Intel® Integrated Performance Primitives (Intel® IPP) Developer Guide and Reference

ID 790148
Date 3/22/2024
Public

A newer version of this document is available. Customers should click here to go to the newest version.

Document Table of Contents

Special Arguments

Some Convolution and Correlation functions described in this section have two implementations of the algorithm:

  • For small data size, function processes data as described by the formula
  • For big data size, function uses FFT-inherited algorithms

The optimal algorithm is selected automatically according to the input data size. You can manually choose which algorithm to use by passing one of the following predefined values to the algType parameter of the function:

ippAlgAuto

Select the optimal algorithm automatically.

ippAlgDirect

Use direct algorithm as described by the formula.

ippAlgFFT

Use FFT-based algorithm implementation.

These values are declared in the IppAlgType enumerator.

Several functions support normalization of the output data. You can choose which normalization to apply by passing one of the following values to the function:

ippsNormNone

No normalization (default).

ippsNormA

Biased normalization.

ippsNormB

Unbiased normalization.

These values are declared in the IppsNormOp enumerator.