremely large parallel machines ... Back patching. Network simulator. Simulation can be in separate resolutions. 8. 9 ... – PowerPoint PPT presentation

Number of Views:34
Avg rating:3.0/5.0
Slides: 21
Provided by: KALE2
Learn more at: http://charm.cs.uiuc.edu
Category:
Tags: bigsim | patching

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: BigSim


1
BigSim
  • Presented by
  • Yogesh Mehta (ymehta_at_uiuc.edu)
  • and Gengbin Zheng (gzheng_at_uiuc.edu)
  • Parallel Programming Laboratory
  • Department of Computer Science
  • University of Illinois at Urbana-Champaign
  • http//charm.cs.uiuc.edu

2
Overview
  • BigSim
  • Component based, integrated simulation framework
  • Performance prediction for a large varieties of
    extremely large parallel machines
  • Study alternate programming models

3
Our approach
  • Applications based on existing parallel languages
  • MPI
  • Charm
  • Adaptive MPI
  • Facilitate development of new programming
    languages
  • Detailed/accurate simulation of parallel
    performance
  • Sequential part performance counters,
    instruction level simulation
  • Parallel part simple latency based network
    model, network simulator

4
Parallel Simulator
  • Parallel performance is hard to model
  • Communication subsystem
  • Out of order messages
  • Communication/computation overlap
  • Event dependencies, causality.
  • Parallel Discrete Event Simulation
  • Emulation program executes concurrently with
    event time stamp correction.
  • Exploit inherent determinacy of application

5
Architecture of BigSim Simulator
6
Emulation on a Parallel Machine
7
Emulator to Simulator
  • Predicting time of sequential code
  • User supplied estimated elapsed time
  • Wallclock measurement time on simulating machine
    with suitable multiplier
  • Performance counters
  • Hardware simulator
  • Predicting messaging performance
  • No contention modeling, latency based
  • Back patching
  • Network simulator
  • Simulation can be in separate resolutions

8
(No Transcript)
9
Simulation of different applications
  • Linear-order applications
  • No wildcard receives
  • No timestamp correction necessary
  • Reactive applications
  • Message driven objects
  • Methods execute as corresponding messages arrive
  • Multi-dependent applications
  • Irecvs with WaitAll
  • Uses of structured dagger to capture dependency

10
Timestamp correction
  • Needed for out-of-order message delivery
  • Two messages are not executed in the order of
    their timestamps
  • Need to capture event dependency
  • Use structured dagger
  • Construct backward and forward dependents

11
Post-mortem Timestamp Correction
  • Simulate execution again using POSE, a parallel
    discrete event simulation environment
  • Read log files containing messages and tasks and
    their dependency relationships
  • Simulate their execution in correct order to
    generate revised timestamps for each task

12
Big Network Simulation
  • Simulate network behavior packetization,
    routing, contention, etc.
  • Incorporate with post-mortem timestamp
    correction
  • Switches are connected in torus network

13
BigNetSim Network Model
  • Optimal available path is locked
  • Message is packetized and sent
  • Path is freed after all packets are received

14
Projections Performance visualization
15
Simulation Process
  • Compile MPI or Charm program and link with
    simulator library
  • Online mode simulation
  • Run the program with bgcorrect
  • Visualize the performance data in Projections
  • Postmortem mode simulation
  • Run the program with bglog
  • Run POSE based simulator with network simulation
    on different number of processors
  • Visualize the performance data

16
Projections before/after correction
17
Validation
18
LeanMD Performance Analysis
  • Benchmark 3-away ER-GRE
  • 36573 atoms
  • 1.6 million objects
  • 8 step simulation
  • 64k BG processors
  • Running on PSC Lemieux

19
Predicted LeanMD speedup
20
Other Research Areas
  • Programming environment
  • Scalable load balancing
  • Parallel algorithms on 128K processors
  • FFT,
  • Fault Tolerance
  • Applications
  • Molecular dynamics (LeanMD)
  • Quantum Chemistry (CPAIMD)
  • Finite Element Framework (FEM)
About PowerShow.com