User Tools

Site Tools


haas:spring2016:cprog:projects:pnc0

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision
Previous revision
haas:spring2016:cprog:projects:pnc0 [2016/02/22 18:14] – [Program] wedgehaas:spring2016:cprog:projects:pnc0 [2016/02/27 13:13] (current) – [Program] wedge
Line 17: Line 17:
   * ability to iterate sections of code (for/while statement)   * ability to iterate sections of code (for/while statement)
  
 +=====Algorithmic Complexity=====
 +A concept in Computer Science curriculum is the notion of computational/algorithmic complexity.
 +
 +Basically, a solution to a problem exists on a spectrum of efficiency (typically constrained by time vs. space): if optimizing for time, the code size tends to grow.
 +
 +Additionally, if optimizing for time (specifically to reduce the amount of time taken), strategic approaches are taken to reduce unnecessary or redundant operations (yet still achieving the desired end results).
 +
 +This project will endeavor to introduce you to the notion that the algorithms and constructs you use in coding your solution can and do make a difference to the overall runtime of your code.
 =====Background===== =====Background=====
 In mathematics, a **prime** number is a value that is only evenly divisible by 1 and itself; it has no other factors. Numbers that have divisibility/factors are known as **composite** numbers. In mathematics, a **prime** number is a value that is only evenly divisible by 1 and itself; it has no other factors. Numbers that have divisibility/factors are known as **composite** numbers.
Line 109: Line 117:
  
 One assumption I will allow you to make for your optimized solution is that the single-digit primes (2, 3, 5, 7) can be assumed prime, and just printed out if having them be calculated would otherwise break your algorithm (might be helpful to some people; I certainly found it useful in some of my solutions). One assumption I will allow you to make for your optimized solution is that the single-digit primes (2, 3, 5, 7) can be assumed prime, and just printed out if having them be calculated would otherwise break your algorithm (might be helpful to some people; I certainly found it useful in some of my solutions).
 +
 +===some optimization ideas===
 +  * [[https://primes.utm.edu/notes/faq/six.html|Prime patterns of 6]]
 +  * [[https://lab46.g7n.org/~wedge/php/web.php?width=800&height=600&div=6&max=145&prime=1&factor=0&spiral=0|Visualization of Primes in a web of 6 divisions]]
 +  * [[https://lab46.g7n.org/~wedge/php/web.php?width=800&height=600&div=12&max=145&prime=1&factor=0&spiral=0|Visualization of Primes in a web of 12 divisions]]
 +  * [[https://en.wikipedia.org/wiki/Ulam_spiral|Ulam spiral]]
 +  * [[https://en.wikipedia.org/wiki/Sieve_of_Eratosthenes|Sieve of Eratosthenes]]
 +
 +Of particular note: the sieve algorithms take advantage of a increased storage space, where others (like brute force) are predominantly time-based. The sieve is also more detailed... even if you don't decide to implement a sieve, take a look and compare the algorithm to what you've done to see the differences in approaches.
 =====Program===== =====Program=====
 It is your task to write 3 separate prime number calculating programs: It is your task to write 3 separate prime number calculating programs:
  
-  - primebrute.c: for your brute force implementation +  - **primebrute.c**: for your brute force implementation 
-  - primesqrt.c: for your square root-optimization of the brute force +  - **primesqrt.c**: for your square root-optimization of the brute force 
-  - primeopt.c: for your optimized solution, be it basing off the existing ones, or taking a different approach+  - **primeopt.c**: for your optimized solution, be it basing off the existing ones, or taking a different approach
  
 Your program should: Your program should:
Line 122: Line 139:
     * these values should be positive integer values; you can make the assumption that the user will always do the right thing.     * these values should be positive integer values; you can make the assumption that the user will always do the right thing.
   * start your stopwatch (see **timing** section below):   * start your stopwatch (see **timing** section below):
-  * perform the correct algorithm against the input+  * perform the algorithm against the value
   * if enabled, display the prime numbers found in the range   * if enabled, display the prime numbers found in the range
   * output the processing run-time to STDERR (do this always).   * output the processing run-time to STDERR (do this always).
Line 128: Line 145:
     * if primes are being displayed, they are space-separated (first prime hugs the left margin), and when all said and done, a newline is issued.     * if primes are being displayed, they are space-separated (first prime hugs the left margin), and when all said and done, a newline is issued.
     * the timing information will be displayed in accordance to code I will provide (in the **timing** section).     * the timing information will be displayed in accordance to code I will provide (in the **timing** section).
 +
 +====Other considerations====
 +All your programs MUST perform the calculations to determine primality- you may not always be printing it out (depending on argv[2]), but work must be done to ensure the value is identified as a prime/composite value.
 +
 +For example:
 +
 +<code>
 +if (show == 1)
 +{
 + work to determine if it is prime
 + if prime
 + print number
 +}
 +</code>
 +
 +will actually skip the core processing, and you’ll see some amazing runtimes as a result. They may be amazing, but they’re not real, because you’re not actually doing anything.
 +
 +What you want instead:
 +
 +<code>
 +work to determine if it is prime
 +if (show == 1)
 +{
 + if prime
 + print number
 +}
 +</code>
 +
 +there are many ways to express the above, through compound if statements and other arrangements, but notice how nothing is holding back “work to determine if it is prime”.
 +
 +That also isn’t to say you can’t avoid doing a work run if you’re able to determine its non-primality with a simple pretest (even value, factor of 3, etc.), but that’s actually considered more of the core “work”, so it is more than okay (and encouraged in the primeopt).
 +=====Command-Line Arguments=====
 +To automate our comparisons, we will be making use of command-line arguments in our programs. As we have yet to really get into arrays, I will provide you same code that you can use that will allow you to utilize them for the purposes of this project.
 +
 +====header files====
 +We don't need any extra header files to use command-line arguments, but we will need an additional header file to use the **atoi(3)** function, which we'll use to quickly turn the command-line parameter into an integer, and that header file is **stdlib.h**, so be sure to include it with the others:
 +
 +<code c>
 +#include <stdio.h>
 +#include <stdlib.h>
 +</code>
 +
 +====setting up main()====
 +To accept (or rather, to gain access) to arguments given to your program at runtime, we need to specify two parameters to the main() function. While the names don't matter, the types do.. I like the traditional **argc** and **argv** names, although it is also common to see them abbreviated as **ac** and **av**.
 +
 +Please declare your main() function as follows:
 +
 +<code c>
 +int main(int argc, char **argv)
 +</code>
 +
 +The arguments are accessible via the argv array, in the order they were specified:
 +
 +  * argv[0]: program invocation (path + program name)
 +  * argv[1]: our maximum / upper bound
 +  * argv[2]: visibility (1 to show primes, 0 to be silent)
 +
 +There are ways to do flexible argument parsing, and even to have dashed options as we have on various commands. But such things are beyond the scope of our current endeavors, so we will stick to this basic functionality for now.
 +
 +====Simple argument checks====
 +Although I'm not going to require extensive argument checking for this project, here's how we would check to see if the minimal number of arguments has been provided:
 +
 +<code c>
 +    if (argc < 3)  // if less than 3 arguments have been provided
 +    {
 +        fprintf(stderr, "Not enough arguments!\n");
 +        exit(1);
 +    }
 +</code>
 +
 +If you're wondering, "why 3? I thought we only had 2.", C includes the program's name as the first argument, so we want program + max + visibility, or 3.
 +
 +====Grab and convert max and visibility====
 +Finally, we need to put the arguments representing the maximum value and visibility settings into variables.
 +
 +I'd recommend declaring two variables of type **int**.
 +
 +We will use the **atoi(3)** function to quickly convert the command-line arguments into **int** values:
 +
 +<code c>
 +    max  = atoi(argv[1]);
 +    show = atoi(argv[2]);
 +</code>
 +
 +And now we can proceed with the rest of our prime implementation.
 +
 +=====Timing=====
 +Often times, when checking the efficiency of a solution, a good measurement (especially for comparison), is to time how long the processing takes.
 +
 +In order to do that in our prime number programs, we are going to use C library functions that obtain the current time, and use it as a stopwatch: we'll grab the time just before starting processing, and then once more when done. The total time will then be the difference between the two (end_time - start_time).
 +
 +We are going to use the **gettimeofday(2)** function to aid us in this, and to use it, we'll need to do the following:
 +
 +====header file====
 +In order to use the **gettimeofday(2)** function in our program, we'll need to include the **sys/time.h** header file, so be sure to add it in with the existing ones:
 +
 +<code c>
 +#include <stdio.h>
 +#include <stdlib.h>
 +#include <sys/time.h>
 +</code>
 +
 +====timeval variables====
 +**gettimeofday(2)** uses a **struct timeval** data type, of which we'll need to declare two variables in our programs (one for storing the starting time, and the other for the ending time).
 +
 +Please declare these with your other variables, up at the top of main() (but still WITHIN main()-- you do not need to declare global variables).
 +
 +<code c>
 +    struct timeval time_start; // starting time
 +    struct timeval time_end;   // ending time
 +</code>
 +
 +====Obtaining the time====
 +To use **gettimeofday(2)**, we merely place it at the point in our code we wish to take the time.
 +
 +For our prime number programs, you'll want to grab the start time **AFTER** you've declared variables and processed arguments, but **JUST BEFORE** starting the driving loop doing the processing.
 +
 +That call will look something like this:
 +
 +<code c>
 +    gettimeofday(&time_start, 0);
 +</code>
 +
 +The ending time should be taken immediately after all processing (and prime number output) is completed, and right before we display the timing information to STDERR:
 +
 +<code c>
 +    gettimeofday(&time_end, 0);
 +</code>
 +
 +====Displaying the runtime====
 +Once we having the starting and ending times, we can display this to STDERR. You'll want this line:
 +
 +<code c>
 +    fprintf(stderr, "%10.6lf\n", time_end.tv_sec - time_start.tv_sec + ((time_end.tv_usec - time_start.tv_usec) / 1000000.0));
 +</code>
 +
 +For clarity sake, that format specifier is "%10.6lf", where the "lf" is "long float", that is **NOT** a number one but a lowercase letter 'ell'.
 +
 +And with that, we can compute an approximate run-time of our programs. The timing won't necessarily be accurate down to that level of precision, but it will be informative enough for our purposes.
 =====Execution===== =====Execution=====
 Several operating behaviors are shown as examples. Several operating behaviors are shown as examples.
Line 173: Line 329:
 The execution of the programs is short and simple- grab the parameters, do the processing, produce the output, and then terminate. The execution of the programs is short and simple- grab the parameters, do the processing, produce the output, and then terminate.
  
 +=====Check Results=====
 +If you'd like to compare your implementations, I rigged up a script called **primerun** which you can run.
 +
 +In order to work, you **MUST** be in the directory where your **primebrute**, **primesqrt**, and **primeopt** binaries reside, and they must be named as such.
 +
 +For instance (running on my implementations):
 +
 +<cli>
 +lab46:~/src/cprog/pnc0$ primerun
 +============================================
 +   range       brute        sqrt         opt 
 +============================================
 +          0.000002    0.000002    0.000002  
 +      16    0.000002    0.000002    0.000002  
 +      32    0.000003    0.000004    0.000002  
 +      64    0.000005    0.000020    0.000003  
 +     128    0.000012    0.000023    0.000003  
 +     256    0.000037    0.000029    0.000006  
 +     512    0.000165    0.000036    0.000014  
 +    1024    0.000540    0.000080    0.000033  
 +    2048    0.001761    0.000187    0.000078  
 +    4096    0.006115    0.000438    0.000189  
 +    8192    0.021259    0.001036    0.000458  
 +   16384    0.077184    0.002520    0.001153  
 +   32768    0.281958    0.006156    0.002826  
 +   65536    1.046501    0.015234    0.007135  
 +  131072    5.160141    0.045482    0.021810  
 +  262144    --------    0.119042    0.057520  
 +  524288    --------    0.301531    0.146561  
 + 1048576    --------    0.758027    0.370700  
 + 2097152    --------    1.921014    0.943986  
 + 4194304    --------    4.914725    2.423202  
 + 8388608    --------    --------    --------  
 +============================================
 + verify:       OK          OK          OK     
 +============================================
 +lab46:~/src/cprog/pnc0$ 
 +</cli>
 +
 +For evaluation, each test is run 4 times, and the resulting time is averaged. During development, I have it set to only run each test once.
 +
 +If the runtime of a particular prime variant exceeds an upper threshold (likely to be set at 2 seconds), it will be omitted from further tests, and a series of dashes will instead appear in the output.
 +
 +If you don't feel like waiting, simply hit **CTRL-c** and the script will terminate.
 +
 +In the example output above, my **primeopt** is playing with an implementation of the **6a+/-1** algorithm.
 +
 +I also include a validation check- to ensure your prime programs are actually producing the correct list of prime numbers. If the check is successful, you will see "OK" displayed beneath in the appropriate column; if unsuccessful, you will be "MISMATCH".
 +
 +If you'd like to experiment with other variations, the script also recognizes prime variants of the following names:
 +  * primeopt0 (for an additional optimization)
 +  * primeopt1 (and another)
 +  * primeopt2 (if you'd like another entry for another optimization)
 +  * primeopt3 (for yet another optimization)
 +  * primeopt4 (and one more; hey, I want you to have nice things)
 =====Bonus Points===== =====Bonus Points=====
 There will be an additional bonus point opportunity with this project, based on processing run-time of your optimized solution. There will be an additional bonus point opportunity with this project, based on processing run-time of your optimized solution.
haas/spring2016/cprog/projects/pnc0.1456164865.txt.gz · Last modified: 2016/02/22 18:14 by wedge