Slow LabVIEW 8.2 execution with UI events

[Travis_M.]

Good morning,

Some users have noticed that there seems to be a decrease in performance in LabVIEW 8.2 (in comparison to other versions of LabVIEW) when a menu has been activated.  The issue seems to be dependant on which operating system is used (Win 2K is apparently the slowest).  

This thread is intended to be a means of troubleshooting this issue. If you have experienced this issue, please let us know as many details as possible about the issue here.  If you are experiencing any other problems that are not related to this, please post those issues in a new thread – this should be reserved for the run-time slowdown of LabVIEW 8.2 compared to previous LabVIEW versions when runtime menus are active.

Thanks,

[PJM_Labview]

This is my take on the issue (Note this might not be the same issue, but this is nevertheless a major one).

PJM

[Henrik_Molsen]

I am glad to see that this is beeing taken serious by NI - thanks.

From the postings on info-labview in dec 2006 and from my own experiments I am not shure the problem is related the run-time menues alone. The run-time menues was my first experience on the slow down problem but I think it is a more common problem. Looking at the link above to LAVA and the findings of the example "Order of Transformations.vi" it seems to me that the NI guys should look at a more general problem in the LabVIEW <-> Windows interface. It must be something that has changed from LV7.1.1 to LV 8.2.

I created a VI that runs much faster when the Windows focus is moved outside the LV front panel. (This has been reported to NI). Comparing this with the other findings added to the info-labview thread I think it must be something with the assignment of focus to the application in Windows and the associated task priority.

I have made several experiments with VI's that showed slow performance and these were all executed on the same PC with different operating systems and LabVIEW versions.

To make shure other settings did not influence on the result the PC was ghosted with a clean installation of Windows (XP and 2K) and LV 7.1.1 and 8.2. By doing these experiments we can exclude PC hardware (it is the same) and settings of windows drivers (it is the same). We can also exclude tricky settings in LabVIEW (it is default settings) and tricky settings in the VIs (they are the same except for the mass compile to LV 8.2).

Best regards
Henrik Molsen
Certified LabVIEW Developer
microLEX Systems A/S

[Henrik_Molsen]

Hi NI guys

For your information: Extracks from info-labview in case you didn't read it there::

-----------------------------------------------------------------

Subject: Re: Execution speed
From: "Henrik Molsen" <hmo@microlex.dk>
Date: Thu, 21 Dec 2006 08:20:29 +0100

Hi Russel

I think your are right about the 'Windows craziness' but it must be related to the new LabVIEW versions (8.2 and maybe 8.0) and maybe its interface to Windows.

I have tried to run the same VIs with execution problems on the same operating system on the same PC in LV 7.1.1 and LV 8.2. There were no problems in LV 7.1.1 but execution problems in LV 8.2. Notice that the only thing that changed was LV version. The PC was ghosted between the tests and started out with a 'clean' Windows and a default LV installation (=no nonstandard settings anywhere, not even in windows drivers).

Strange!!!!?????

(PC: P4 3GHz, 1Gb. Win2k and WinXP was tried both with LV711 and LV82. Win 2K is slowest with LV 8.2).

Best regards
Henrik Molsen
Certified LabVIEW Developer
microLEX Systems A/S
Dr. Neergaards Vej 5C
DK-2970 Hoersholm
Denmark

Tel:   +45-45 76 21 00
Fax:   +45-45 76 22 00
Email: hmo@microlex.dk
WEB:   http://www.microlexsys.com

------------original message----------------------------------------------------------

Subject: Execution speed
From: "Russell" <wagner@lmi.net>
Date: Wed, 20 Dec 2006 17:30:23 -0800

>
> Speaking of LV8.2 execution speed, I came across this effect described
> below, and I am wondering if some LV internals guru could explain  
> to me
> what is happening:
>
> 1) Open the example "Order of Transformations.vi" (search Examples for
> "3D").
> 2) Run the VI. Observe how the speed of motion in the 3D picture
> increases when you click on a drop-down menu and then goes back again
> when you click outside of it (at least that's what it does on my
> computer).
> 3) Play with VI execution priority.

I discovered a similar thing recently (Using LV 8.0.1, I believe)  
where a VI would execute (much!) faster while moving its FP window  
around during execution. That VI was having its FP controls updated  
from another VI, and I was timing the execution speed of that  
operation in a loop on a graph. The FP object updates took about 30ms  
normally and dropped to 1-2ms as the screen window was dragged. I  
figured it had something to do with screen updating in the OS (I seem  
to recall that the FP controls and indicators were not updating  
during the drag operation), but the affect was extreme and appalling!  
This was on a Windows machine, so I just took it to be windows  
craziness. Perhaps you are seeing something similar related not to  
LabView but to the OS, but how to untangle these affects is beyond me.

----------------------------------------------------------------------

Subject: Re: LV 8.2 VI execution much slower than with LV 7.1
From: <stefan.tauche@infineon.com>
Date: Thu, 21 Dec 2006 11:13:10 +0100

Hi Jeff,

No menues are used in the application.
I think it will be a problem to isolate this behaviour,
since I can only see the delay of execution speed in the changing period
of a signal on an oscilloscope (signal generation in turn involves
calling another
application via active-X)....
Any other idea, how to track down this problem ? is there any execution
speed
timing taken internally by labview, which I may read out and report ?

Stefan 😉
----------------------------------------------------------------------

Subject: Re: LV 8.2 VI execution much slower than with LV 7.1
From: <Jeff.Boettcher@ni.com>
Date: Mon, 18 Dec 2006 09:31:59 -0600

Hello Stefan,

As you may have seen, the other issue that was related to slowness of
execution in 8.2 turned out to be related to run-time menus.  Are you
using custom run-time menus in your application?

Regardless of the cause, it does seem wrong to me that execution speed
wouldn't return to what it initially was after the application loses
focus
and then regains it.  I tried a simple VI with virtually nothing in it,
and focus didn't seem to impact execution speed much.  If you can
reproduce this behavior in a simple VI (you might start from your app
and
start stripping things out) I'd suggest submitting it to the NI
Discussion
Forums ( http://forums.ni.com ) for further investigation.

Regards,

Jeff Boettcher
Staff Software Engineer - LabVIEW R&D
National Instruments

[Travis_M.]

Hello,

Thanks for the post Hendrik, and for updating those who are joining in on the conversation.  Forgive me for speculating a little as well, because I am not experiencing the same behavior on my machine (or on another test machine here).  However, I do know that activation of menus, and certain other UI events occur in the UI thread, and can have a large impact on how that thread is executing.  I wonder (though ultimately I am unsure) how the differences in scheduling algorithms between Windows 2K and XP might affect a very active UI thread.  I wonder if these UI events (such as activating a menu or ‘dragging the window’ are having some affect on the priority that Windows is giving that thread.  Before I begin speculating too much, does anyone have another example that might demonstrate this behavior?

[Travis_M.]

Happy Friday –

I’m not sure how many people are following this thread, or have experienced this problem.  For those following this thread, an update was made on the lava forum and a bug report was filed that we think fixed one UI slowdown issue.  I’m not sure if this is the same issue as discussed in other posts in the thread.  We are still interested in determining if the other issues are problems that need our attention.  If you are experiencing these slowdowns and provide us with an example we would be very grateful.  If not, just keep those ears open and let us know if something comes up.

Thanks for everyone’s contributions.

[ejb068]

Hi,

I've noticed a significant slow down in performance when I switch to 8.2 verses 7.1. My application has several windows that are parallel processes to the main program.  These other windows act a logs for comminications/staus/results etc... I use multi-column listboxes to dispay these logs. I also use the properties to change the colours of the text and backgound based on the data being displayed.  In 7.1 this did not slow the main program much, but in 8.2 when I have these windows open, the program slows down significantly.  After I get the program started I close all the extra windows to save execution time.  All my windows do have run-time menus but haven't tested it without these to see if this is the cause. The program was written in 7.1 then opened on a computer with 8.2.  Is there anything I need to do is improve performance?

If there are any updates on this issue I'd really be happy to try them out.  Right now I'm seriously considering to go back to 7.1.

Thanks,

James

[ejb068]

I have been able to improve the graphical performance considerably on my system.  I found that the slow-down was caused by a incompatability with my ATI graphics card.  It seems to only affect Labview.  I adjusted the slider of the amount of graphics acceleration.  Taking the slider down one notch improve performance significantly. If anyone else is using an ATI card you may want to check it out. I don't understand the real reason for the issue but not that I've found a way around it I'm not too concerned. Attached is the screen shoot of the way I fixed it.

-James

[Kikujiro]

When can your hope for a corrective measure of the largest bug of LABVIEW 8.X.

This is very penalizing, to see intolerable!.

Thank's.

A. Carbonnel

Labview developer for scientific applications since 12 years

Electricité de France, Division Technique générale

alain.carbonnel@edf.fr

[NIJanell]

Hello Kikujiro,

I'm not exactly sure which bug you are referring to - there are multiple issues discussed in this forum and some have been determined to not be bugs.  Can you please explain the behavior you are experiencing and possibly post a small example?

Thanks!

Janell R | Applications Engineer