Any idea for performance difference between CBC and CTR (aes-256) + sha256

Look like both are with supported HW offload (for example hap ac^2, cap ac .. etc.)
I do not see a big difference, however with both I hit my IPS limit (~40mbits)

Probably speed tests is not so linear (with CTR) but iperf between hosts behind both sites look ok
rb5009 (ros7) <> cap ac (ros6) ( cap ac is same cpu as hap ac^2)

If I not mistake CTR use multiple CPU cores however

1. my test cap ac is with ros 6
2. my isp limit form one side is only 40mbits

I wondering if someone test AES-CTR (ros7 vs ros6 ?)

Regards,