I have a need to store and retrieve 4-5 parameters of each of about 30 data items. I would normally use an array of clusters to hold the data. However, in this instance, I need to be fast as possible - retrieving the parameters for the data items will occur almost constantly in the background. I got to thinking if it would be any faster to use separate arrays for each parameter and performing five indexing operations. I know clusters do add some overhead, but how much difference is there between 1 array index followed by a 4-element cluster unbundle vs (much less elegant!) 5 array indexes?
You are right: The runtime disadvante for the array benchmark should come from the additional buffer allocations. Very interesting benchmark, by the way - my prediction on the runtime would have been different.
I could not sleep in without having modified your benchmark and having added some Replace Array Subset code (and using smaller array dimensions at the same time - my PC isn't that fast). Once it is inserted, the array solution is faster on my PC. (I set the default values of execution time indicators to those on my PC. The execution was done in LV 8.0.1, the VI is saved for 7.1.1)