Helsingin yliopisto Tietojenkäsittelytieteen laitos
 

Tietojenkäsittelytieteen laitos

Tietoa laitoksesta:

 
Homework   [ Kerola | PerfEv home (Kerola) | Descr | PerfEv home (dept) | Exams | Spring 02 | Lectures | Homeworks ]

Perf. Eval., Homework 1 (due 13.3.2002)

  1. Data base query times are normally distributed with average 4 seconds (4 s) and with standard deviation 1 s.
    1. What is the probability that query time is more than 8 seconds?
    2. What is the probability that query time is at most 6 seconds?
    3. What is the probability that query time is 2-6 seconds?
    4. What is the 95% percentile for the query time?

  2. You have made a sample to measure some unknown value. Is it better to have
    1. large mean or small mean? Why?
    2. large variance or small variance? Why?
    3. large confidence interval or small confidence interval? Why?
    4. large 95% percentile than small 95% percentile? Why?

  3. What is the difference between
    1. mean and sample mean?
    2. mean and confidence interval for mean?
    3. sample mean and confidence interval for mean?
    4. deviation and sample deviation?
    5. 90% confidence interval and 95% percentile

  4. Is it possible that
    1. The system is saturated, but no device is saturated?
    2. Some devices are saturated, but the system is not?
    Explain, or give an example in if the answer is yes.

  5. [2 htp] Assume that we want to evaluate a new cell phone for its performance characteristics. Concentrate on the hand-held unit.
    1. What is the overall system?  What are the inputs? What are the outputs?
    2. What is the workload?  Does it consist of different types of tasks?
    3. What are the performance metrics?  When is the system considered saturated?
    4. Describe sitations when the performance is acceptable/unacceptable.
    5. What about future workload?  How will it change?
    6. Give examples on how performance evaluation should be used at cell phone manufacturer corporate level plans? (I.e., what are the real questions that are solved with performance evaluation)

Teemu Kerola