Why Ref. operations are taking so much time

[Tushar_Jambhekar]

In my program i want to execute some commands. for executing command i do following things

1.search a vi with perticular name in certain folder.

2.open ref. to selected vi

3. Pass the args data into cmdArgs string control in the vi

4. Set the error status of the vi

5. Run the VI

6. Read the Error status of the vi

7. Read the data in "Data" indicator in the vi

i am attaching the block diagram of the perticular vi.

but when call the dynamic vi it takes excessively large time here is what i found after further analysis

1. Open ref takes around 700 msec

2. ctr val set for cmd args takes around 550 msec

3. ctr val set for error in takes hopeless 2.5 SECONDS

4. RunVI takes 300 msec

5. both ctr val get takes less than 1 mSec

and close ref takes 700msec.

now if just i open the VI which is called dynamically by double clicking on it and if i run my code it takes hardly 150msecs.

i cant figure out why its happening so

1. file path is correct

Controls and indicators which are used for ctr val Get\Set are existing without any spelling mistake or without any extra whiterspace chars

[altenbach]

Just a guess, but could it be that your VI was last saved in a previous version of LabVIEW? In this case it would need to be converted every time you call it, a slow process.

Make sure all you VIs are saved in the current LabVIEW version. You could even masscompile the entire folder containing all your subVIs.

Message Edited by altenbach on 09-05-2006 09:06 AM

[Tushar_Jambhekar]

This is surely not the case.

I am using LabVIEW 8.0.1, and the entire code has been developed from scratch in LabVIEW 8.0.1 itself.

[Tushar_Jambhekar]

just now i conducted one more experiment and here are the results

1. I created one simple VI and Called it dynamically (Entire VI was Written in LabVIEW 7.1) i found that overhead of dynamic calling is around 2 msec

2. then i wrote (NOT CONVERTED) same code in LabVIEW 8 and i found that the overhead of the dynamic calling is around 12 Msec.

3. I Converted the 7.1 code into 8.0 and saved it to find that it takes around 12 msec as an overhead

please find attached sample code on which i took these trials. it is in LV7.1

some more things i observed are as follows

1. If i increase the Execution time of VI being called dynamically (by increasing the iterations) then overhead also increases (try out 10000, 100000, 1000000 etc.) this seems funny to me, since i expect the overhead must be due to opeining of ref. passing the values and closing the vi ref.

Message Edited by Tushar Jambhekar on 09-10-2006 02:36 AM

[chilly_charly]

The apparent increase in overhead time is due to the passing of the array. Increasing the iteration number also increases the size of the data chunk to be transfered. Additionnally, the array indicator should have been placed after the second timer, outside the sequence structure, otherwise you also measure the display time.

If you use a fixed size array in the test.vi, then the overhead time stays the same, whatever the iteration number.

Sorry for the two stars, that's just the result of an unintentionnal drop-through click 😞

Message Edité par chilly charly le 09-10-2006 11:41 AM

[Tushar_Jambhekar]

thanks. certainly now i know one more point to take care of while calling vis dynamically

but why there is so much of a difference in Version 7.1 and 8.0

i observed 7.1 takes 1/6 the time taken by 8.0

and in the actual VIs that i itend to call is taking the overhead of the order of hundreads of miliseconds, the data passed is certainly not that huge.

is there any way through which i can reduce this overhead?

[chilly_charly]

I don't see such differences between 7.1 and 8.2. May be you could post a vi that shows the problem ?

[Tushar_Jambhekar]

please check out attached file