My UW | UW Search

   
   

Computer Science Home Page
> ~markhill

 

 

Home

Contact Info

Lecture & Notes

Schedule

Text, Reader, Sources

Computing

Grading

Project

Examinations

Homework

Reviews

Miscellanea

UW-Madison
Computer Sciences Dept.

CS/ECE 752 Advanced Computer Architecture I Fall 2004 Section 1
Instructor Mark D. Hill and T. A. Andrey Litvin
URL: http://www.cs.wisc.edu/~markhill/cs752/Fall2004/

Homework 2 // Due at Lecture Tue October 5

Problem 1 (6 points)

H&P Problem 3.1.

Problem 2 (6 points)

A global GAg predictor stores the outcomes of n last executed branches in a history register (HR). A local PAg predictor stores n last outcomes of a particular branch in an entry of history register table (HRT), indexed by the branch address. In the second stage, both of these predictors use the history bits to index into the table of 2-bit automata to obtain the actual prediction.

a)Give an example program fragment in C where the global predictor would achieve greater prediction accuracy than the local predictor.
b)Vice versa: Give an example where the local predictor would do significantly better.
c)Both GAg and PAg predictors only use history bits to select the predictor automaton. What are the pros and cons of introducing some branch address bits in the indexing process?

Problem 3 (8 points)

    L.D    F2, 16(R6)
    MUL.D  F4, F2, F10
    ADD.D  F4, F4, F12
    S.D    F4, 32(R5)
    ADD.D  F4, F2, F8
    S.D    F4, 128(R5)

Using Tomasulo's Algorithm, show what the status tables (Instruction status, Reservation Stations, and Register Status tables as in figure 3.3) look like right after the second ADD.D has entered the "Execute" phase. Use the hardware configuration from Figure 3.2 and the FP latencies from Figure 3.63

Problem 4 (10 points)

(This problem is important well beyond the points we assign to it, because it begins our gentle ramp for teaching you about simulators, in general, and SimpleScalar, in particular.)

Using a CS Unix/Linux machine, download and install the Simplescalar v 3.0 Source code from http://www.simplescalar.com/. Descriptions for the Integer Benchmarks and Floating Point Benchmarks can be found at the Standard Performance Evaluation Corporation's website. To run the benchmarks, the command line arguments for each benchmark must be specified after the executables name. The command line options necessary for each benchmark are at /unsup/spec2000/benchspec/benchmark.cmdlines. From there, the associated Alpha executables and data files for the integer benchmarks and floating point benchmarks are in the CINT2000 and CFP2000 directories respectively.

a) The benchmarks have initialization phases at the beginning of their execution that run only once. Researchers tend to avoid collecting data from this region because it is not representative of the entire program. Choose a benchmark and find the likely end of its initialization period by observing its IPC in sim-outorder while it executes. Describe which benchmark you chose, your methodology, and your results. You might consider fast-forwarding to a certain instruction and limiting the number of instructions run at any one time. Assume the initialization period to be a factor of at least 10 million.

b) Using the performance of gcc in sim-outorder with default settings as a base, calculate the speedup(slowdown) if:

  • the simulator is run in in-order mode
  • the D1 cache replacemnent policy is changed to FIFO
  • the branch predictor is changed to "always taken"

    For all of the above tests, fast-forward 100M instructions, and simulate for next 50M instructions. Use cc1_base executable and integrate.i input file for gcc.

    •  
    Computer Sciences | UW Home

    Page last modified: Friday, 01-Oct-2004 01:35:05 CDT
    Feedback or content questions: send email to Email Address of Mark Hill
    Technical or accessibility issues: lab@cs.wisc.edu
    Copyright © 2002, 2003 The Board of Regents of the University of Wisconsin System.