Fast Magnitude Routine for IPP PCA 16bit FFT Result

[plesna]

Hello,

Does anyone know a fast magnitude calculation routine to turn a Ipp16sc FFT result into a Ipp16s or Ipp32s vector with PCA IPP calls quickly? The method I have been using (with the ippsNorm_L2_16s32s function, with each sc point as a vector) is too slow, almost doubling my processing time.

Thanks!

Adam

[Ying_S_Intel]

Dear Customer,

Try this one:

Call "ippsMagSquared_32sc32s_Sfs" and set the scaleFactors 16, and then call the 16s square root function by forcing the output of "ippsMagSquared_32sc32s_Sfs" of Ipp16s formatand input the forced vector as the inputof ippsSqrt_16s.

Hope it helps.

Thanks,

Ying S
Intel Corp.

Message Edited by ysong on 04-25-2005 07:27 AM