I am currently experimenting with some small asm sequences and investigating the performance event counters. I have a small loop consisting of a (add,add,mov,sub,jnb) sequence, being repeated x number of times.
My initial guess was that I would be able to see the number of uops sent to the instruction dispatch queue (IDQ), could be observed by looking at the decoded icache (IDQ:DSB_UOPS) event. However, the results were not as I expected:
Can anyone please point out where these uops are sent to the backend from, and which event I should check to observe that?