How can I get better time permformances for my vi?

[Zinkanagwaka]

I have a vi simulating an electronic device we have in our lab. When I launch the vi on a Win2k OS running on an Athlon1800+ (256RAM), the vi works at least twice faster than on my own PC, same OS, Win2k, but running on a P4 2.53GHz including 1Go DDR-RAM!!! I could not figure out that! How can I get rid of this issue? Thanks for your help.

[Ben]

IS everything else the same?

Is the color depth set the same on both machines?

What does the app do?

If it is doing disk I/O alot, the disk I/O subsystem could make a difference.

Does the Ahelon have a big cache?

Does the Athlon have a graphics adapter?

Just guesses,

Ben

[Zinkanagwaka]

Hi Ben,
Thanks for your interest in my request. Actually the vi simulates the full electronics of an atomic force microscope, namely the signal detection (photodiodes), a phase locked loop, an automatic gain control including a RMS-to-DC converter and a PI controller, a phase shifter and the oscillator itself. It's quite a big computation as 5 second order differential equations have to be numerically solved simultaneously. At that stage of the simulation, there is no I/O to the disk like read/write/delete buffers, everything is done through the cache of the computer. But talking about disks differences, the Athlon disk is a standard IDE disk (40Go) and my own one is also an IDE one, but driven through the RAID controller of my main board (ASUS P4PE).
Do you think that this could be a problem, although there is no I/O to the disk?
The graphic card of the Athlon is an ATI Rage Fury Pro and I have on my own PC an ATI Radeon 9000 PRo (128Mo). If it were a graphic card problem, I should see the difference!!! The cache memory of the Athlon is about 240Mo, and I didn't check the one of my P4... Hope this might give you some ideas for an answer.
Best regards,

[Ben]

You did not answer my question regarding the color depth on the two machines.

LV 6.X uses 16 bit graphics. If one machine is set for 16 bit, it can just use the controls the way they are set-up.

If the other machine is not set for 16-bit, it will have to work harder to convert the colors.

Otherwise, post you code, we may be able to spot something.

Ben

[Jac.m.a.m.Oppers]

On Sun, 16 Feb 2003 13:19:30 -0600 (CST), Zinkanagwaka
wrote:

>I have a vi simulating an electronic device we have in our lab. When I
>launch the vi on a Win2k OS running on an Athlon1800+ (256RAM), the vi
>works at least twice faster than on my own PC, same OS, Win2k, but
>running on a P4 2.53GHz including 1Go DDR-RAM!!! I could not figure
>out that! How can I get rid of this issue? Thanks for your help.

In a parallel operating world such as LabVIEW a relative fast machine
is capable of spending much resources at a wrong place, resulting in
relative less speed compared to machines who do not operate that fast.

Jac.

[unclebump]

Try looking at this webpage from the NI site. When I switched to another windows 2000 computer I had to correct this issue. Changing this setting had a big impact on performance. http://zone.ni.com/devzone/conceptd.nsf/webmain/28A6979ECA9A0CAF862569050053D92D?opendocument

[Zinkanagwaka]

It looks interesting. I'll check, thanks.

[Zinkanagwaka]

The color depth of both machines being similar, I get the the same behavior, the P4 is still much slower.

[pn2500]

> Try looking at this webpage from the NI site. When I switched to
> another windows 2000 computer I had to correct this issue. Changing
> this setting had a big impact on performance.
> http://zone.ni.com/devzone/conceptd.nsf/webmain/28A6979ECA9A0CAF862569050053D92D?opendocument

XP seems to have the same issue (?), though they renamed the buttons in
the control panel dialog box, the concept seems the same, (though
I havent tested it).

Can one manually assign thread priority from within Labview via a
Win32 call, as from within CVI?

[Wojcio]

Hi,

So Below are the answers from the customer. It looks like the drivers are ok - but the customer has LabView 7.1.1 - upgrade to LV 8 is not an option - the customer is finishing big project and has little timeframe left.
Is it possible to do anything to make the half duplex work for him? Maybe some changes ini files on CVS - any ideas? If it will possible to do it - the customer will stay with Labview in next projects.


Below the information from customer:

Hi Wojtek,

as agreed, I have checked software vesions.
To supply you with full info I enclose pitures showing:
1. Soft versions (NI IMAQ 2.0.2, NI VISA 3.4.1, etc)
2. Max version (4.0.2.)
3. MAX - network configuration (permanent, all changes to "half duplex"
after rebooting CVS return to "full duplex").
4. Front panel of MAX during picture acquisition
(3,75 fps acquied and displayed, no problems, no time out, only the
picture is dark
but on the connected screen (t VGA output) the pictures change 4 /
sec.)

If you need more info - pelase let me know.
I do hope for recieving anysolution very soon.
Best regards
Marek


Many thanks
Wojtek Rachwalski