2 Feb
2002
2 Feb
'02
1:53 a.m.
--- Dave McGuire <mcguire(a)neurotica.com> wrote:
I'd argue that a grant-chain scheme has an advantage over the IRQ
scheme found in Intel hardware - in a PC, you run out of interrupts
pretty quick. In a VAX or PDP-11, you can add devices until you
either run out of convenient vector addresses, or the bus gets too
long to propogate signals reliable (or the AC and DC loading gets
too high).
An example - our 11/750 at work had a TU80 controller, 4 Emulex CS21s
(2 interrupts each) for 64 serial ports (no terminal servers for us!),
3 to 5 COMBOARDs, an RUX50 Unibus RX50 interface, an RL11 RL02 controller,
a DMF32 8-port serial/LPT controller (3 interrupts), a DDCMP interface
for 56Kbps DECnet, and a UDA-50 SDI disk controller (plus three MASSBUS
disk and tape interfaces for our SI9900 controller, the TU78 and our
emergency RM03s). That's 14 Unibus devices in 16 quad-height and hex-
height cards using 19 interrupts. Physically, the cards were in the
internal DD11DK, and in a BA-11 next to the VAX with two DD11DKs - 27
useful slots, if you have some quad-height cards like we did.
Compare that with how many devices you can stuff into a P4, including
built-in I/O.
I am fairly certain that there are only 4 and 9 slot Unibus backplanes. If
there's anything out there that DEC made with more than 9 slots in one
block, part of it is almost certainly where the CPU goes and is not Unibus.
Speaking of CPU and Unibus together, there are a few machines that follow
their own rules - the Unibus slots in a 11/730 and 11/725 (and one model
of PDP-11) are *not* wired identically to the slots in a DD11DK. Our
COMBOARD-I used certain pins for extracting the Unibus signals from the
backplane. It must be heavily modified to fit into one of these Unibuses.
The COMBOARD-II uses the "right" pins and goes into any Unibus without
modification. I don't think there are any DEC cards that have this
limitation, but I wouldn't be surprised if CBD1 wasn't the only 3rd party
card to get it wrong.
-ethan
__________________________________________________
Do You Yahoo!?
Great stuff seeking new owners in Yahoo! Auctions!
http://auctions.yahoo.com
On February 1, Gunther Schadow wrote:
Oooh! Must be something sitting on pin B2 - -15VDC.
I read a web article the other day where the guy
describes the
various forms of the Qbus and he also said that you could fry
certain cards when you stick'em in a wrong version of the Qbus.
This is indeed correct...if you plug an RLV11 into a Q/Q backplane
instead of a Q/CD backplane, for example, you'll let out the magic
smoke. Also, why was
the need for grant continuity cards an advantage?
It's not an "advantage" per se...just what the bus needs. SPC slots..."small peripheral controller"
I believe is the correct
expansion of that acronym...you'll want to grab a pdp11 unibus
processor handbook for that info...it's all in there.
Yes.
Are there
UNIBUS backplanes with more than 9 rows?
None that I've seen. I've seen 4- and 9-slot ones.