Path: utzoo!mnetor!uunet!lll-winken!lll-lcc!ames!hao!gatech!purdue!i.cc.purdue.edu!j.cc.purdue.edu!pur-ee!uiucdcs!uiucdcsb!kenny
From: kenny@uiucdcsb.cs.uiuc.edu
Newsgroups: comp.arch
Subject: Re: Marketing MIPS
Message-ID: <165100013@uiucdcsb>
Date: 25 Feb 88 03:27:00 GMT
References: <175121@<1988Feb18>
Lines: 24
Nf-ID: #R:<1988Feb18:175121:uiucdcsb:165100013:000:1180
Nf-From: uiucdcsb.cs.uiuc.edu!kenny    Feb 24 21:27:00 1988


/* Written  9:43 am  Feb 23, 1988 by davidsen@steinmetz.steinmetz.UUCP in uiucdcsb:comp.arch */
 As I recall there was once a system from Heneywell which was billed as
a "timesharing" machine. It would run batch, but slowly. It was quite a
bit cheaper than the general purpose machine.

 In the dispatcher was an instruction which set the priority of batch
jobs lower on that model. Addition of a single NOP would perform an
upgrade, at a cost of thousands of dollars.

/* End of text from uiucdcsb:comp.arch */

It was the 66/07, 66/17 and 66/27 (There could have been a /37, but
wasn't.)  It didn't finagle the job priorities, but rather it disabled
cache memory when going to non-timesharing programs (batch or
transaction-processing).

The NOP *could* be patched in, but the dispatcher initialization code
would put it in for you if you were on the right sort of box -- a
change the field engineer could make by clipping one diode and causing
a bit to appear in the configuration status word.

Oh, yes, and what is `Heneywell?' A melange of General Electric and
Honeywell (I preferred General Honey myself -- what's it now, General
Nippon Honey?  Anyone got a better handle?)