swm@unccvax.UUCP (swami manohar) (04/18/89)
Hello folks! Sorry for the delay in posting the summary. My comments are enclosed in angle brackets << >> ------------------------------------------------------------------------------- From: David Ellsworth <mcnc!unc!ellswort> Message-Id: <8904072138.AA14569@dopey.cs.unc.edu> Organization: University Of North Carolina, Chapel Hill I took a course in parallel algorithms last semester here from Dr. Tyagi (tyagi@unc.cs.unc.edu); you might want to talk with him. We used the butterfly at Duke and the Sequent at NC State for our projects. Machines you were interested in: Intel hypercube We have a simulator (the Caltech Cosmic cube, I think) here; talk to Jon Leech leech@unc.cs.unc.edu. BBN Butterfly Sequent Balance I suggest that you use the real machines at Duke & NC State respectively; Dr. Tyagi did the negotiation for our class. <<may be more useful for us here in North Carolina>> The connection machine You can get free use of a connection machine from Thinking Machines (paid by either DARPA or NSF, I forget which). Russ Tuck (tuck@unc.cs.unc.edu) knows more about this. Using the real machines instead of simulators requires a good internet connection. However, I think that simulators for most of the machines either don't exist or run very slowly; you would do just as good by using the machine over the internet. << I got an account on the network CM (it is sponsered by DARPA) but have not started using it: I have to order the documentation. ray@think.com was the person I contacted to get an account>> Good luck, David Ellsworth ----------------------------------------------------------------------- >From cs.utexas.edu!top.cis.syr.edu!nasokan@gatech.edu Thu Apr 6 23:21:24 1989 Organization: CIS Dept., Syracuse University <<A number of parallel machines are available with the Northeastern Parallel Computation Center at Syracuse. I think it they include two connection machines , a hypercube etc. (i lost the original message, but Asokan was kind enough to resend me the contact address given below)>> Contact addresses: Carleton E Geckler - geckler@cmx.npac.syr.edu Bill O'Farrel - billo@cmx.npac.syr.edu Asokan ----------------------------------------------------------------------------- >From ephraim@Think.COM Thu Apr 6 10:45:49 1989 Subject: starlisp simulator for CM Date: Thu, 06 Apr 89 10:44:41 EDT Status: R You can obtain a copy of the starlisp simulator for the CM via anonymous ftp from cmns.think.com. Here's a forwarded message from one of the cmns folks here. --- Start of forwarded message --- gandalf%gingold> ftp cmns.think.com Connected to cmns.think.com. 220 cmns-vax.think.com FTP server (Ultrix Version 4.10 Mon Nov 7 15:52:11 EST 1988) ready. Name (cmns.think.com:gingold): anonymous 331 Guest login ok, send ident as password. Password:guest 230 Guest login ok, access restrictions apply. ftp> ls 200 PORT command successful. 150 Opening data connection for /bin/ls (192.31.177.24,1465) (0 bytes). CM_Lights archives bin etc gnu starsim-f15-sharfile usr vision 226 Transfer complete. 71 bytes received in .31 seconds (.22 Kbytes/s) ftp> get starsim-f15-sharfile 200 PORT command successful. 150 Opening data connection for starsim-f15-sharfile (192.31.177.24,1466) (1034296 bytes). 226 Transfer complete. local: starsim-f15-sharfile remote: starsim-f15-sharfile 1064100 bytes received in 39 seconds (27 Kbytes/s) ftp> bye 221 Goodbye. Now, I don't know a lot about the simulator and how to use it. It should run with Common Lisp, though. Getting documentation for starlisp probably happens through our standard channels for that stuff. You might want to reply to this guy and give him the info. (Ok, I should have done it. I'm either lazy or overburdened. Maybe both.) - -dg ------- End of Forwarded Message << Here are some details on the CM documentation: There are two sets, one for symbolic computing ($130) and another for numeric computing ($170). Both sets have the following manuals: CM-2 technical summary, CM front-end subsystems, CM Parallel instruction set (PARIS), and CM programming in LISP. the numeric computing set has the CM programming in C* extra.>> --------------------------------------------------------------------------- >From shannon@intelisc.intel.com Sat Apr 15 07:40:11 1989 From: Shannon Nelson <shannon@intelisc.intel.com> Organization: Intel Scientific Computer, Beaverton, OR We have a simulator for our iPSC/2. For ordering information, call either your local sales rep, Nolan Drevitch @ 404-594-0518, or call our Customer Support line @ 1-800-421-CUBE. ------------------------------------------------------------------------------ >From mailrus!ames!amdahl!shs@gatech.edu Fri Apr 7 17:39:53 1989 From: mailrus!ames!uts.amdahl.com!shs@gatech.edu (Steve Schoettler) Organization: Amdahl Corporation, Sunnyvale CA Free to universities: PARET: An Integrated, Visual Tool for the Study of Parallel Systems send email to edmark@vax135.att.com for licensing agreement and more info. RCS: The Rochester Connectionist Simulator send mail to Mrs. Rose Peet Department of Computer Science Rochester, NY 14627 I have used PARET. It's very nice. The current version runs on a Sun under suntools (4.0) or X10. Steve ---------------------------------------------------------------------------- >From jerbil@csvax.caltech.edu Thu Apr 6 20:42:42 1989 Organization: California Institute of Technology From: Stainless Steel Gerbil (Joe Beckenbach) <jerbil@csvax.caltech.edu> For a hypercube of a general message-passing machine, how about the ghostcubes from Caltech's Cosmic Environment? Release information is available here at Caltech, and I think the normal route is to send mail to chuck@vlsi.caltech.edu [the head of the department and the hypercube effort] asking about it. If you've got access to an Arpanet host, things will go more quickly, but I think it can be distributed on tape. The Cosmic Environment runs on various Unixes, including 4.1BSD, 4.2BSD, and SunOS 3.x. 4.3tahoe and SunOS 4.x will be later on, after the CS department machines are upgraded. If you need additional help, just send e-mail to me. --------------------------------------------------------------------- Organization: University of Oklahoma, Norman, OK From: Longshot <texsun!uokmax!randy@uokmax@suntri> Sender: texsun!uokmax!randy@uokmax@suntri Last Summer, I did some undergraduate research here at OU, using a package called MPSIM, which simulated a hypercube architecture. I know it runs on several flavors of UNIX (we use BSD4.2). It was written by T. H. Dunigan, but I have since lost his address. << does anyone have more information on the above?>> ------------------------------------------------------------------------ >From montnaro@sprite.crd.ge.com Fri Apr 7 14:07:44 1989 From: montnaro@sprite.crd.ge.com (Skip Montanaro) Reply-To: <montanaro@sprite.crd.ge.com> (Skip Montanaro) Send a mail message to Dave Ray (ray@think.com) to see if you can get one or more accounts on Thinking Machines' front-end machines. They have a DARPA-sponsored project, the Connection Machine Network Service (CMNS), which may be available for teaching purposes. Skip Montanaro (montanaro@sprite.crd.ge.com) ------------------------------------------------------------------------- Comments: First, thanks to everyone who responded to my request. (I sent mail to everyone. hope you got it). It is really a pleasure to be part of this wonderful group. Most of the above pertains to hypercubes or the connection machine (there is only one reference to the BBN butterfly, but that is about a real machine at Duke). I am sure there are simulators for a number of shared-memory multiprocessors out there. Please respond if you have information on these. I have to wait till May to start following up on the above pointers. I will post another summary in early June. More pointers are welcome in the meanwhile. Thanks very much Swami -- Swami Manohar Email: mcnc!unccvax!swm@mcnc.org Computer Science Department Phone: 704-547-4883 University of North Carolina Charlotte, NC 28223