Intel® Integrated Performance Primitives
Community support and discussions relating to developing high-performance vision, signal, security, and storage applications.
Welcome to the Intel Community. If you get an answer you like, please mark it as an Accepted Solution to help others. Thank you!

Suming the Results of Multiple FIR Filters


Hello everyone,

I would like to create various random broadband signals. To create a broadband signal, I am using random (white) noise as input to an FIR filter. In practice, I will have many FIR filters with different coefficients (and likely different sampling rates, although they can be set to 44.1 kHz, 44.1 / 2 kHz, 44.1 / 4 kHz, ... etc).

Because I have so many broadbands, performance is an issue. I would like use IPP's FIR filter functionality to achieve this. However, my research on it indicates that I have to make a function call for each filter separately, and this would write the output samples in an array.

In my case, I am not interestedin the individual outputs for each FIR filter. This is because I will be summing all outputs together and make one new signal that contains the sum of all outputs. I am only interested in the sum of all outputs.

What I am concerned about is that writing the output each time in a separate output array for each FIR filter causes an overhead. Is there a way to configure the FIR filter so that it "adds" the output to an existing array as opposed to simply overwriting any existing data? This way, I won't have to do a whole new loop for each FIR filter to copy its output samples to an existing array.

Please note that each FIR filter has a different input array (i.e. different random noise).

Thank you.


0 Kudos
1 Reply

Hi Greg,

there is comment from our expert

there is no such way. ippsFIR is a primitive function.