LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

Performance effects of using large clusters

I've written an application that uses a large number of disparate variables that I've organized into a large cluster, and saved as a Type Def. I pass this large cluster through a long series of subVIs, each of which unbundles and bundles the particular data that they are working with. Often, I have my data buried several layers of clusters deep within the large cluster.

This application seems unusually slow, with no distinct bottlenecks indicated by the performance meter. Does LabView use a lot of overhead when passing this large cluster in and out of the subVIs? Would using global variables within the subVIs, instead, help increase the speed? Does anyone have general suggestions on how to organize a larg
e amount of disparate data better? Thanks in advance.
0 Kudos
Message 1 of 2
(2,498 Views)
I have made the same mistake once...
Check the memory used by your typedef cluster
My large cluster contained about 30 clusters with each few string, doubles
and booleans.
The data itself was under 1KB but the cluster typedef took about 100KB .
With 25 subVIs
having each one cluster "In" and one cluster "Out" that sums up to 5 MB of
memory
eaten for 1KB of data...

Since I had to deal with only one dataset at a time and only one subVI
modifying it
at a time, I finally put the cluster in a global accessed within subVIs

However, I did not notice speed degradation but oversized LLB and
executable.

Jean-Pierre Drolet
Scientech R&D




"Cathode Guy" a écrit dans le message news:
506500000008000000281A0000-986697009000@quiq.com...
> I've written an a
pplication that uses a large number of disparate
> variables that I've organized into a large cluster, and saved as a
> Type Def. I pass this large cluster through a long series of subVIs,
> each of which unbundles and bundles the particular data that they are
> working with. Often, I have my data buried several layers of clusters
> deep within the large cluster.
>
> This application seems unusually slow, with no distinct bottlenecks
> indicated by the performance meter. Does LabView use a lot of
> overhead when passing this large cluster in and out of the subVIs?
> Would using global variables within the subVIs, instead, help increase
> the speed? Does anyone have general suggestions on how to organize a
> large amount of disparate data better? Thanks in advance.
0 Kudos
Message 2 of 2
(2,498 Views)