Thread: Working with clock(); getting unexpected result

Hello everyone,

I have written a program to answer Problem 54 on Project Euler and I have the code working successfully, but the problem I'm having is related to measuring how long it takes the program to run.

I am working on Fedora 12 running on a Virtual PC. The result I am currently getting is:

Code:

time lapsed: 0.000000

I am expecting a different result as it has to take 'some' time for it to process. Any help would be appreciated. Below is my code:

Code:

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

//defines listed here
//function prototypes listed here

int main(void) {
	char hand1[11] = {0}, hand2[11] = {0};
	char *filename = "poker.txt";
	FILE *fin;
	int ch = 0, counter = 0, i = 0, j = 0;
	static int p1_wins = 0, p2_wins = 0;
	clock_t start, end;
	double lapsed = 0.0;	

	fin = fopen(filename, "r");
	if (fin == NULL) {
		fprintf(stderr, "Error: unable to open %s for reading.\n", filename);
		exit (1);
	}	

	start = clock();

	while ((ch = fgetc(fin)) != EOF) {
		if ((ch == ' ') || (ch == '\r'))
			continue;

		if (counter < 10) {
			hand1[i] = ch;	
			i++;	
			counter++;
		}		

		else if (counter >= 10) {
			if (ch == '\n') {
				counter = 0;
				i = 0;
				j = 0;
				determine_winner(hand1, hand2, &p1_wins, &p2_wins);
			}

			else {
				hand2[j] = ch;
				j++;
			}
		}
	}	

	printf("p1_wins: %d\n", p1_wins);
	printf("p2_wins: %d\n", p2_wins);	

	end = clock();

	lapsed = (double)(end - start)/CLOCKS_PER_SEC;	

	printf("time lapsed: %f\n", lapsed);

	return 0;
}

I think your program will be IO bound.
Try to change fin to no buffer mode.

Code:

setvbuf(fin,NULL,_IONBF,0);

See any difference?

TY for the help Bayint Naung, but unfortunately when using setvbuf() I am still getting the same result. I placed the function after the stream has been opened but before any operation has been done on the stream.

I created a small program just to test if my computer was the issue, and it is not. I got 0.217000 when running the code below:

Code:

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

int main(void) {
	int i = 0;
	clock_t start, end;
	double lapsed = 0.0;
	
	start = clock();
	
	for (i=0; i<10000; i++)
		printf("i: %d\n", i);
		
	end = clock();
	
	lapsed = (double)(end - start)/CLOCKS_PER_SEC;
	
	printf("lapsed: %f\n", lapsed);
	
	return 0;
}

How large is your input poker.txt file ?
Is determine_winner() function doing anything that could take time??

Here are both of the files:

Matrix,
Try putting your little timewaster loop into your program just before the "end" statement.
You know it takes measurable time, so if your program still reports 0, there is a problem.
Otherwise, your code is too fast. (I never thought I'd write that sentence.)
Gil

TY for the reply gil,

There must be a problem. I did as you suggested and I'm still arriving at 0.000000. I even changed my printf() statement to show twenty 0's after the decimal point:

Code:

printf("time lapsed: %.20f\n", lapsed);

and I'm still arriving at the same answer. Unfortunately, I don't know that much about the proper way to use clock(), hence my original question. I hope someone will be able to shed some light on this interesting issue

If you set your print statement to output beyond the range of the data type, you're just telling the printf() to print nonsense digits -- no relation to anything factual.

Clock() comes from the PIC chip, and only gets 1/3rd of that, so it's resolution is limited. Plus, your OS is multi-tasking (unlike DOS or a real-time OS), and so each request is put into a prioritized list.

Your motherboard and Linux may have higher resolution timer chips and code for them, so check them out. There are at least two different types - one was a "multimedia timer" solution and another one involved a chip from Intel. The latter is a good one. The "multimedia timer" was a mess.

Google "high precision Intel timer" and see what you can find. I haven't seen the code and/or driver needed, for this, yet.

WOW! Excellent call Adak!

I took your advise and before doing in-depth research on my mobo and processor I copied the program over to my Intel Atom Netbook......and the function worked perfectly! I got an answer of 0.031000.

The machine I was originally running the program from has a AMD Athlon 64 X2 6000+ Windsor 3.0GHz with a GIGABYTE GA-M61P-S3 mobo....

I'll start googling HPET with my hardware specs to see if I can find any info on getting this working on a AMD machine....thanks again Adak!

Rats - I see your code doesn't show the parentheses, but I saw them when I copied the code to a text editor.
You might try copying the line from your test program over the line in the main program.
Gil

nice catch, gil_johnson.
Why don't you post the exact code that caused problem? matrix333???!!!!!!!

matrix,
I'm sorry - I am away from home, where I worked on your program, and my mouse here has died.
I am getting around by keyboard, but I can't figure out how to select the text in your code section.
What I did was to highlight the code in your original post by clicking and dragging over the whole program, hitting Ctl-C, and pasting into mousepad, a text editor.
I'll post again later today if you can't reproduce the result I saw.
Gil