speed of split/merge signals

[knapkerd]

I was wondering the order of magnitude of time it takes to merging signals and then splitting them?  Is it about 1ms, 10ms, etc.  I am merging and then splitting about 4 signals, all of which are averaged values (so 1 number).  I am try to increase the performance of my code and I am wondering if I might be taking a significant performance hit from doing this (I merge and then split the signals for code readability reasons).

Thanks for your help.

--Robert

[MikeS81]

Hi Robert,

can you show us a screenshot of what you mean?

---

knapkerd wrote:

...(I merge and then split the signals for code readability reasons).

---

Without seeing your code, i think this is no good reason to merge and split signals.

Mike

[knapkerd]

I have attached a screenshot of what I am doing (slightly modified due to the proprietary nature of the code).  Let me know what you think.

Thanks for the quick response.

--Robert

[MikeS81]

Hi Robert,

i think to merge them is ok, because you need it to store the data, but why don´t you use the wires before the merge block if you want to work with the single signals? If you need better performance than remove the split block and use the signal you already have.

Mike

[knapkerd]

The merge and split was something I did a while ago and it made the code more readable.  I kind of figured I should connect them directly to save on the performance.  But I am not sure it will matter since my boss just told me that we are probably going to scrap the program and forget about doing real-time analysis!  Yay, 2 months of work down the drain.

Thanks for you help Mike.

--Robert

[Wiebe@CARYA]

If the signals are one sample, it would take about 10 ns. If the signals are
100Ms, it would depend completelly on your machine.

You'll have to do the benchmarking yourself. Use the profiler, or place the
code in sequence frames, and put a timer before and after it, then subtract
them to find out how long it took. If it executes too fast, put it in a for
loop, and execute it 10000 times or so. Make changes, and spot the
difference.

Or post the code so we can review it.

Regards,

Wiebe.

[knapkerd]

That sounds like a good idea.  If we end up using the code I will try some benchmarking.  Thanks for the suggestion.

--Robert