Hi Ramasubramanian Subramanian,
For support on your ARM+DSP Keystone 2 device, you may get better and faster support by contacting your local TI FAE directly.
That said, the PA on the C6678 should be very similar in terms of operation. Is the log you posted for C6678 from the project, PA_multicoreExample_exampleProject? Would you mind letting me know your toolchain? (CGT, CCS, MCSDK versions, etc.) I can see a similar error on my end and I will double-check the code internally to see what went wrong.
As for using the PA to send to different cores, the e2e post you linked is very relevant. You should be able to use PA to push packets to different queues based on the UDP port, and have different cores popping different queues assigned for them. But first, we should get the PA multicore example working.
Ivan
EDIT 8/7: I realize the error from PA_multicoreExample_exampleProject is from running on all 8 cores. The variable, NUM_CORES, is set to 4 by default. Running in all 8 cores will create a deadlock. Please try running this program on cores 0-3 and verify its operation.