Sun-Spots-Request@RICE.EDU (William LeFebvre) (06/09/88)
SUN-SPOTS DIGEST Wednesday, 8 June 1988 Volume 6 : Issue 106 Today's Topics: The archive server again Franz Lisp is available through SUG Re: Xylogics 753 The following SUN3/SUN4 benchmark was submitted by a colleague 3/50, 3/60 keyboard: <return> & <del> sunview window dumper wanted Vt100tool wanted Looking for a graphics editors for Suns Anti-Glare/Polarizing Screen for 19" screens? A question about NeWS availability in Switzerland Sun 3-50 SCSI peripherals? Sun 4: timesharing & I/O limitations? Send contributions to: sun-spots@rice.edu Send subscription add/delete requests to: sun-spots-request@rice.edu Bitnet readers can subscribe directly with the CMS command: TELL LISTSERV AT RICE SUBSCRIBE SUNSPOTS My Full Name Recent backissues are available via anonymous FTP from "titan.rice.edu". For volume X, issue Y, "get sun-spots/vXnY". They are also accessible through the archive server: mail the request "send sun-spots vXnY" to "archive-server@rice.edu" or mail the word "help" to the same address for more information. ---------------------------------------------------------------------- Date: Wed, 8 Jun 88 23:47:02 CDT From: William LeFebvre <phil@Rice.edu> Subject: The archive server again Astronomical archive requests, Batman! Well, it turns out that there is quite possibly another reason that people get an acknowledgement but no later delivery from the archive server. It seems that I haven't checked the queue of pending requests very recently. There are currently over 400 requests waiting for final processing and delivery. The server sends out the smallest requests first, regardless of the order in which they were received. This means that someone with a fairly large request might wait a long time before seeing it. The oldest requests waiting date all the way back to May 9 and they are about 97K in total length. So how did this happen? Well, I think I have a parameter set too low. I'm going to increase it a substantial amount and these things should start going out a little faster. I am also preparing the latest version of the server software for use. This new version should fix a few bugs and have a few features that people have asked for. I should install it sometime next week. So be patient with those large requests. I will probably be mailing out a queue summary to those who have requests still waiting so that you will have some sort of idea where you stand. William LeFebvre <phil@Rice.edu> ------------------------------ Date: Wed, 1 Jun 88 14:40:43 EDT From: yunexus!mike%ists.YORKU.CA@uunet.uu.net Subject: Franz Lisp is available through SUG The PD version of Franz Lisp is available on the 1987 Sun User Group tape. It is reliable, customizable, and very cheap. Many thanks to Barry Shein et. al. for makeing this run. See you Sun salesman for info on how to join the Sun User Group. PS. PD Ingres is on it too. ------------------------------ Date: Tue, 31 May 88 18:19:26 BST From: mcvax!ritd.co.uk!mr@uunet.uu.net Subject: Re: Xylogics 753 Our experience with Xylogics, even on the "wrong" side of the pond, is pretty good. We have had a 7053 under evaluation for a few months running under SunOS 3.2 (Xylogics supplied device driver) on a 3/1260 with a Fujitsu M2333K. No problems. I don't expect any further problems under 4.0 :-). Seriously, I've been assurred by both Sun and Xylogics that the 7053 is fully supported under 4.0. I have confirmed this against a recently arrived 4.0FCS Release Notes. Note that these notes DO NOT mention the Interphase, which up to then I had expected to have a supported device driver. For those interested, the 7053 is a Sun sized card (no adaptor needed) which they commissioned from Xylogics. Sun are uppity about the world at large buying this board, so Xylogics are putting together the 753. The 753 is really the 754 with the 7053 firmware, so that it is functionally identical to the 7053. However, the current 7053 cannot cope with 3Mb/sec drives whereas the 753/754 can (see the new Fujitsu M238xK drives if you wonder why this is relevant). Of course, to get a 753 in to a Sun requires a VME 2-3 adaptor and internal cabling; I gather Xylogics are working hard to supply complete and neat solutions Looks like we are going to find out very soon. BTW, we use lots of Fujitsu M23{33,44}K "bricks" and find them pretty good. Anyone care to speculate why Sun have migrated to (by my standards) bulky units for their new 900Mb drives? BTW2, anyone tried a 754 yet? I don't work for Xylogics or Sun, etc.. Martin Reed, Racal Imaging Systems Ltd uucp: mr@ritd.co.uk,{mcvax!ukc!ritd,sun!sunuk!brains}!mr Global String: +44 252 622144 Paper: 309 Fleet Road, Fleet, Hants, England, GU13 8BU ------------------------------ Date: Thu, 2 Jun 88 21:01:05 BST From: James Davenport <jhd%maths.bath.ac.uk@nss.cs.ucl.ac.uk> Subject: The following SUN3/SUN4 benchmark was submitted by a colleague Prof. R. Sibson - rs@uk.ac.bath.maths These benchmark results are obtained from a program written in FORTRAN 77 to load up a pseudorandom (but repeatable) symmetric matrix, find its eigenvalues and eigenvectors by Householder tridiagonalisation followed by implicit QL iteration, and check each eigenvector by direct multiplication by the matrix. The computation is in single precision except that certain inner products are accumulated in double precision. Sun-3's are running Release 3.5 and the standard f77. The Sun-4 is running Sys4-3.2 and Fortran 1.05. Times (seconds) are for a 100 by 100 matrix, with repetitions of the same computation giving times which differ by less than 5%. Each figure here is for the first such test which was run. No opt -O Sun-3/160, 68881 136.9 85.7 Sun-3/160, fpa 68.8 20.7 Sun-3/260, 68881 113.1 76.9 Sun-3/260, fpa 45.2 15.6 Sun-4/260 38.7 13.5 A possible interpretation of these results is as follows. The program is heavily floating-point intensive, with integer computation involving little more than loop control and array addressing. The Sun-3 68881 results are dominated by the difference in clock rates between the 68881 chips on the two machines, which is not great; optimisation effects a significant improvement by reducing data transfer between the 68020 and 68881 chips. The FPA on the two machines runs at the same rate, but communication with it is more efficient on the Sun-3/260. Because it has a lot of registers, optimisation can dramatically improve performance (this is the biggest -O effect I have ever seen, and certainly does not reflect inept source-coding, which is based closely on Wilkinson and Reinsch). What is puzzling is the poor performance of the Sun-4. On any RISC machine, one would expect optimisation to have a big effect, and it may well be just coincidence that the effect is parallel to that on FPA-equipped Sun-3's. If the optimised code on each architecture is spending most of its time in the Weitek chips, and these are running at similar clock rates, then this could explain the similarity. Presumably the message in that case is that the Weitek chips have now run out of steam! ------------------------------ Date: 2 Jun 88 01:01:40 GMT From: Maarten Litmaath <mcvax!cs.vu.nl!maart@uunet.uu.net> Subject: 3/50, 3/60 keyboard: <return> & <del> Why have the keys mentioned above, been placed so irritatingly close together? The location of the <del> key is 'normally' covered by the extension of the <return> key, to let the latter be more easily spotted. Due to SUN's setup every now and then I have to retype a complete line, after hitting <del> instead of <return>... (Yes, I use <del> as interrupt character, and I guess I'm not the only one!) South-Africa: |Maarten Litmaath @ Free U Amsterdam: revival of the Third Reich |maart@cs.vu.nl, mcvax!botter!ark!maart ------------------------------ Date: Wed, 1 Jun 88 18:38:31 PDT From: ihnp4!homxc!lewisd@ucbvax.berkeley.edu Subject: sunview window dumper wanted We'd like to be able to print the windows of applications running under suntools. Has anyone devised a program that would grab and dump a particular window in some rasterfile format? We could then pick it up and print it. (For X users, I'm looking for an xwd equivalent for the suntool environment.) Also: I can't seem to get screendump to work on any color monitor; it works without any problems on monochrome screens. Ideas? David B. Lewis {ihnp4,allegra,ulysses,rutgers!mtune}!homxc!lewisd "Kenneth, the frequency! What's the frequency, Kenneth?!" [[ There's a program called "dumpregion" in the source archives. It allows the user to select the specific region of the screen that is to be dumped into a rasterfile. It is in "sun-source" and is called "dumpregion.c". --wnl ]] ------------------------------ Date: 31 May 88 15:45:04 GMT From: eplrx7!mcneill@uunet.uu.net (Keith McNeill) Subject: Vt100tool wanted Could someone send me vt100tool for sunview? Please send me mail first so I can get it from the closest site. Thanks Keith D. McNeill E.I. Dupont Co. uunet!eplrx7!mcneill Engineering Physics Lab (302) 695-7395 Wilmington, Delaware 19898 Mail Stop: E357-302 ------------------------------ Date: Thu, 02 Jun 88 17:53:37 -0500 From: Gurudatta Parulkar <guru@flora.wustl.edu> Subject: Looking for a graphics editors for Suns I am looking for a good graphics editor which - can help draw all kinds of figures and pictures - produces postcript output which can be used with other TeX or LaTeX files - can be used from suntools or X (reason for the cross posting) - is a public domain software or a COMMERCIAL product (that is, willing to pay $$) - is considerably better than graphedit or fig Could you suggest anything which meets MOST of these requirements ? (We are exploring Autocad but not sure if it is worth its high cost.) Thanks! -guru Dr. Guru Parulkar Asst Professor guru@flora.wustl.edu Dept of Computer Science parulkar@udel.edu Washington University wucs1!guru@uunet.uu.net St. Louis MO 63130 (314) 889-4621 ------------------------------ Date: Thu, 02 Jun 88 09:49:21 -0400 From: Sandeep Mehta <sxm@philabs.philips.com> Subject: Anti-Glare/Polarizing Screen for 19" screens? I have been looking for a large size anti-glare or polarizing filter for my workstation screen. The standard stuff they sell for PC's and Macs in the computer catalogs does not exceed 12-14", whereas my Sun screen is 19". I know one can buy polarizing sheets in large sizes, but that is a rather expensive solution. I was wondering if someone has any ideas, solutions, or is a large polarizing sheet the best solution ? I have seen a fine nylon mesh being sold for the PC's also, and the vendor claims that it cuts out glare/reflections. Any help is appreciated. Please send me e-mail, and I will post all feasible solutions if there is interest. Thanks. sandeep Sandeep Mehta (914)-945-6478 Robotics & Flexible Automation uunet!philabs!bebop!sxm Philips Laboratories sxm@bebop.philips.com ------------------------------ Date: Thu, 2 Jun 88 20:59 N From: <DIMITRI@CGEUGE51.BITNET> Subject: A question about NeWS availability in Switzerland I have a question-complain conserning the SUN policy(?) in Europe. I am trying (for the last two months) to get the NeWS 1.1 release. Still SUN Suisse can not tell me if they can give it to me, and when. I wonder why that is happening. Is there a problem with Europe? Are there any sites in Europe running NeWS 1.1? What about 1.1b+ ? Will that ever get in Europe? Is true that NeWS is NOT to be distributed with the 4.0 release? And by the way, how many UNIX pieces have been "unbundled" from 4.0? (I heard something about FORTRAN and Pascal) [[ Just about everything except the C compiler has been unbundled for 4.0 (well, so I exaggerate sometimes). --wnl ]] Dimitri Konstantas | UUCP : mcvax!cernvax!cui!dimitri Centre Universitaire d'Informatique | EARN : dimitri@cgeuge51.bitnet Univeristy of Geneva | MHS : dimitri@cui.unige.ch ------------------------------ Date: 31 May 88 17:09:32 GMT From: well!ejf@lll-crg.llnl.gov (Erik James Freed) Subject: Sun 3-50 SCSI peripherals? I know that this has been gone over before but this situation is new for me. Can anyone email me SCSI peripherals both Hard disk and Tape that work witht the Sun3-50 reliably and with no kernal modifications? Thanks in advance. erik freed well!ejf ------------------------------ Date: 02 Jun 88 16:20:19 CDT (Thu) From: albers@umn-cs.cs.umn.edu Subject: Sun 4: timesharing & I/O limitations? Briefly: We're buying a new departmental system to support timesharing and file serving. We'd like to consider a Sun 4, but don't believe that the I/O subsystem can support our needs. I'd like some response from Sun 4/280 sites with timesharing loads. I'll post a synopsis to sun-spots. BACKGROUND The Computer Science Department at the Univeristy of Minnesota -- Twin Cities is retiring a VAX-11/780 (4.3BSD) & DG/MV10000 (DG/UX) and will buy a replacement system. At this point, we've narrowed our selection to one of an Encore Multimax 310, a Sequent Symmetry S27, and one or more Sun 4/280s. Our initial requirements are to support about 40 network pty logins, maybe 10 or so tty logins, and 10-20 Sun clients mounting shared mail/news/source/executable NFS volumes. With user file systems, we will be supporting about 2.4 GB of file systems on this new machine. We expect that the timesharing requirement will decrease as we continue to add workstations, but the NFS serving requirement will grow considerably. Furthermore, this system will be the primary departmental server for uucp, news, mail, BIND 4.8, and etc. We will be making a tentative final selection soon. I'd like very much to go with the Sun, but don't have an reliable evidence that it can handle the load or provide any growth. There has been some discussion in sun-spots describing experience using Sun 3/280s and Sun 4/280s as timesharing systems. At the time of the original postings, most sites had only just received their 4/280s and had little experience with them (January -- March). Now, you sites with 4/280s, now that you have (hopefully) run your systems under some fairly heavy loads and have installed SunOS 4.0.... Here are some questions: 1. What is your configuration (CPU,memory,disk ctrlr type (e.g. Xy451 or Ciprico 3220?),disk drive type, arrangement of disks on controllers, mux ports, ...) 2. What is the maximum reasonable load you have observed on your system with different ratios of user logins and file serving? E.g. logins:NFS traffic at 1:0, 1:1, 0:1, etc. If you can, include the usual subjective guess concerning the kind and ratios of processes the user logins are running (mail, emacs, lisp, c compiles, etc.). 3. What do you know or guess about Ciprico's ability to stay abreast of 4.X upgrades now that Sun has committed to Xylogics and will presumably not cooperate further with Ciprico? (that is, don't want to have to run xy451s until the new xylogics controller is available) 4. How would you trade off the performance costs of ttys vs. ptys? We could front-end an ALM-2 MUX with an Annex or Ciprico box to offload some pty overhead. What is your guess about the relative performance cost of ptys (M ttys = N ptys)? Would M/N be closer to 2.0 or 1.5? (Yes, there are lots of qualifications... but the question is really whether we must front-end the box with Annexes/Ciscos to get the performance we need.) 5. SIMPLE QUESTION: Would you, from your experience, have any hesitation using the following configuration for the purposes I described: Sun 4/280S-64 2 Ethernet interfaces (we may have host-terminal traffic on a separate segment than the NFS traffic... TBD) XY451 with 1 892MB drive XY451 with 2 892MB drives (assumes upgrade to a viable 32-bit path controller asap.) ALM-2 if necessary to support more user sessions when we can't support any more ptys. 6. Extra credit. Do you have any experience with Multimaxes or Symmetries in a similar environment to compare relative performance/functionality? Thanks! Jim Albers, Computer Science Laboratories, CSci Dept, University of Minnesota albers@umn-cs.UUCP, albers@umn-cs.cs.umn.edu ------------------------------ End of SUN-Spots Digest ***********************