CCR one core load 100%

What am I doing wrong?
Can’t get CCR1016 to route more than 200Mbps bandwidth.
I have >700 pppoe sessions, nat + queue.

Maybe some misconfiguration?
I have large address-lists for routing marks, maybe firewall eats one core?
How to see profile for all cores, like on x86?
ccr.jpg

Have you BGP? It is not multicore optimized