The UNIX and Linux Forums  

Go Back   The UNIX and Linux Forums > Top Forums > High Level Programming
.
google unix.com



High Level Programming Post questions about C, C++, Java, SQL, and other programming languages here.

More UNIX and Linux Forum Topics You Might Find Helpful
Thread Thread Starter Forum Replies Last Post
byte swapping 32-bit float and weird od results GoDonkeys UNIX for Advanced & Expert Users 2 05-20-2009 08:54 PM
Help with __builtin_prefetch function and it's timing Tavo High Level Programming 3 11-25-2008 05:31 AM
timing your functions bebop1111116 High Level Programming 3 11-01-2006 04:19 PM
Timing out a SSH rcunn87 Shell Programming and Scripting 9 07-31-2006 06:11 PM
scp timing out jph UNIX for Advanced & Expert Users 1 06-09-2005 11:59 PM

 
English Japanese Spanish French German Portuguese Italian Dutch Swedish Russian Norwegian Hungarian Hebrew Danish Bulgarian Greek Powered by Powered by Google
 
LinkBack Thread Tools Search this Thread Rate Thread Display Modes
Prev Previous Post   Next Post Next
  #1 (permalink)  
Old 09-07-2009
CRGreathouse CRGreathouse is offline
Registered User
  
 

Join Date: Mar 2009
Posts: 105
Weird timing results in C

I was running some timings in my code to see which of several functions was the best and I've been getting some odd results. Here's the code I'm using:

Code:
static double time_loop(int (*foo)(int)) {
	clock_t start, end;
	int n = 0, i = 0;
	start = clock();
	for (; i <= MAXN; i++)
		if ((*foo)(i)) n++;
	end = clock();
	if (n != 50847534)
		return -1;	// Error
	return ((double) (end - start)) / CLOCKS_PER_SEC;
}

static double median3(int (*foo)(int)) {
	double t1 = time_loop(foo);
	double t2 = time_loop(foo);
	double t3 = time_loop(foo);
	printf("  ((%f %f %f))\n", t1, t2, t3);
	int errors = 0;
	if (t1 < 0)
		errors++;
	if (t2 < 0)
		errors++;
	if (t3 < 0)
		errors++;
	if (errors > 0)
		return -errors;
	if (t1 > t2) {
		if (t2 > t3)
			return t2;
		return t3 < t1 ? t3 : t1;
	} else {
		if (t1 > t3)
			return t1;
		return t3 < t2 ? t3 : t2;
	}
}

The functions that I'm passing to time_loop are pure functions -- they're not making any changes to global variables. And yet every time I run this, my debugging line in median3 shows that the third invocation takes significantly longer than the others. Sample run:

Code:
  ((4.560000 3.750000 6.540000))
Original: 4.560000 s
  ((4.920000 3.760000 6.690000))
Original: 4.920000 s
  ((4.190000 3.750000 6.620000))
Original: 4.190000 s
  ((3.820000 3.130000 5.640000))
New:      3.820000 s
  ((3.580000 3.150000 5.580000))
New:      3.580000 s
	  ((3.670000 3.140000 5.510000))
New:      3.670000 s

Why is that? These functions are entirely deterministic, and nothing is changing between invocations. But I see a strong pattern of time - smaller time - much longer time.

As for cache effects, the functions rely on a 125 MB global array, so that's not going to fit even in my 6MB L3. (Even if it did you'd expect the second and third invocations to be fast.)
 

Bookmarks

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On




All times are GMT -4. The time now is 05:09 AM.


Powered by: vBulletin, Copyright ©2000 - 2006, Jelsoft Enterprises Limited. Language Translations Powered by .
vBCredits v1.4 Copyright ©2007 - 2008, PixelFX Studios
The UNIX and Linux Forums Content Copyright ©1993-2009. All Rights Reserved.Ad Management by RedTyger

Content Relevant URLs by vBSEO 3.2.0