FFT results from LabVIEW & GPU do not match

[MathGuy]

You've provided the input sequence but only the magnitude of the output. To investigate the differences (e.g. is a scaled window being applied to the FFT), we need the complex results.

-----------------------------------------

Original post by gaisi

---------------

Hi, Have you meet such problem that the FFT results got from Labview and GPU are different?

My input is 1,2,3,2,1,2,3,2,1,2, and the results from GPU are 19,1.23607,3.23607,3.23607,1.23607,1,3.23607,3.23607,1.23607;

And the ones from GPU computing is 19,2.80079,3.74502,3.74502,2.80079,3.29691,2.80079,3.74502,3.74502,2.80079;

[gaisi]

Oh. my bad. This is the results from Labview FFT computing.

And this is the results from GPU computing. The first one is the image part:

And this is the real part:

You could see that the image part of the results are different.

If it is necessary, I can post my GPU and Labview program here.

[MathGuy]

Can you widen the element views for the array elements? It looks like they imaginary parts agree but the Nyquist value is clipped. If it is being displayed in scientific notation, the exponent may small enought that the value should be treated as zero (relative to a scaled version of machine epsilon for the precision being used).