CCS has an option to profile the code(Run->Clock). The clock shows irregular cycle counts for any function executed. For example, NOP on any core is supposed to take 1 cycle, even with deep pipeline(C7x), the cycle should be relevant. For 11 NOPs, I expect 11 cycles or something close, but CCS reports 5897 cycles. For comparison, there is a time stamp counter in C7x(TSC), it reports 12 cycles for the execution. 

Steps to reproduce on AM62D/AM62A(C7x):

• Replace the hello_world.c file in examples/hello_world. Compile it in debug mode.
• Load it using CCS.

• Enable the profile clock(Run->Clock->Enable)

• Set a breakpoint before the NOP and at the end of NOP.