LabWindows/CVI
timed loop compared in CVI and labview
Hi all,
I have read few pots online regarding the timing control and CPU usage in CVI and labview. I am trying to compare it in my computer since timing is pretty critical in my application. But I don't have real time system so I would like to know how much difference will be by using labview and/or CVI. My testing is pretty simple, start a loop to lopp1000 time in which loop I will pause for 20 millisecond, I try the following labview code
I repeat running this for many times and it will roughly return 20500 millisecond while we expect 20000 millisecond.
Now in CVI, I use the following code
int CVICALLBACK BTNCallBack(int panel, int control, int event, void *callbackData, int eventData1, int eventData2)
{
if (event==EVENT_COMMIT)
{
clock_t start = clock();
for (int i=0; i<2000; i++)
{
Sleep(20);
}
clock_t end = clock();
char str[128];
sprintf(str, "time elapsed %.10f", (double)(end-start)/(double)CLOCKS_PER_SEC);
MessagePopup("time elapsed", str);
}
return 0;
}
This code will return about 41.443 seconds elapsed instead of 20seconds!!! I also try to use SyncWait but it doesn't help much
int CVICALLBACK testcb(int panel, int control, int event, void *callbackData, int eventData1, int eventData2)
{
if (event==EVENT_COMMIT)
{
double mark, interval=20/1000.0;
clock_t start = clock();
for (int i=0; i<2000; i++)
{
mark = Timer();
SyncWait (mark, interval);
}
clock_t end = clock();
char str[128];
sprintf(str, "time elapsed %.10f", (double)(end-start)/(double)CLOCKS_PER_SEC);
MessagePopup("time elapsed", str);
}
return 0;
}
it returns about 42.005 seconds. So why is that? I know it is not possible to control the timing better than millisecond, but why labview will get better result in timing issue but CVI not. How can precisely control timing in CVI (at least the same performance observed in lbaivew)?
Hello,
I think everyone is missing the problem. 
It took me a few seconds to see it as well...
Your CVI code is running the loop 2000 times instead of 1000 times:
for (int i=0; i<2000; i++)
{
Sleep(20);
}
I hope that helps!
Jeff
@Jeff_P. wrote:
Hello,
I think everyone is missing the problem.
Your CVI code is running the loop 2000 times instead of 1000 times:
for (int i=0; i<2000; i++) { Sleep(20); }
I hope that helps!
Jeff
Thanks Jeff. I admit that it is my mistake ... well actually the very first code I wrote for testing was correct but before I post it, I modify the loop time for testing. Let me restate my question here.
At the very beginning, I am trying to compare how close the timed loop control in labview and CVI by delay a loop (1000) with timer delay of 5 millisecond (not 20). I have my labview code posted as before but just chhanged the timer to 5 millisecond. I got the total time elapsed about 5100 ms (roughly 100 ms more spent)
In CVI, here is the correct code I used
double mark, interval=5/1000.0;
SetSleepPolicy(VAL_SLEEP_NONE);
clock_t start = clock();
for (int i=0; i<1000; i++)
{
Sleep(5);
//mark = Timer();
//SyncWait (mark, interval);
}
clock_t end = clock();
char str[128];
sprintf(str, "time elapsed %.10f millisecond", 1000.0*(double)(end-start)/(double)CLOCKS_PER_SEC);
MessagePopup("time elapsed", str);
SetSleepPolicy(VAL_SLEEP_MORE);
The actual time elapsed it about 5900 millisecond, it is 800ms more than the labview case. I try to change the SleepPolicy to SLEEP MORE, SLEEP SOME and NONE SLEEP mode, about the same. I wonder if using clock() method to estimate time is not a good way or not.
I also post my project file here. The labview code as well as the CVI are running in 2013 version. Thanks.
