[Jcp: Re: [decvax!pur-ee!p: UNIBUS problems on new processors]]

[utzoo!decvax!ucbvax!unix-wizards]

>From mike.bmd70 at BRL Fri Sep  4 14:14:05 1981

----- Forwarded message # 1:

Date:      4 Sep 81 14:44:42-EDT (Fri)
From:      Jcp at bmd70
To:        Michael Muuss <mike at bmd70>
Subject:   Re:  [decvax!pur-ee!p: UNIBUS problems on new processors]

Mike:

	From the drawings I have seen, (and the 11/24 I have seen), the
UNIBUS on both 11/24's and 11/44's goes thru a PAX (Program Address
Extension) Module if you are using more than 128Kw of memory.  This
produces similar problems as the UNIBUS MAP on an 11/70, and the other
extended memory processors.  In additon, on the 11/24, memory managment
is done by a VLSI chip, which incurs additional delay.

	On an 11/70, the KT (mem. manager) introduces ~90nsec of delay
	and the UNIBUS MAP is good for another 90.

	My impression is that on the 11/24, the KT is good for ~100nsec
	and the UNIBUS MAP for another ~100nsec, making the 11/24 UNIBUS
	good for a little less than 1Mhz.  (11/70 is 1Mhz).

	Incidentally, the same is true for an 11/23 (Qbus), except that
	there is no PAX for the 11/23, yet.  (Yes, DEC is planning to
	introduce one...)

	You may want to relay this info.

						-JCP-

----- End of forwarded messages