Intel IPP overflow of Ipp32fc (consultation is required)
$10-30 USD
Dibayar semasa penghantaran
I work with complex radio signals in Intel IPP, I store and I carry out over them operations in the Ipp32fc format (complex float).
In an investment example of a complex radio signal of [login to view URL] (first column Real, the second column imag).
I need to build a signal in degree which can reach great values, up to 128 (complexSignal^128). I do it through the ippsMul_32fc function.
When squaring (powSignal (2).csv) and the fourth degree (powSignal(4).csv) of problems is not present, everything is carried out correctly.
At higher degrees problems are observed, there is an overflow. Format 32fc allows to accept values from 1,175494351e-38 to 3,402823466e38.
How to normalize a complex radio signal that at its construction in degree to 128 there was no overflow and no NaN values?
Option to use Ipp64fc (complex double) does not approach.
The consultation is required.
ID Projek: #19401481
