Forum Discussion
Altera_Forum
Honored Contributor
7 years agoThanks HRZ for the great response,
Here I have some questions: 1) I totally understood the effect of unrolling, but I wanna know more about its implementation. When you unroll a loop, does that mean it will replicate the loop pipeline structure spatially and can run "X" number of iterations in parallel? This is what comes to my mind right now. 2) You said the latency of my "for" loop is something around 60, but when I unroll the for-loop, the loop trip count does not stay as 16384. In fact, it will be 16384/X. Don't you think for X equal to 128, the result loop trip count is still small to hide the pipeline latency? Thanks, Saman