I've been implementing the transcoding engine based on API 3.1 and the demo code of 'simple_5_transcode_opaque - async - vppresize' application.
By visual inspection so far, it looks to me that the recipe has been tightly followed in terms of what happens when. The value of asyncDepth has also been chosen to match the demo code (i.e. asyncDepth=4).
However, the execution encounters the problem. When the execution runs out of all the available 'Tasks', the attempt to synchronize the session results with MFX_ERR_ABORTED. Before that, the transcoding seems to run smooth (based on the return values). The attached log file illustrates the details.
Any hint/suggestion as to what might have been the cause of this problem is highly appreciated.