pase@ogcvax.UUCP (Douglas M. Pase) (03/24/88)
Many, many thanks to all those who helped out on this request. I really appreciate your kind suggestions. As it turns out, Wim Bohm (by some fortunate coincidence) happened to be here this last weekend, and he was also able to help me. In short, thanks to all of your help, and the smiles of good fortune, I was able to get all the information I need in all its gory detail. Because others might be interested (however doubtful that might be), I am posting a summary of the replies I received. (Sorry it's so long, I cut it down a great deal.) ------------------------------------------------------------------------------ In article <hubcap.1127> segall@steppenwolf.rutgers.edu (Ed Segall) writes: >There is a discussion of the U-Interpreter in chapter 2 of >MIT/LCS/TR-332 "Resource Management for the Tagged Token Dataflow >Architecture" by David E. Culler. This is an MS Thesis from 1985. ------------------------------------------------------------------------------ From: Eugene N. Miya <tektronix!ames!pioneer.arc.nasa.gov!eugene> Have you spoken to Arvind? Some of his papers cover some of the details, but I think more are lacking. Dennis's dataflow bibliograph in mine (the Rdf file) might have some. ------------------------------------------------------------------------------ From: munnari!smokey.ua.oz.au!andrew@uunet.uu.net (Andrew L. Wendelborn) A reasonably detailed reference which explains the basic ideas quite clearly is Arvind, K.P. Gostelow and W. Plouffe "An asynchronous programming lang- uage and computing machine", TR#114A, Dept. Inf. and CS, Uni. Calif. at Irvine, Dec. 1978 this report is probably best obtained these days by writing to Prof. Arvind at MIT (Lab. for CS). A less detailed, but recent, report on that work is in Arvind and R.S. Nikhil "Executing a program on the MIT tagged-token dataflow architecture", Lecture Notes in Computer Science vol. 259, 1987. You might also try Veen, A.H. "Dataflow machine architectures", Computing surveys, Dec. 86. and a special isssue on data flow in IEEE Computing, Feb. 1982. Also Vegdahl, S.R. "A survey of proposed architectures for the execution of functional languages", IEEE Trans. Comp., Dec. 84. ------------------------------------------------------------------------------ From: Marta Kallstrom <marta%research2.computer-science.manchester.ac.uk@nss.cs.ucl.ac.uk> Hi Doug, [...] But the faculty person I work with, Wim Bohm, is going to give a talk at OGC next Friday morning. He is a dataflow expert! Why don't you talk to him? Incidently, I noticed that you posted some articles about virtual time in event-driven distributed systems. That's another issue we're working on. All of the theories break down when applied to our dataflow simulator. Wim thinks he has a solution. We're going to try it. -Marta ------------------------------------------------------------------------------ From: Amos Omondi <omondi@cs.unc.edu> Try: "Generation of Dataflow Graphical Object Code .... " by Gurd, et al, in : Springer-Verlag LNCS 111. Otherwise get the TR "A Multilayered Data Flow Computer Architecture" from Gurd. ------------------------------------------------------------------------------ From: Arthur Veen <mcvax!uva!arthur@uunet.uu.net> Details on tagging are indeed hard to find in the widely available literature. In my Computing Survey article of Dec 1986 I give a rather abstract (but correct) description of the method. If you want more details you can try to find: J.Gurd, J.Glauert & C.C.Kirkham Generation of Datflow Graphical Object Code for the Lapse Programming Language CONPAR81, Conference on Analysing Problem Classes and Programming for Parallel Computing 1981 pp.155-168 On pages 127-140 in the same proceedings there is an article of mine that describes code generation for an hypothetical tagged dataflow machines. These 2 articles are however a bit out of date. The compiler I wrote for the Manchester Dataflow Machine is described in my dissertation which can be ordered from Centre for Mathematics and Computer Science P.O.Box 4079 1009 AB Amsterdam The Netherlands The reference is: The Misconstrued Semicolon: Reconciling imperative languages and dataflow machines CWI Tract 26 Chapter 8 of my dissertation describes dataflow code generation and in conjunction with sections 6.4 and 6.5 should give you probably as much detail as you want. Arthur Veen Parallel Computing Postbus 16775 1001 RG Amsterdam The Netherlands ----------------------------------------------------------------------------- From: Choong Huei Seow <sequent!uwvax!XX.LCS.MIT.EDU!EDDIE.MIT.EDU:CHOONG!ll-xn!> I'm with the Dataflow group at MIT. Catch me around sometime if you need help. Choong ----------------------------------------------------------------------------- Again, thanks to all who contributed to my understanding. -- Doug Pase -- ...ucbvax!tektronix!ogcvax!pase or pase@cse.ogc.edu (CSNet)