02-16-2007 08:30 AM
02-16-2007 08:54 AM
02-16-2007 09:10 AM
02-16-2007 09:39 AM
02-16-2007 10:13 AM
02-16-2007 11:01 AM - edited 02-16-2007 11:01 AM
Hi Norbert,
Actually, I had already implemented a similar solution..
The elements of the array point to a case structure which populates the "string" array. It's sufficiently fast. I was curious if there was another trick that I was not aware of.
EDIT: Just noticed that Tomi suggested the same <Thanks>
Once I convince them that the string array is no longer required, I will remove it. 😉
Thanks guys for all the help. The overall progress on the older slower machine is in the magnitude of 100 times!!! 😮 - wow!!!! -
Test time dropped from 79 min to 6 min!! I'd say that's "pretty good"..
THANKS!! 🙂
RayR
Message Edited by JoeLabView on 02-16-2007 12:03 PM
02-16-2007 11:12 AM
Thank you Ray for the entertainment!
This thread reminds of a quote from "The Silicon Valley Guy handbook" that ascribed the following characteristic to SVG's.
"One of my favorite pass times is Gang Applience re-wiring."
Ben
02-16-2007 11:15 AM
I'd suspect that the latest numeric substitution ideas would go a bit faster if you passed the array through Shift Registers and used Replace Array Subset to make changes rather than auto-indexing your output arrays. Guess it depends how smart the LV compiler optimization is. It'd take a fair bit of smarts for it to recognize the acceptability of reusing the same memory space in the auto-indexed output method. With shift registers, it has no choice.
-Kevin P.
02-16-2007 11:22 AM - edited 02-16-2007 11:22 AM
Perhaps we need a new loop tunnel type in LabVIEW that would be in-place auto-indexing. This tunnel would be a combination of shift register and auto-indexing tunnel. Like shift-register it would always be visible at both ends of the loop and it would operate in-place. Like auto-indexing tunnel it would auto-index elements from the array. It may not be a great leap for man kind but it would be a small step for LabVIEW performance tuners... 🙂
@Kevin Price wrote:
I'd suspect that the latest numeric substitution ideas would go a bit faster if you passed the array through Shift Registers and used Replace Array Subset to make changes rather than auto-indexing your output arrays. Guess it depends how smart the LV compiler optimization is. It'd take a fair bit of smarts for it to recognize the acceptability of reusing the same memory space in the auto-indexed output method. With shift registers, it has no choice.
Message Edited by Tomi M on 02-16-2007 07:28 PM
02-16-2007 11:35 AM