recently some problems with ippiMalloc made me write mixed code sometimes with ippiMalloc's and sometime with simple malloc's. I know they aren't aligned, but at least they can allocate memory correctly. The same now is happening to ippiResize. An access violation is reported by MSVC and then the function returns the code which means
"Not enough memory allocated for the operation"
which I could understand if my program would be eating too much memory, but it is not. As I analyzed in the same thread of the ippiMalloc problem , there is plenty of memory left, no heap corruption (indicated by MS heap analysis functions) and, as far as I could debug, no code setting weird parameteres to disrupt all the IPP library (and I have seen this before with Ipp). Even I tried to see the assembly inside, but realized it was too much a waste of time, when it started to go into the profoundness of Windows DLLs.
So all was left for me was the same solution with the ippiMalloc problem: substitute it.
Since I still wanted functions that could do better performance, I took the Framewave version of it, fwiResize, and guess what: it worked perfectly, after 2 days struggling with all sorts of tests with Ipp. Now I'm using both libraries, which messes up with my code elegance. :)
I am compiling in MSVC 9.0 and calling my library, which calls IPP and Framewave, from a C# Windows Forms Applications. It is about image processsing, but currently I am dealing with very small images.
If someone is having the same problem or have a better solution (or hints to solve without Framewave), I would be very thankful.
the problem was solved. Details were post in another thread: http://software.intel.com/en-us/forums/showthread.php?t=57121#53657. Memory corruption and incompatible library with IDE were the cause.
Thank you for your help.