This project has moved. For the latest updates, please go here.

Problems with 8 core and and getting FFT results.

May 27, 2011 at 5:11 PM


if I do this:

Dim FFTBufferInLeft(FFTLength) As Numerics.Complex

'Here I assign the buffer to a MLS signal (a pseudo random signal with only 1 and -1 in it)

Dim RFT As New MathNet.Numerics.IntegralTransforms.Algorithms.DiscreteFourierTransform

RFT.Radix2Forward(FFTBufferInLeft, MathNet.Numerics.IntegralTransforms.FourierOptions.Matlab)

I've tried it with simple signals like an sine waves and it works fine, but with more complicated signals, the first couple of thousand samples seems ok, but then it starts to deviate more and more and almost goes unstable.

This applies to both real and imaginary part.

Am I using the lib wrong?

I used the old irridium with the same signals, and it worked fine, but now on a newer computor, I thought I should upgrade to numerics and then it stopped to work.

Any ideas? I can supply a fully working project in Visual Basic express 2010 with .net 4 if anyone cares to have a closer look, unless ofcourse, you spot my stupidity!



May 31, 2011 at 12:00 PM

Sorry guys, Please ignore all this, it's just that I missed that when you do a vector in VB with dim x(2^N) as integer you get a 2^N+1 long vector, and that screws all my results up since I get a 0 inserted in the end of the signal. The lib works perfectly!

Again, I do apologise for any wasted time! (if it's any comfort, I've wasted 2 days on this bl"#" VB, I'm normally a C/C++ kinda guy)

kind humble regards