I'm looking for FFT computation speeds in LabVIEW and LabWindows/CVI.

[dunx]

I'd like benchmark comparisons of FFT speeds in LabVIEW and CVI (on a given platform) compared with regular C-compiler code.

[waldemar.hersacher]

Your question is not really applicapable. This is because LabVIEW and CVI use the same DLL for FFT. These DLLs are written in regular C and compiled with a regular C-compiler (I think they use MS Visual Studio).

So any benchmarks make sense to compare specific implementations and compilers. Some implementations restrict the length to be a power of 2.

I can do some benchmarking of LV and CVI against MS Visual Studio if you can supply the C source code of the algorithm you are interested in.

Waldemar

[dunx]

Thanks Waldemar,
I think you confirmed what I had suspected: that LV and CVI basically just act as C compilers. I was worried whether the nice graphical front-end slowed down computation, when compared with a simple compiled C program.
Do you have any example times for FFT exectution? I guess standard lengths would be 1k, 4k or 16k points, and I accept that it will vary from machine to machine. I'm just trying to get a feel for how long it takes - a few ms, 100s of ms or what?
Cheers,
Duncan

[Greg_McKaskle2]

> I think you confirmed what I had suspected: that LV and CVI
> basically just act as C compilers. I was worried whether the nice
> graphical front-end slowed down computation, when compared with a
> simple compiled C program.

As stated in the other posting, the FFT and other analysis libraries are
shared between LV and CVI, and can in fact be called from other
environments capable of calling a DLL.

But, just to make sure you don't jump to the wrong conclusion, CVI is a
C compiler, an ANSI compliant compiler. LV is not a C compiler, but a
compiler of the G language, the graphical dataflow diagrams in LV. As
with other languages, some of the functionality will come from code
being generated, and other by linking to OS libraries or
NI shared
libraries.

Greg McKaskle