2. TAs Fardad Jalili: fardad@ufl.edu Aisharjya Sarkar: sarkar@cise.ufl.edu Soham Das: soham@ufl.edu

3. Performance Measurement

4. Performance Analysis Paper and pencil. Don’t need a working computer program or even a computer.

5. Some Uses Of Performance Analysis determine practicality of algorithm predict run time on large instance compare 2 algorithms that have different asymptotic complexity  e.g., O(n) and O(n 2 )

6. Reminder

7. Limitations of Analysis Doesn’t account for constant factors. but constant factor may dominate 1000n vs n 2 and we are interested only in n < 1000

8. Limitations of Analysis Modern computers have a hierarchical memory organization with different access time for memory at different levels of the hierarchy.

9. Memory Hierarchy R L1 L2 MAIN ALU 8-3232KB~512KB~512MB 1C2C10C100C

10. Limitations of Analysis Our analysis doesn’t account for this difference in memory access times. Programs that do more work may take less time than those that do less work.

11. Performance Measurement Measure actual time on an actual computer. What do we need?

12. Performance Measurement Needs l programming language le.g. C++/Java l working program lInsertion Sort l computer l compiler and options to use gcc –O2

13. Performance Measurement Needs l data to use for measurement worst-case data best-case data average-case data l timing mechanism --- clock

14. Timing In C++ double clocksPerMillis = double(CLOCKS_PER_SEC) / 1000; // clock ticks per millisecond clock_t startTime = clock(); // code to be timed comes here double elapsedMillis = (clock() – startTime) / clocksPerMillis; // elapsed time in milliseconds

15. Shortcoming Clock accuracy assume 100 ticks Repeat work many times to bring total time to be >= 1000 ticks

16. More Accurate Timing clock_t startTime = clock(); long numberOfRepetitions; do { numberOfRepetitions++; doSomething(); } while (clock() - startTime < 1000) double elapsedMillis = (clock()- startTime) / clocksPerMillis; double timeForCode = elapsedMillis/numberOfRepetitions;

17. Bad Way To Time do { counter++; startTime = clock(); doSomething(); elapsedTime += clock() - startTime; } while (elapsedTime < 1000)

18. Accuracy Now accuracy is 10%. first reading may be just about to change to startTime + 100 second reading (final value of clock())may have just changed to finishTime so finishTime - startTime is off by 100 ticks

19. Accuracy first reading may have just changed to startTime second reading may be about to change to finishTime + 100 so finishTime - startTime is off by 100 ticks

20. Accuracy Examining remaining cases, we get trueElapsedTime = finishTime - startTime +- 100 ticks To ensure 10% accuracy, require elapsedTime = finishTime – startTime >= 1000 ticks

21. Timing in Java s Same semantics, slightly different syntax long startTime = System.nanoTime(); //Code to be measured System.out.println( "Elapsed time: " + (System.nanoTime() - startTime)); or Use micro-benmarking frameworks such as JMH

22. What Went Wrong? clock_t startTime = clock(); long numberOfRepetitions; do { numberOfRepetitions++; insertionSort(a, n); } while (clock() - startTime < 1000) double elapsedMillis = (clock()- startTime) / clocksPerMillis; double timeForCode = elapsedMillis/numberOfRepetitions;

23. The Fix clock_t startTime = clock(); long numberOfRepetitions; do { numberOfRepetitions++; // put code to initialize a[] here insertionSort(a, n) } while (clock() - startTime < 1000) double elapsedMillis = (clock()- startTime) / clocksPerMillis;

24. Time Shared System UNIX time MyProgram

25. To-do s 1. C++: –http://www.cplusplus.com/reference/ctime/clock/http://www.cplusplus.com/reference/ctime/clock/ s 1’.Java (JMH): –http://www.oracle.com/technetwork/articles/java/ architect-benchmarking-2266277.htmlhttp://www.oracle.com/technetwork/articles/java/ architect-benchmarking-2266277.html –http://nitschinger.at/Using-JMH-for-Java- Microbenchmarking/http://nitschinger.at/Using-JMH-for-Java- Microbenchmarking/ s 2. Linux: –http://stackoverflow.com/a/556411http://stackoverflow.com/a/556411