Linear resampling

daven-hughes — Thu, 17 Mar 2011 02:09:58 GMT

Here is my use case: I wish to resize a vector before transforming with an FFT. Due to the power of two input size contraints, I need to resize the vector to the next greatest power of two.

I realize I can perform a direct transform at arbitrary sizes but speed is of critical importance. I have seen mention of ippsResamplePolyphase but this looks like one of the most complex, state-driven IPP functions that likely sacrifices speed for generality.

I welcome any suggestions, thank you.

Linear resampling

igorastakhov — Mon, 21 Mar 2011 08:36:25 GMT

What is the purpose of using ippsResamplePolyphase? The purpose of this function is to change signal frequency. What are further steps of processing in frequency domain (after FFT)? What is the final goal? In the most cases (correlation, convolution, etc) it's enough to pad a vector with zeroes to the nearest power of two. Also you can use DFT function.

Regards,
Igor

topic Linear resampling in Intel® Integrated Performance Primitives

Linear resampling

Linear resampling