Mark: 3 revision(s)

2008-11-03T05:18:43Z

3 revision(s)

New page

Ok so the task is to report back on "how long it takes to execute an addition (+) operation in C?"

Hints have suggested looping, using the '''clock()''' function and using the Linux server.

So Step 1: I'm going to try and remember how to write a C script to loop an addition.

After trying quite hard and referring back to old notes, i kind of remembered how to write a C script again. (YAY!!) so here's what i did the first time.

#include <stdio.h>
#include <time.h>

int main(void)
{
int answer = 0;
long time;

time=clock();

while (answer<10000000){
answer = answer+1;
}
time=clock()-time;

printf("The time taken to reach a million is %ld milliseconds \n", time);
}

Hence used a While loop. I ran it 5 times.
Here are the results in order in milliseconds: '''2, 3, 2, 2, 3'''

Not so bad I think. I'm gonna try and use a larger number to reach as 2 or 3 milliseconds is not enough to measure accurately.

This time I'm timing how long it takes for C to reach a billion using simple addition.
Here are the resultsin milli seconds and in order: '''2195, 2303, 2217, 2296, 2207'''

So a billion is a good number to count upto.

Now I'm going to try out the '''long''' data type.
Here are the results in milliseconds: '''2194, 2178, 2167, 2304, 2188'''

Now I'm trying out the '''short''' data type.
Here are the results in milliseconds: ''' NOTHING!!''' This must be because short data type ends at 32767 and cant go to a billion.

Now trying out the '''float''' data type.
And again nothing works!

There's something wrong with my code and after consulting John, I've been advised to separate the counter so that the changing of the data types doesn't interfere. Also John has changed my While loop into a For loop. Here's the code:

''#include <stdio.h>
#include <time.h>

int main(void)
{
float answer = 0;
long i;
long time;

time=clock();

for( i = 0; i < 1000000000; i++ ){
answer = answer + 1;
}
time=clock()-time;

printf("The time taken to reach a billion is %ld milliseconds \n", time);
}''

And now the results:
--- using '''int''': '''2198, 2345, 2319, 2144, 2347'''
--- using '''long''': '''2165, 2278, 2144, 2183, 2149'''
--- using '''short''': '''2424, 2447, 2446, 2547, 2532'''
--- using '''float''': '''8251, 8193, 8414, 8210, 8491'''
--- using '''double''': '''8224, 8205, 8253, 8328, 8317'''

Hence, using these results I can conclude that the '''float''' and the '''double''' types take longer to calculate. It also seems that the '''long''' type for some reason is the fastest. So I'm going to test the '''long''' type again to see if it spits out similar results.
Here are the results: '''2168, 2171, 2187, 2214, 2158'''.

So it is true that the '''long''' data type did give the fastest time overall using my code.

Question really is, would using the long data type as much as possible make our programs run any faster?
I think to answer that question correctly, we'd have to consider all the factors that could affect such an experiment.[[User:Rbha033|Rbha033]] 11:49, 4 March 2008 (NZDT)

SE250:lab-1:rbha033 - Revision history

Mark: 3 revision(s)