this is my first post in this forum so please bear with me. While exploring the FilterBorder family of functions, I used ippRndFinancial round mode with ippiFilterBorderInit_64f. However it returned error -213. My question is can ippRndFinancial be used with ippiFilterBorderInit_64f (the documentation did not mention anything regarding this)? Attached is my Test.cpp file.
ipptypes.h defines status of this operations: ippStsRoundModeNotSupportedErr -213 /* Rounding mode is not supported. */
What kind of rounding mode do you need for supporting this function?
thanks for reporting, yes, it is the problem in the IPP documentation and in the ippi.h header file. Actually this function supports only the ippRndNear mode. We will correct documentation. If you need supporting other round modes please submit an official request through https://supporttickets.intel.com/
Dear Pavel and Gennady,
thanks for your replies. I decided to use ippRndNear rounding mode. While exploring the FilterBorder family of functions, I also observed that each kernel data type only supports certain image data types as shown below.
kernel data type supported image data types
ipp16s ipp8u, ipp16u, ipp16s
ipp32f ipp8u, ipp16u, ipp16s, ipp32f
Wrong combination of kernel data type and image data type results in error -59 (ippStsDataTypeErr). If this is indeed the correct behavior for this family of functions, I was unable to find any documentation about it as well. (as a suggestion, this information could be added to the documentation for the sake of completeness.)