Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!sun-barr!decwrl!shlump.nac.dec.com!shodha.dec.com!alan
From: alan@shodha.dec.com ( Alan's Home for Wayward Notes File.)
Newsgroups: comp.sys.dec
Subject: Re: vs3520 ucode & processor questions
Message-ID: <533@shodha.dec.com>
Date: 14 Dec 89 16:08:34 GMT
References: <35139@cornell.UUCP> <SAUS.89Dec14000337@media-lab.media.mit.edu>
Organization: Digital Equipment Corp. - Colorado Springs, CO.
Lines: 65

In article <SAUS.89Dec14000337@media-lab.media.mit.edu>, saus@media-lab.media.mit.edu (Mark Sausville) writes:
> In article <35139@cornell.UUCP> kowalski@svax.cs.cornell.edu (Jeffrey Kowalski) writes:
> 
>    Date: 14 Dec 89 01:01:52 GMT
>    Reply-To: kowalski@svax.cs.cornell.edu (Jeffrey Kowalski)
> 		.
> 		.
> 		.
>    [ Is the 3520 a multi-processor or not? ]
> 
> Yes, I hear you.  I have the same question regarding our VAX6320 running
> Ultrix 3.1.  While troubleshooting a problem, we ran without the second 
> CPU for a while.  Guess what, nobody noticed.  Deeply embarrassed the
> people who spec'd the buy.

	All of the ULTRIX multiprocessor support since V2.0
	has been asymetric.  Only one of the processors (the 
	master) takes care of system calls while the other 
	processor(s) only run user mode code.  

	Any given "operation" will only run as fast as the
	fastest CPU.  Where you get the benefit of the 2nd
	CPU is when you start doing two of them at the same
	time; if they don't do any (or many) system calls.

	You're right, ps(1) doesn't give any indication which
	CPU something is running on.  The only distributed
	program that I know that even know how to display time
	in the different CPU modes for all the processors is
	iostat(1).  There is a field in the proc structure that
	tells which processor a process is on.
> 
> I have heard that there exists a piece of code called "the Scheduler Patch"
> which is supposed to vastly improve the utilization of the second CPU
> (at least on 6x20s and it is the same proc, right?).

	If you find out more please let me, as I've never heard
	of it either.  Normal operation of my 8800 is a user's
	CPU bound process running 100% "nice mode" on the slave
	CPU and the master taking care of everybody else.
> 
> The same source reports that this non-product is from Ultrix pre-sales,
> wherever they might be.  I was told by my source that my sales-expletive
> ("We just sell 'em!") could get it for me.  
> 
	Actually this is beginning to sound familiar.  There is
	Software Services support group that contracts with customers
	to specialized drivers and enhancements.  Your sales-person
	should be able to contact them.

> To make it even more bizarre, I've also heard that this thing is supposed
> to be in the phantom maintenance release 3.2, of which no one else has
> ever heard.

	I've never heard of this phantom V3.2 either.

> 
> 				Disappointed,
> 						Mark.
> 
> Mark Sausville                           MIT Media Laboratory


-- 
Alan Rollow				alan@nabeth.enet.dec.com