Rice University
COMP 528
Computer Systems Performance Analysis
Lecture Notes and Handouts

This page contains copies of class lecture notes (if they exist in electronic form) as well as copies of any handouts provided in class.
  1. Introductory lecture
    Course Information
  2. Modern Processors & Hardware Support for Performance Measurement
  3. Software Support for Performance Measurement
  4. Workloads and Workload Selection
  5. Workloads Characterization
    Mathematica Notebook on Principal Components Analysis
  6. More Workload Characterization & Basic Probability and Statistics
    Mathematica Notebook on Normal Distribution pdf and cdf
    Matlab simulation of Markov network model 1
    Matlab simulation of Markov network model 2
  7. Summarizing Measured Data
    Mathematica Notebook computing properties of packet bursts
  8. Comparing Systems Using Sample Data
  9. Linear Regression Models
  10. Other Regression Models
    Matlab classroom regression demos: mat.m
    Trivial, but useful Matlab routines: sse.m, sst.m, zquant.m
  11. Introduction to Experimental Design and Analysis
  12. 2^k Factorial Designs and 2^kr Factorial Designs with Replications
  13. 2^k Factorial Designs and 2^kr Factorial Designs with Replications (continued)
  14. 2^(k-p) Fractional Factorial Designs
  15. One-factor Designs
  16. Two-factor Full Factorial Design without Replications
  17. Two-factor Full Factorial Design with Replications
  18. General Full Factorial Design with k Factors
    Matlab routines for analyzing the Tsao and Margolin data. The data: logdata. Utilities: twofour.m (map from 2D table in Jain to 4D matrix), fourtwo.m (map from 4D matrix back to 2D table), matmean.m (mean of an n-D matrix), elems.m (number of elements in an n-D matrix), grandmean.m. Main effects: effects.m. Interactions: interact1.m, interact2.m, interact3.m. Model: model.m. Analysis: ssq.m, sumofsquares.m, errors.m.
  19. Introduction to Simulation
  20. Analyzing Simulation Results
  21. Random Number Generation
  22. Testing Random Number Generators
    Matlab routines for generating plots in lecture. Generic LCG: lcg.m. Visual test of an LCG using overlapping pairs: lcg2d.m. 2-distributivity test of Matlab rand: rand3d.m. Tausworthe RNG: tausworthe.m. 2D analysis of Tausworthe generator using overlapping pairs: tw2d.m. 3D analysis of Tausworth generator using non-overlapping pairs: tw3d.m. Helper function for Tausworthe generator: xor_reduce.m.
  23. Generating Random Variates
  24. Models for Understanding Parallel Performance