You are doing the test wrong. Btest generates random data, so to generate this random data for the test, you are maxing out the CPU. You can't run the test on the same router which you are testing. Find a 12 core Xeon PC to generate the Btest
Why is there such a big difference of CPU use between TCP and UDP tests of the same speed?
I have an RB1200 set to Local/Remote Tx of 3M/3M and a UDP test uses 2% in profiler at most (overall CPU use is 4-9%). When I change it to TCP however overall CPU is always 100%.
One odd thing is that the top process when running a TCP test is not always btest. It seems to alternate between dhcp, ppp, hotspot, unclassified.