>Second, this design elimanates contention between the Ethernet
>controller and the SCSI controller for IOCTL bus.  Third, the TCDS
>ASIC supports much longer TURBOchannel DMA bursts (64-byte burst
>rather than 16-byte burst).

Interesting.  Clearly true for SCSI, but given the brain-numbing
constraints of the IOCTL lance dma, I'm not sure this would help
network performance.

>Anyway, I'm dreaming of the day when NetBSD/pmax fully utilizes the
>IOCTL ASIC capabilities will be realized.  I firmly believe in there is
>ore room for improvements of TC DECstation and TC Alpha NetBSDs. 

Yes, absolutely. 

I'm still stuck waiting for time to finish reworking the 3MIN
interrupt structure (it needs to recompute ipl masks for both the CPU
and the ioctl asic IMSK), so's I can finish the bus-dma support I
hacked out for the ioctl asic.  Time, time...  

Failing that, I'll have to replace the fan on the MAXINE chassis I
have (the noise is driving me nuts) which, I think, means drilling out
rivets.  Didn't someone do this successfully on a 3100?