epmp ports strange error rate/traffic

[silverstr8p]

I finally deployed my DC 8 port unit today (running latest stable 1.4.x firmware) at a tower site and I had a couple problems:

my ePMP radios behaved strange. My thoughput dropped to *maybe* half of what it was on Mikrotik gear, and it started showing tons of errors on one link, and no RX throughput on the other link, both of which were transferring traffic, just about half of what it was (according to an upstream MRTG server).

Also, the fan seemed to go nuts and just get faster and faster, and I guess that means the power consumption must go up.

I took it out and put the Mikrotik back in with a piggyback gig PoE switch (both fanless) and everything went back to normal and the power consumption dropped and throughput doubled again. The old set up has been working for years, but I wanted an integrated gig PoE switch. My ports are all set to Auto negotiate, and otherwise just default settings. The Netonix power consumption seems much higher than my other gear, so it would take a solar panel just to run this switch, which is hard since I have limited space. It takes as much power as an ePMP radio. Am I doing something wrong?