Performance Optimization for the Origin 2000

9/9/98

Click here to start

Table of Contents

Performance Optimization for the Origin 2000

SGI Optimization Tutorial

Getting Started

Getting Started (Continued)

Version 1 of Mpifft

Version 2 of Mpifft

Version 3 of Mpifft

Speedshop Tools

Speedshop Experiment Types

Speedshop Experiment Types

Using Speedshop Tools for Performance Analysis

Using ssusage on Your Program

ssusage (Continued)

Using ssrun on Your Program

Choosing an Experiment Type

Running Experiments on a MPI Program

Experiments That We Will Run for the Tutorial.

usertime (Continued)

fpcsamp (Continued)

fpcsamp (Continued)

ideal (Continued)

ideal (Continued)

Secondary Data Cache Misses (dsc_hwc)

Secondary Data Cache Misses (dsc_hwc) (Continued)

Secondary Data Cache Misses (dsc_hwc)

Translation Lookaside Buffer Misses (tlb_hwc)

Translation Lookaside Buffer Misses (tlb_hwc)

SGI Optimization Tutorial

More Info on Tools

Getting the new files

Using nupshot (continued)

Using nupshot (continued)

Running the example

Running the Example (continued)

Running the Example (continued)

Vampir Tutorial (Getting Started)

Vampir Tutorial Creating a logfile

Vampir Tutorial Viewing the Logfile

Vampir Startup Screen

Vampir Timeline Display

Using Vampir viewing statistics for selected portion of the timeline.

Vampir Global Activity Chat

Vampir, additional info on messages

Identifying a message in Vampir

Identifying messages in Vampir

Matrix-Matrix Multiply Demo

Matrix-Matrix Multiplication - Simple Optimization by Cache Reuse

Example of the loop

Instructions for the exercise

Exercise 1 (continued)

Exercise 2 Matrix-Matrix Multiplication Optimization using Blocking and Unrolling of Loops

Author: Kevin London

Email: london@cs.utk.edu

Home Page: http://www.cs.utk.edu/~london

Author: Philip J. Mucci

Email: mucci@cs.utk.edu

Home Page: http://www.cs.utk.edu/~mucci

Download presentation source

Download presentation source