10-20-2009 08:32 AM
Ray.R wrote:I have never seen code running faster by moving a scrollbar.
Thanks Partha.. 😉
Like I said. I never saw the code run faster by moving a scrollbar.
Running slower. Yes, I've seen that.
I don't know what would cause the code to run faster except as Broken Arrow mentionned "sounds like Windows is setting the priority of the open windows".
John Pierson wrote:Actually i don't have property nodes, i'm using three invoke nodes to grab an image from a 3d graph.
.....maybe property nodes slow down when mouse is clicked, and method nodes speed up when when the mouse clicks?
anyone else seen it?
Just a shot in the dark: Usually, this would slow down the process, as it would need to refresh the graphs often while obtaining the data. Maybe by some strange coincidence, it would help the refresh and as a consequence run faster. ?? ... That doesn't make sense.. I would rather suspect that it keeps your code executing in one of the sequence structures with a loop which gives you the impression of running faster.
Is it visually running faster or is it actually processing more data? Or same quantity of data but in a shorter time? Do you use something like a Wait Until Multiple?
10-20-2009 08:40 AM
i'll try removing all three when my code stops running, and i'll see if that knocks the timing thing on the head.
i'll report back.
10-20-2009 08:45 AM
Ray.R wrote:
Running slower. Yes, I've seen that.
I don't know what would cause the code to run faster except as Broken Arrow mentionned "sounds like Windows is setting the priority of the open windows".
OK, now understand. To be honest, I have seen also "speedup" by pressing mouse. More exactly (it was reported by customer) - when mouse pressed and hold on the Title Bar, then application was 10-15% faster. I was able to reproduce that (LabVIEW 6.1). This weird problem goes away with upgrade to LabVIEW 7, so we haven't investigate what exactly was wrong.
Andrey.
10-20-2009 09:01 AM
10-20-2009 10:17 AM
Andrey Dmitriev wrote:OK, now understand. To be honest, I have seen also "speedup" by pressing mouse. More exactly (it was reported by customer) - when mouse pressed and hold on the Title Bar, then application was 10-15% faster. I was able to reproduce that (LabVIEW 6.1).
When using LV6.0, I quickly upgraded to 6.1 and then 7.0 to 7.1. I used 7.1 for a long time. Still using it.
I cannot recall seeing a speedup using LabVIEW. However, I did see speedup in some applications (can't remember which ones) where you hold the left mouse button down and scroll or move the mouse left to right, thus causing some sort of speedup in the application. I've never quantified the perceived or actual improvement in speed. It was rather amusing to observe.
It would actually be interesting to find out the root cause, because we could actually "speed up" our applications. 😄
I did try changing the priorities of running applications, within the Task Manager. I had not noticed any significant changes or improvement. 😞
"Hidden" tricks are always good to know.
R
10-20-2009 10:22 AM
I'm on LV8.2
its odd that it doesn't happen each time with me either, if it happens first time, but it may not happen second time with similar conditions, making it harder to test the speed difference.
10-20-2009 10:23 AM
Now we really want to see the code to try to figure out this mystery...
😉
10-20-2009 10:27 AM
Still 22 hours untill the VI is done i'm afraid.
i'm trying to make a test programe based around simply plotting a random 3d graph, then loop it a number of times and log the amount to time it takes.
a nice simple test. - which probably means it won't show the same effect.......
...we'll see.
10-20-2009 10:53 AM - edited 10-20-2009 10:55 AM
right, i've made a simple VI - see attached VI.
I tried it a couple of diffent ways:
1. Full screan, left alone
2. Windowed, left alone
3. Mouse depressed, Full Screan
4. Mouse depressed, windowed
5, Wiggling mouse around while depressed, full screan
I tried thease with firstly, just a Invoke node, then secondly with both an invoke node and a Property node.
The results were as follows:
First ones - just invoke node
1. 15.8 sec
2. 15.8 sec
3. 14.6 sec
4. 14.8 sec
5. 15.4 sec
Second ones - both invoke and property nodes
1. 15.8 sec
2. 15.6 sec
3. 14.4 sec
4 14.7 sec
5. 15.2 sec
Thease are an average over 5 times for each.
anyone else fancy trying to see if they get the same results?
It seems that my computer likes to have the mouse depressed while VI's are running. - weird!
plus if someone would liketo improve my test programme, please do, if you'd liketo take away time stamp 2 from time stamp 1 automatically, it could give the result easier, and then a for loop around the whole lot could let us run multiple test ans average them quicker. - i would but i'm off home now. 😄
10-20-2009 11:15 AM
John Pierson wrote:
anyone else fancy trying to see if they get the same results?
It seems that my computer likes to have the mouse depressed while VI's are running. - weird!
I can confirm, that the given code running a bit faster when mouse pressed on the scroll bar:
Sure, caused by strange interference with UI thread...
Andrey.