On May 17, 2013, at 1:43 PM, John Wilson wrote: ... The DEQNA was such a mistake. Why DEC would break from eons of tradition and make the Q and U versions of something be not in the slightest compatible is beyond me. Apparently fitting on a dual-height card was more important. I think it does have a CPU (i8051?) but its ROM is mostly full of the PDP-11 boot/diag code (since the boot is way too big to fit in a typical PDP-11 boot PROM so all that does is suck the real boot out of the 8051). So they didn't have space for doing MOP in firmware. I don't think the QNA had any form of microprocessor on it. It's basically a brand X Ethernet NIC chip, plus interface logic to talk to the Qbus. And yes, the requirement was to make it small. Remember that the only alternative at the time was the UNA, which is 6 times as big! The big issue with the QNA is not that it's different. Different is easy to handle, just write another driver. The real issue is that it's unreliable. Even after 12 revisions (ECO level L) it still didn't work right. The LQA exists simply because it had become clear that the QNA would never work, no matter how many ECOs were created for it. How much of that came from faults in the NIC chip itself, and how much from the logic around it, I have no idea. QNA is not the only horrible Ethernet interface DEC designed. The CNA (for the Pro bus) is, if anything, even worse. It uses the Intel 82586, which has a horribly misdesigned excuse for a programming interface, full of race conditions that are obvious to anyone qualified to pass Comp Sci 101. paul