Hi! I am having some trouble with the crosscorr_fr16 function.

The first argument is pattern, that is present in the second vector.

The second vector is much larger than the first.

Thus the 'n' is much smaller than the 'lags' number.

The result vector is large enough to handle the result

and the c version of crosscorr handles this correctly

but the fast crosscorr_fr16 generates an unknown_exception

after calculating n samples in the output vector.

It seems like there is a limit so that lags can not be

greater than the size of the first vector.

I have tried to understand what causes this exception and how

to modify the library function, but failed sofar.

best regards

Jarl

I think I understand the problem now. The crosscorr is not intended to

have different length vectors. By removing the P2 += -1; instruction

close to the end, the function does what I want it to do.

best regards

Jarl