aps (04/14/83)
I am corrected by Carl on the point of "distributing UNIBUS
arbitration". It seems that Dave was correct regarding devices not
passing BG? when NPR is asserted.
Date: Tue Apr 12 22:42:01 1983
To: sdcsvax!sdchema!decvax!aps
Subject: stealing bus grants
This is ancient history, but some of us hardware types remember it.
Along about 1972, somebody at DEC decided that PDP-11 Unibus performance
would be improved for NPR (DMA) devices if ordinary interrupt (BG[4-7])
devices would block bus grants if they saw NPR asserted. For this
reason, they devised the M7821 interrupt controller, and noted that
"Only some PDP-11 processors will work with the special circuitry
described above."
The same circuit was put into the DL-11 terminal controller, and probably
other things as time went on. I have never found it to work properly,
and have taken advantage of the jumper to disable it. Apparently it
still does funny things on a VAX unibus, since it does a passive release
of the bus if it receives a BG while NPR is asserted.
Carl Lowenstein
Marine Physical Lab., U.C. San Diego
p.s. Thanks for the UDA50 stuff, I will try it out some time soon, and
let you know how it works.
cdl