hppiAnd, hppiOr, hppiXor don't process the last byte in a matrix row if matrix width is not divisible by 8. If the width is a multiple of 8 all is right. The library version is IPP Asynchronous Preview 2014 February
Thanks for finding the issue. This problem was present for those functions on CPU accelerator only, now it's fixed and the fix will be available in the next release of IPP Asynchronous Preview.