Sun-Spots-Request@RICE.EDU.UUCP (10/28/87)
SUN-SPOTS DIGEST Wednesday, 28 October 1987 Volume 5 : Issue 55
Today's Topics:
Re: Sun multiprocessor rumor
Re: linpack and eispack libraries
Re: Experiences with CDC SMD disks on SUN 3s
Fixing Monochrome (or Color) Monitors
FTP problems between Suns and Symbolics LISP machines...
"text table full" and scsi disks...
"fig" 1.3 problems
When is a ND server overloaded?
Has anyone implemented CD-ROM for a Sun?
----------------------------------------------------------------------
Date: 22 Oct 87 02:44:53 GMT
From: sundc!potomac!jtn@uunet.uu.net (John T. Nelson)
Subject: Re: Sun multiprocessor rumor
> From: Neil G. Sullivan <ngs@cs.duke.edu>
>
> I have heard an unsubstantiated rumor that Sun is soon coming out with a
> multiprocessor architecture machine. Could someone substantiate this
> rumor and if possible fill in with some more information.
Haven't heard a thing. Speculation on what Sun will do next is always fun
though. I think Sun will have to do something, either to incorporate
multiprocessing architectures or specialized processors (LISP engines on a
chip) into their computers. Users running LISP environments with
sophisticated windowing and presentation systems will never be satisified
with the the number of crunchons they're getting from current Suns or even
future machines based on 68030 chips. A future Sun product might be
plug-in LISP processor/parallel processor boards (given that VME can
support this sort of thing).
Sun advocates "the network as the computer." Perhaps they will one day
introduce a (closely-coupled) parallel machine that will perform much of
the brute-force processing. Symbolics has introduced something like this,
and as I understand it their array-processor is a honey.
Sun may find the VME bus limiting, so they might begin designing a much
larger, parallel, machine from the ground up. This machine would probably
run an integrated Unix/Window/multi-language environment in a more elegant
way than the current window system is set up. But this is going to be
difficult because Unix has this historical tendency to think of things as
"ttys." I'm sure its been tuff getting where they are now. If the trend
continues, when will Sun Unix cease to be Unix?
Enough speculation...
------------------------------
Date: Thu, 22 Oct 87 16:34:19 cdt
From: rackow@anl-mcs.arpa (Gene Rackow)
Subject: Re: linpack and eispack libraries
Since there are bound several of you looking for software for your
systems, I am sending the top level index on what is available from
netlib. This is an automated software distribution system, NOT a real
person.
------------INDEX-----FOLLOWS-------------
===== general NETLIB index =====
Welcome to netlib, a system for distribution of mathematical software
by electronic mail. This index is the reply you'll get to:
mail netlib@anl-mcs.arpa
send index.
To examine the full index for any library send a request of the form:
send index from eispack.
To search for all software with certain keywords:
find cubic spline.
To search for somebody in Gene Golub's address list:
Who is Joan Doe?
displays entries containing "Joan" and "Doe". (no spelling correction!)
Here are some addition forms a request may take...
send dgeco from linpack
(Retrieves routine DGECO and all routines it calls from the LINPACK library.)
send only dgeco from linpack
(Retrieves just DGECO and not subsidiary routines.)
send dgeco but not dgefa from linpack
(Retrieves DGECO and subsidiaries, but excludes DGEFA and subsidiaries.)
send list of dgeco from linpack
(Retrieves just the file names rather than the contents;
this can be helpful when one already has an entire library and just
wants to know what pieces are needed in a particular application.)
whois france
(Retrieves all addresses of people in the database living in France.)
You may include several requests in a single piece of mail, but put
each on a separate line.
Send the requests to "netlib@anl-mcs.arpa" even though replies appear to
be coming from "netlibd@anl-mcs.arpa". You'll be talking to a program,
so don't expect it to understand much English. If your Unix system
doesn't talk directly to research, you might try forwarding through ihnp4
(AT&T Bell Labs in Chicago) or mcvax (Math Centrum in Amsterdam). Someone
will be paying for long distance phone calls, so be reasonable! Those
with access, possibly through some gateway, to the arpanet can use
netlib@anl-mcs.arpa (at Argonne National Lab) instead of research!netlib
(at Bell Labs in Murray Hill, New Jersey).
The default precision is double; to get single, prefix the library name
with "s". However, if the library only comes in one precision, that's
what you will be sent. To save space we remove sequence numbers and
maintain a central set of machine dependent constants. Otherwise the
codes, which are almost all in Fortran, are as received from the authors.
Bugs found in core libraries like eispack will receive prompt attention;
in general, we will forward comments (and annual lists of recipients) to
the code authors. The "Caveat receptor" you see means: this code may not
be worth more than you are paying for it!
-------quick summary of contents---------
alliant - set of programs collected from Alliant users
apollo - set of programs collected from Apollo users
benchmark - various benchmark programs and a summary of timings
bihar - Bjorstad's biharmonic solver
bmp - Brent's multiple precision package
cheney-kincaid - Programs from the book by, Ward Cheney & David Kincaid
conformal - Schwarz-Christoffel codes by Trefethen; Bjorstad+Grosse
core - machine constants, blas
domino - communication and scheduling of multiple tasks; Univ. Maryland
eispack - matrix eigenvalues and vectors
elefunt - Cody and Waite's tests for elementary functions
errata - corrections to numerical books
fishpack - separable elliptic PDEs; Swarztrauber and Sweet
fitpack - Cline's splines under tension
fftpack - Swarztrauber's Fourier transforms
fmm - software from the book by Forsythe, Malcolm, and Moler
fn - Fullerton's special functions
go - "golden oldies" gaussq, zeroin, lowess, ...
harwell - MA28 sparse linear system
hompack - nonlinear equations by homotopy method
itpack - iterative linear system solution by Young and Kincaid
lanczos - Cullum and Willoughby's Lanczos programs
laso - Scott's Lanczos program for eigenvalues of sparse matrices
linpack - gaussian elimination, QR, SVD by Dongarra, Bunch, Moler, Stewart
lp - linear programming
machines - short descriptions of various computers
microscope - Alfeld and Harris' system for discontinuity checking
minpack - nonlinear equations and least squares by More, Garbow, Hillstrom
misc - everything else
ode - ordinary differential equations
odepack - ordinary differential equations from Hindmarsh
sodepack - ordinary differential equations from Hindmarsh(S.P.)
paranoia - Kahan's floating point test
pchip - hermite cubics Fritsch+Carlson
polyhedron - there is a file for each solid.
pltmg - Bank's multigrid code; too large for ordinary mail
port - the public subset of PORT library
pppack - subroutines from de Boor's Practical Guide to Splines
quadpack - univariate quadrature by Piessens, de Donker, Kahaner
siam - typesetting macros for SIAM journal format
slatec - machine constants and error handling package from the Slatec library
specfun - transportable special functions
toeplitz - linear systems in Toeplitz or circulant form by Garbow
toms - Collected Algorithms of the ACM
y12m - sparse linear system (Aarhus)
[[ The last section, which described certain items in more detail, was
omitted for space reasons. Those interested can get more information from
anl-mcs.arpa as per the instructions above. I have been advised to NOT
place any omitted part of this message in the Sun-Spots archives, as the
list of available software changes "constantly." --wnl ]]
------------------------------
Date: Fri, 23 Oct 87 09:50:13 EDT
From: nemnich@nrl-cmsun.arpa
Subject: Re: Experiences with CDC SMD disks on SUN 3s
From: rbthomas@jove.rutgers.edu (Rick Thomas)
Does anyone have experience using CDC 9720 SMD disk drives with XY451
controllers on SUN 3/180 fileservers?
At Thinking Machines, we have been using lots of the CDC 515Mb 8" drives
on Sun 3/280 fileservers with no problems at all.
--Bruce Nemnich, Thinking Machines Corporation, Cambridge, MA
Temporarily at NRL, Washington, DC: +1 202 767 9044
bruce@think.com, ihnp4!think!bruce, bjn@mitvma.bitnet
------------------------------
Date: Thu 22 Oct 87 21:59:19-EDT
From: C. P. Yeske <CY13@te.cc.cmu.edu>
Subject: Fixing Monochrome (or Color) Monitors
Recently there has been some discussion on where to go for spare parts for
Sun monitors. Carnegie Mellon has an extensive self maintenance (to the
component level) program for the entire Sun product line. I must warn the
net that working on monitors by untrained and unwary personnel can lead to
serious injury or additional damage to equipment. Often monitors have
sections or parts that are tuned to other sections or parts. We often
find that untrained people will spread the problem to other more difficult
to repair areas, (this process sometimes is called "cancer."). In the
case of the transformer, only a competent engineer could determine WHY it
fried -- and fix that problem first.
Please be careful out there.
[[ I second that one. Make sure you know what you are doing! --wnl ]]
Curt Yeske
Technical Administrator
Carnegie Mellon Computing Services
CY13@te.cc.cmu.edu
Disclaimer: The opinions expressed are my own and are not Carnegie Mellon's.
Home-Phone: (412)422-4667 (422-GOOP, -IONS, -HOOP)
Office: UCC A72 x2647
Staff-Assistant: Cathy Hays, x5147, CH1G@te.cc.cmu.edu
------------------------------
Date: 22 Oct 87 03:00:54 GMT
From: sundc!potomac!jtn@uunet.uu.net (John T. Nelson)
Subject: FTP problems between Suns and Symbolics LISP machines...
Hmmmm... it seems I can't run FTP (or anything built on top of TCP/SMTP)
between my Sun III/180 and Symbolics 3670 LISP machine. I just upgraded
to version 3.4 of Sun Unix and now whenever I open a TCP channel between
the two machines, the operation hangs in a "TCP In" state.
The FTP bug fixes provided by Sun (3.4.2) don't help although Sun tells me
that these fixes will solve hanging connections between Suns running 3.4
which is good.
Has anyone else encountered this? Sun has been informed of the problem
but thus far we have jointly come up with no solutions.
------------------------------
Date: Wed, 21 Oct 87 19:11:47 EDT
From: cayman!brad@harvard.harvard.edu (Brad Parker)
Subject: "text table full" and scsi disks...
Two quick questions:
(Sun 3/50 w/SUNOS 3.2)
1. When I run X windows (doesn't everyone? ;-) I sometimes get "text table
full" when I fork new processes. I have *a lot* of processes running
(simple name server, domain name server, CAP aufs server, atalkad, atis,
screen save, etc....).
How to fix this? Simply increase MAXUSERS? or, should I increase the
"fudge" factor which defines ntext? ("24 + MAXUSERS")
2. I'd like to buy another 140Mb disk for my sun. Is SUN the only source
for this? Can I just buy another ESDI disk and stuff it in the box? Does
anyone else make SCSI disks for the sun (ps: I know that SCSI is not a
standard, assume I've written a SCSI driver or two).
Thanks!
------------------------------
Date: Thu, 22 Oct 87 12:43:03 +1000
From: munnari!mulga.oz.au!isaac@uunet.uu.net
Subject: "fig" 1.3 problems
Does anyone know of a later version? In particular, has anyone noticed
that text input in a figure does not come out in exactly the same position
as on the screen? I am using pic -Plw (to a laserwriter) and ditroff,
note though that the same thing occurs on an imagen. Finally, on my wish
list, has anyone made mods to fig to produce different fonts/sizes ... (I
know I can edit the pic file, but this is not the ideal). By the way, why
do people output "pic" and not "ideal" - postscript/psfig would be the
best a-la the recently posted graphedit, but that requires adobe software
which we don't have (too expensive for our department). While I am at it,
anyone use anything other than fig or graphedit (ie non-commercial
products?)
Isaac Balbin
munnari!isaac@uunet.uu.net or
isaac%munnari.oz.au@australia
------------------------------
Date: 22 Oct 87 19:35:22 GMT
From: cos!asp@uunet.uu.net (Andrew S. Partan)
Subject: When is a ND server overloaded?
I have been trying to figure out how many diskless clients I can support
on each of my servers and on each of my ethernets.
Right now I have one 3/280 with 4 3/50s ND clients, and 2 3/280s each with
13 3/50s ND clients, all on one ethernet. I am planning on adding 6 more
3/50 ND clients. Each 3/280 has 2 451 disk controllers, each controller
has 1 super eagle disk. I am running Sun OS 3.4. The 1st 3/280 is my
comm server; the other 2 are ND & NFS servers.
My feeling is that this is very overloaded. The problem is that I:
1) do not know what I really need to measure,
2) am not sure how to measure it, and
3) do not know what the 'danger' levels are (i.e.: the point at
which I should get another server or split my ethernets or
do something like that).
I think that I need to pay attention to:
- the load averages (this probably does not tell me much, if
anything??),
- the disk utilization (io? - I am not sure what I mean by
this),
- the ethernet utilization (I have heard that 37% is the
'danger' point).
Can anybody help? Does anyone know of any papers that tell me all that I
want to know? Does anyone have any programs to help monitor stuff? [I
know of traffic(1) for the ethernet load, but that gives a graphical
display, and I really want to collect statistics over a long period - days
or weeks - and then do some statistical analysis on them.]
I am also not sure what effects 4.0 will have with getting rid of ND.
Thanks for any help that anyone can give me.
--
--asp (Andrew Partan @ Corporation for Open Systems)
-- asp@cos.com
-- {uunet, sundc, decuac, hqda-ai, hadron}!cos!asp
------------------------------
Date: Fri, 23 Oct 87 08:52:21 edt
From: geac!daveb@uunet.uu.net (Dave Collier-Brown)
Subject: Has anyone implemented CD-ROM for a Sun?
Out of sheer, ignorant curiosity ,I should like to know if anyone has
implemented a CD-ROM or WORM disk for a Sun.
--dave (they're available for Macs, you see) c-b
David Collier-Brown. {mnetor|yetti|utgpu}!geac!daveb
Geac Computers International Inc., | Computer Science loses its
350 Steelcase Road,Markham, Ontario, | memory (if not its mind)
CANADA, L3R 1B3 (416) 475-0525 x3279 | every 6 months.
------------------------------
End of SUN-Spots Digest
***********************