01-21-2015 10:11 AM
What about LabVIEW 2014? Is it better then 2013/2012/2011?
01-22-2015 09:52 AM
Nope it's even slower.
LV2014 have roughly 1/3 times the performance compared with LV2011 according to my measurements.
Br,
/Roger
01-26-2015 02:12 AM
NI, guys, this is a real issue.
I'm hoping that this slowing down is some behind-the-scenes optimisation or re-architecting which will eventually get us to where we should be but at the moment, it's giving me cause for concern. Real concern.
01-26-2015 02:29 AM
Regarding programming languages:
Amateurs worry about syntax and semantics. Professionals worry about money. Masters worry about performance.
/Roger
01-26-2015 05:19 AM
@Intaris wrote:
NI, guys, this is a real issue.
I'm hoping that this slowing down is some behind-the-scenes optimisation or re-architecting which will eventually get us to where we should be but at the moment, it's giving me cause for concern. Real concern.
LabVIEW 2011 can come in handy as a saviour in a performance crisis.
/Roger
01-26-2015 10:19 AM
I completely agree with Intaris. It is troubling that the problem has grown worse in later versions of LabVIEW rather than better. Is there any status update NI? Even a brief "we are having trouble fixing it, here's why" goes a long way in my book.
01-26-2015 10:48 AM
@shansen1 wrote:
I completely agree with Intaris. It is troubling that the problem has grown worse in later versions of LabVIEW rather than better. Is there any status update NI? Even a brief "we are having trouble fixing it, here's why" goes a long way in my book.
This is probably due to a lot of LVOOP related bugfixes, such as extra conditional checks during runtime (slowing things down significantly if you have OOP code in a tight loop), this probably caused by the "popularity" (leading to a lot of bugreports/fixes) of various frameworks such as Actor.
The only way ahead is probably if NI decides to re-architect parts of LV (thus no hope here in the imminent future) or you "downgrading" to earlier versions, or perhaps refactoring, call it "optimizing", the code into obscurity and arcaneness for performance issues with LV.
But it's just me speculating.
01-26-2015 01:06 PM
@User002 wrote:
...
LV2014 have roughly 1/3 times the performance compared with LV2011 according to my measurements.
Br,
/Roger
1/3 the performance?! That's not the direction I want to go.
01-26-2015 01:43 PM - edited 01-26-2015 01:49 PM
@PaulG. wrote:
@User002 wrote:
...
LV2014 have roughly 1/3 times the performance compared with LV2011 according to my measurements.
Br,
/Roger
1/3 the performance?! That's not the direction I want to go.
Didn't you see the disclaimer?
I'm leaning quite heavily on LVOOP and rely on good performance there. Perhaps you don't?
The current state of affairs makes me doubt there is a way of coding up a LV application that is well structured, performant and have lots of functionality. I get a feeling that I can pick one or two, but three is too much asking. In either way I'm doing it wrong according to NI.
Anyway, It is sad that such a promising way of developing highly parallell applications have become stagnant and have started to emit an ooze of staleness.
01-27-2015 02:33 AM
Roger,
can you share a benchmark or some benchmark data including LV 2014 so that we can discuss the numbers?