In new release, the redundant Atom/Merom CPU library optimizations have been combined (v8+s8 and u8+n8) into a single "smart dispatch" library; the uncombined library optimizations will continue to be supported in the IPP 6.1 product.
When running Intel IPP crypto sample for OpenSSL on Hyper-Threading systems, the AES benchmark application reports slow performance. Users need to use correct threading setting to avoid the problem.
JPEG APIs are implemented using new threading model in IPP 7.0 and later
This page offers the list of Intel IPP functions specially optimized for Intel AVX.