I don't know why, but if you initialize to decoder with ((Number of Cores of Die) >>1), or 1,it will work acceptible, otherwise if you choose auto mode (0) it will freeze the hole system.
I set the PREPROCESSOR to "_OPENMP" and working with IPP 7.0.7 and UMC 7.0.7.
I think there are someproblems in general design depending on Threading.
For the decoder "will freeze the hole system". Does it mean the decoder has a deadlock, or the decode use all of the CPUs? The later one is fine, since the decoder has same numbers with CPU cores.