I am using the intel IPP function, ippsConvert_32f24s_Sfs (), which converts the 32-bit float data to 24-bit signed char (3 signed character bytes) after scaling the actual data with an arbitrary/programmer-choice scaling factor.

At the receiver end, I wanted to revert the 24 bit data to 32 bit float information with less loss in information. Even though IPP package provides another built-in function (ippsConvert_24s32f) to achieve this requirement, I can not use it since I receive the data in totally different environment.

In this scenario, I would like to know that how the function: ippsConvert_24s32f works to convert 24 bit signed char data to 32 bit float data.

Any information in connection with above query will be a great help.

Thanks to everyone.

The basic expectation is understanding the operations hid in ippsConvert_24s32f.
Since my float data contains signed values, any missing/mishandling of data may end up with spurious errors.