Here is a (belated) summary of the replies I received to my request for information on parallelizing compilers. Thanks to the following people who provided information: Simon N Smith nadkarni@cse.uta.edu (Sachit R. Nadkarni) Qiang Alex Zhao maniattb@cs.rpi.edu lorie@cs.rice.edu (Lorie Liebrock) subhas@cs.wm.edu (Subhas Roy) Christian Lengauer Joe Hummel dcp79807@csie.nctu.edu.tw good@enuxha.eas.asu.edu (David R. Good) Jean-Louis.Pazat@irisa.fr (Jean Louis Pazat) kremer@cs.rice.edu (Uli Kremer) kumar@cat.syr.edu (Pankaj Kumar) lilja@everest.ee.umn.edu (David Lilja) chialing@iti.gov.sg (Lee Chia Ling) Rudi Hackenberg Gholamali Hedayat (JRG ra) cch@dcs.edinburgh.ac.uk Suresh Damodaran Kamal Steven Collins paco@cs.rice.edu (Paul Havlak) Ramakumar Kosuru wismuell@Informatik.TU-Muenchen.DE (Roland Wismueller) Roger Peel Here's the original request: |> I have found lots of articles on vectorizing compilers. I have not, |> however, been able to find any information on compilers which try to |> discover parallelizable code and then attempt to map it to processes |> on a distributed memory machine. Can someone help me to find the |> relevant references? And here are the replies. The first section contains concrete bibliographic references, mostly in BibTeX format. I have not checked for duplicates. The subsequent sections contain general comments. They have been edited down for brevity. --------------------------------------------------------------------------- "Compile Time Techniques for Data Distribution in Distributed Memory Machines," Ramanujam and Sadayappan, IEEE PDS, vol2,No.4, October 1991. /A/ Chapman, B. M. /A/ Herbeck, H. /A/ Zima, H. P. Automatic Support for Data Distribution Proceedings of the Sixth Distributed Memory Computing Conference, April 28 - May 1, 1991, Portland, Oregon, IEEE Computer Society Press, 51-57 /A/ Zima, H. P. /A/ Bast, H. J. /A/ Gerndt M. SUPERB: A tool for semi-automatic MIMD/SIMD parallelization Parallel Computing, No. 6, 1988, 1-18 /A/ Gerndt, M. Parallelization of Multigrid Programs in SUPERB Proc. of the International Workshop on Compilers for Parallel Computers, Dec. 3-5, 1990, Paris, France, 347 - 356 On the Serialisation of Parallel Programs P.H. Welch & G.R.R. Justo, in "Occam and the Transputer - Current Developments", (proceedings of the 14th World Occam and Transputer User Group Technical Meeting, Loughborough, September 1991). J. Edwards (ed), IOS Press (Amsterdam) 1991. ISBN 90 5199 063 4 @InProceedings{Parafrase2, Author = {C. Polychronopoulos and M. Girkar and M. Haghighat and C. Lee and B. Leung and D. Schouten}, Title = {The structure of {Parafrase-2}: An advanced parallelizing compiler for {C} and {Fortran}}, BookTitle = LCPC, Editor = {D. Gelernter and A. Nicolau and D. Padua}, Year = 1990, Publisher = MIT-P} @InCollection{ABCC:Overview, Author={F. Allen and M. Burke and P. Charles and R. Cytron and J. Ferrante}, Title={An Overview of the {PTRAN} Analysis System for Multiprocessing}, BookTitle={Proceedings of the First International Conference on Supercomputing}, Publisher={Springer-Verlag}, Address={Athens, Greece}, Year={1987}} @Article{AlKe:Automatic, Author = {J. R. Allen and K. Kennedy}, Title = {Automatic Translation of {Fortran} Programs to Vector Form}, Journal = TOPLAS, Volume = 9, Number = 4, Pages = {491--542}, Month = Oct, Year = 1987} @InProceedings{AlCK:Automatic, Author = {J. R. Allen and D. Callahan and K. Kennedy}, Title = {Automatic Decomposition of Scientific Programs for Parallel Execution}, BookTitle = POPL87, Address = {Munich, Germany}, Month = Jan, Year = 1987} @string {JOS = "The Journal of Supercomputing"} @Article{Hend:PTOOL, Author = {L. Henderson and R. Hiromoto and O. Lubeck and M. Simmons}, Title = {On The Use of Diagnostic Dependency-Analysis Tools in Parallel Programming: Experiences Using {PTOOL}}, Journal = JOS, Volume = 4, Pages = {83-96}, Year = 1990} @string {TOPDS = "{IEEE} Transactions on Parallel and Distributed Systems"} @Article{PEDJ2, Author = {K. Kennedy and K. S. M\raisebox{.2em}{c}Kinley and C. Tseng}, Title = {Interactive Parallel Programming Using the {ParaScope Editor}}, Journal = TOPDS, Volume = 2, Number = 3, Pages = {329--341}, Month = Jul, Year = 1991} @string {ICS91 = "Proceedings of the 1991 ACM International Conference on Supercomputing"} @InProceedings{KMT:A+T2, Author = {K. Kennedy and K. S. M\raisebox{.2em}{c}Kinley and C. Tseng}, Title = {Analysis and Transformation in the {ParaScope Editor}}, BookTitle = ICS91, Address = Cologne, Month = Jun, Year = 1991} @string {LCPC91 = "Proceedings of the Fourth Workshop on Languages and Compilers for Parallel Computing"} @InProceedings{HKK:Overview, Author = {S. Hiranandani and K. Kennedy and C. Koelbel and U. Kremer and C. Tseng}, Title = {An Overview of the {Fortran D} Programming System}, BookTitle = LCPC91, Address = SC, Month = Aug, Year = 1991} @InCollection{HKT:icase, Author = {S. Hiranandani and K. Kennedy and C. Tseng}, Title = {Compiler Support for Machine-Independent Parallel Programming in {Fortran D}}, BookTitle = {Compilers and Runtime Software for Scalable Multiprocessors}, Editor = {J. Saltz and P. Mehrotra}, Publisher = {Elsevier}, Address = Amsterdam, Year = {to appear 1991}} @string {SC91 = "Proceedings of Supercomputing '91"} Proceedings{HKT:SC91, Author = {S. Hiranandani and K. Kennedy and C. Tseng}, Title = {Compiler Optimizations for {Fortran D} on {MIMD} Distributed-Memory Machines}, BookTitle = SC91, Address = Albuquerque, Month = Nov, Year = 1991} @Article{Nickhil89, author = "A. S. Nickhil and R. S. Nickhil and K. K. Pingali", title = "{I}-structures: Data Structures for Parallel Computing", journal = "ACM trans. on Programming Languages and Systems", year = "1989", volume = "11", number = "4", pages = "598--632" } @InProceedings{Rogers89, author = "A. Rogers and K. Pingali", title = "Process Decomposition Through Locality of Reference", booktitle = "Conference on Programming Language Design and Implementation", year = "1989", pages = "69--80", organization = "ACM", month = "June 21--23" } @TechReport{Reeves89, author = "A. P. Reeves", title = "Paragon: A Programming Paradigm for Multicomputer Systems", institution = "Cornell University", year = "1989", number = "EE-CEG-89-3", month = "January" } @TechReport{Reeves89b, author = "A. L. Cheung and A. P. Reeves", title = "The Paragon Multicomputer Environment: A First Implementation", institution = "Cornell University", year = "1989", number = "EE-CEG-89-9", month = "July" } @TechReport{Reeves90, author = "A. P. Reeves", title = "The Paragon Programming Paradigm and Distributed Memory Multicomputers", institution = "Cornell University", year = "1990", number = "EE-CEG-90-7", month = "June" } @article{Hatcher88, AUTHOR= "M. J. Quinn and P. J. Hatcher and K. C. Jourdenais", TITLE = "Compiling {C}$^*$ Programs for a Hypercube Multicomputer", BOOKTITLE="ICCL90", YEAR= 1990 } @InProceedings{Hatcher90, author = "M. J. Quinn and P. J. Hatcher", title = "Compiling {SIMD} Programs for {MIMD} Architectures", booktitle = "ACM Sigplan (PPEALS)", year = "1990", OPTpages = "57--65" } @article{Hatcher90b, AUTHOR= "M. J. Quinn and P. J. Hatcher", TITLE = "Data Parallel Programming on Multicomputers", JOURNAL="IEEE Trans. on Software Engineering", volume= " ", number= " ", YEAR= 1990, month= "September" } @InProceedings{Rosing88, author = "M. Rosing and R. B. Schnabel and R. Weaver", title = "Dino: Summary and Examples", booktitle = "3' Conference on Hypercubes Concurrent Computers and Applications", year = "1988", pages = "472--481" } @inproceedings{Rosing90, AUTHOR= "M. Rosing and R. P. Weaver", TITLE = "Mapping Data to Processors in Distributed Memory Computations", BOOKTITLE= "5' International Conference on Distributed Memory Computing", YEAR= 1990, month= "April" } @techreport{Rosing90b, AUTHOR= "M. Rosing and R. B. Schnabel and R. P. Weaver", TITLE = "The {DINO} Parallel Programming Language", INSTITUTION= "University of Colorado at Boulder", YEAR= 1990, number= "CU-CS-457-90" } @techreport{Rosing92, AUTHOR= "M. Rosing and R. B. Schnabel", TITLE = "Efficient Language Constructs for Large Parallel Programs - An Overview of Dino2", INSTITUTION= "University of Colorado at Boulder", YEAR= 1992, number= "CU-CS-578-92" } @techreport{Mehrotra89, AUTHOR= "P. Mehrotra and J. {Van Rosendale}", TITLE = "Parallel Language Constructs for Tensor Product Computations on Loosely Coupled Architectures", INSTITUTION= "ICASE", YEAR= 1989, number= "89-41", month= "September" } %KALI90a @Techreport{Koelbel90, author = "C. Koelbel and P. Mehrotra", title = "Supporting Shared Data Structures on Distributed Memory Architectures", institution= "Department of Computer Science, Purdue University", year= 1990, number= "csd-tr 915" } % KALI90b @inproceedings{Koelbel90b, AUTHOR= "C. Koelbel and P. Mehrotra and J. Saltz and H. Berryman", TITLE = "Parallel Loops on Distributed Machines", BOOKTITLE= "5' International Conference on Distributed Memory Computing", YEAR= 1990, month= "April" } % KALI90c @techreport{Mehrotra90, AUTHOR= "P. Mehrotra, J. {Van Rosendale}", TITLE = "Programming Distributed Memory Architectures Using Kali", INSTITUTION= "ICASE", YEAR= "1990", number= " ", note= "Draft" } @article{Saltz90, AUTHOR= "J. Saltz and K. Crowley and R. Mirchandaney and H. Berryman", TITLE = "Run-Time Scheduling and Execution of Loops on Message Passing Machines" , JOURNAL="Journal of Parallel and Distributed Computing", YEAR= "1990", volume= 8, month= "April" } %Kennedy88, @article{Callahan88, AUTHOR= "D. Callahan and K. Kennedy", TITLE = "Compiling Programs for Distributed-Memory Multiprocessors", JOURNAL="Journal of Supercomputing", YEAR= "1988", volume= 2, pages= "151--169" } @TechReport{FortranD91, author = "S. Hirananandani and K. Kennedy and C. Koelbel and U.Kremer and C-W. Tseng", title = "{An Overview of Fortran D Programming System}", year = "1991", number = "TR91121", institution = "CRPC", address = "RICE University", month = "March" } %SUPERB88, @Article{Zima88, author = "H. P. Zima and H.-J. Bast and M. Gerndt", title = "{SUPERB}: A Tool for Semi-Automatic {MIMD/SIMD} Parallelization", journal = "Parallel Computing", year = "1988", volume = " ", number = "6", pages = "1-18" } @article{Gerndt90b, author = {Michael Gerndt}, journal = {Concurrency Practice and Experience}, title = {Updating Distributed Variables in Local Computations}, year = {1990} } @TechReport{Vienna91, author = "B. Chapman and P. Mehrotra and H. Zima", title = "{Vienna Fortran: A Fortran Language Extension for Distributed Memory Multiprocessors}", institution = "ICASE", year = "1991", number = "91-72", month = "September" } %BOOSTER90, @inproceedings{Paalvast90, AUTHOR= "E. M. Paalvast and A. J. {Van Gemund}", TITLE = "A Method for Parallel Program Generation with an Application to the {\sl Booster} Language", BOOKTITLE="International Conference on Supercomputing", pages= "457--469", YEAR= 1990, month= "June" } %Booster90b, @techreport{Paalvast90b, author = "Edwin M.R. Paalvast and Leo C. Breebaart and Henk J. Sips", address = "Delft University of Tecnology -- The Netherlands", institution = "TNO Institute of applied science", number = "90 ITI 1864", title = "The {\sc {Booster}} Language: a User Language Reference", year = "1990" } @Misc{poster-ifip89, author = "Fran{\c c}oise Andr{\'e} and Henry Thomas", title = "{The Pandore System}", howpublished = "IFIP working conference on Decentralized Systems", year = "1989", month = "December", note = "Poster Presentation" } @InProceedings{dmcc5-90, author = "Fran{\c c}oise Andr{\'e} and Jean-Louis Pazat and Henry Thomas", title = "{Data Distribution with Pandore}", booktitle = "The Fifth Distributed Memory Computing Conference", year = "1990", month = "April" } @TechReport{publi-inria90, author = "Fran{\c c}oise Andr\'{e} and Jean-Louis Pazat and Henry Thomas", title = "{Pandore: A System to Manage Data Distribution}", institution = "INRIA", year = "1990", type = "Rapport de Recherche", number = "1195", month = "March" } @TechReport{publi-interne90, author = "Fran{\c c}oise Andr\'{e} and Jean-Louis Pazat and Henry Thomas", title = "{Pandore: A System to Manage Data Distribution}", institution = "IRISA", year = "1990", number = "519", month = "February" } @InProceedings{amsterdam90, author = "Fran{\c c}oise Andr\'{e} and Jean-Louis Pazat and Henry Thomas", title = "{Pandore: A System to Manage Data Distribution}", booktitle = "International Conference on Supercomputing", year = "1990", organization = "ACM", month = "June 11-15" } @InProceedings{amsterdam90b, author = "F. Andr\'{e} and J.L. Pazat and H. Thomas", title = "{Pandore: A System to Manage Data Distribution}", booktitle = "International Conference on Supercomputing", year = "1990", organization = "ACM", month = "June 11-15" } @InProceedings{icase90, author = "Fran\c{c}oise Andr\'e and Jean-Louis Pazat and Henry Thomas", title = "{PANDORE}: A System to Manage Data Distribution", booktitle = "Workshop on Compiler for Distributed Memory Machines", year = "1990", organization = "ICASE (NASA Langley Research Center)", month = "May" } @InProceedings{paris90, author = "Fran\c{c}oise Andr\'e and Claude Jard and Jean-Louis Pazat and Henry Thomas", title = "Data-driven Distribution of Programs", booktitle = "International Workshop on Compiler for Parallel Computers", year = "1990", organization = "Ecoles des Mines de Paris", month = "December" } @InProceedings{munich91, author = "Jean-Louis Pazat", title = "Code Generation for Data Parallel Programs on DMPCs", booktitle = "European Distributed Memory Computers Conference", year = "1991", OPTorganization = "", OPTpublisher = "", OPTaddress = "", OPTmonth = "April", OPTnote = "" } @PhdThesis{hthomas91, author = "H. Thomas", title = "Une approche de la compilation de programmes s\'{e}quentiels pour machines \`{a} m\'{e}moire distribu\'{e}e", school = "IFSIC/Universit\'{e} de Rennes I", year = "1991", month = "June" } @TechReport{publi-interne91, author = "Fran{\c c}oise Andr\'{e} and Olivier Ch\'{e}ron and Jean-Louis Pazat and Henry Thomas", title = "Efficient Code Generation for Distributed Memory Machines", institution = "IRISA", year = "1991", number = "596", month = "July" } @InProceedings{londres91, author = "Fran{\c c}oise Andr\'{e} and Olivier Ch\'{e}ron and Jean-Louis Pazat and Henry Thomas", title = "Efficient Code Generation for Distributed Memory Machines", booktitle = "Parallel Computing '91", year = "1992", organization = "Parallel Computing Society", publisher = "Elsevier Science Publishers B.V.", month = "September" } @TechReport{publi-interne92, author = "Fran{\c c}oise Andr\'{e} and Olivier Ch\'{e}ron and Jean-Louis Pazat", title = "Compiling Sequential Programs for Distributed Memory Parallel Computers with Pandore II", institution = "IRISA/INRIA", year = "1992", number = "651", month = "April" } @phdthesis{Tseng89, AUTHOR= "P.-S. Tseng", TITLE="A Parallelizing Compiler for Distributed-Memory Parallel Computers", SCHOOL= "CMU", YEAR= 1989 } %POSC90, @inproceedings{Sarkar90, AUTHOR= "V. Sarkar and D. Cann", TITLE = "A Partitioning and Optimizing {{\large S}ISAL} Compiler", BOOKTITLE="Int. Conf. on Supercomputing", YEAR= 1990, pages= "148--163", month= "June" } %OXYGEN90, @inproceedings{Ruhl90, AUTHOR= "R. Ruhl and M. Annaratone", TITLE = "Parallelization of {FORTRAN} {C}ode on Distributed Memory Parallel Processors", BOOKTITLE="Int. Conf. on Supercomputing", pages= "342--353", YEAR= 1990, month= "June" } %ASPAR90, @inproceedings{Ikudome90, AUTHOR= "K. Ikudome and G. C. Fox and A. Kolawa and J. W. Flower", TITLE = "An Automatic and Symbolic Parallelization System for Distributed Memory Parallel Computers", BOOKTITLE="5' Distributed Memory Computing Conference", YEAR= 1990, month= "April" } %SPOT90, @inproceedings{Socha90, AUTHOR= "D. G. Socha", TITLE = "An Approach to Compiling Single-point Iterative Programs for Distributed Memory Computers", BOOKTITLE="5' Distributed Memory Computing Conference", YEAR= 1990, month= "April" } %HYPERTASKING90, @inproceedings{Barber90, AUTHOR= "M. Barber", TITLE = "Hypertasking: Automatic Data Parallel Domain Decomposition on the Intel Supercomputer", BOOKTITLE="5' Distributed Memory Computing Conference", YEAR= 1990, month= "April" } %MIMDIZER90, @inproceedings{Sawdayi90, AUTHOR= "R. Sawdayi and G. Wagenbreth and J. Williamson", TITLE = "{MIMD}izer: Functional and Data Decomposition; Creating Parallel Programs from Scratch, Transforming Existing Fortran Programs to Parallel ", BOOKTITLE="Icase Workshop", YEAR= 1990, month= "May" } @InProceedings{pri92a, author = "T. Priol and Z. Lahjomri", title = "Experiments With Shared Virtual Memory and Message-Passing on iPSC/2 Hypercube", booktitle = "International Conference on Parallel Processing", year = "1992", pages = "To appear", month = "August" } @InProceedings{lah92, author = "Z. Lahjomri and T. Priol", title = "KOAN: a Shared Virtual Memory for the iPSC/2 hypercube", booktitle = "CONPAR/VAPP92", year = "1992", month = "September" } @InProceedings{bod92, author = "F. Bodin and T. Priol", title = "Overview of the KOAN Programming Environment for the iPSC/2 and Performance evaluation of the BECAUSE Test Program 2.5.1", booktitle = "Because Workshop", year = "1992", month = "October" } @book{zima:compilers, author = "Barbara Chapman and Hans Zima", publisher = "Addison-{W}esley", series = "Frontier Series", title = "Supercompilers for {P}arallel and {V}ector {C}omputers", year = "1990", isbn = "0-201-17560-6" } @Book{szymanski:91, editor = "B. K. Szymanski", title = "Parallel Functional Programming Languages and Environments", publisher = "ACM Press", address = "New York", year = 1991 } @Unpublished{mckenney:szymanski, author = "B. McKenney and B. K. Szymanski", title = "Generating Parallel Code for SIMD Machines", note = "Submitted to ACM LOPAS" % may be published by now, check } @Article{szymanski:prywes:88, author = "Szymanski, B. and Prywes, N.", title = "Efficient Handling of Data Structures in Definitional Languages", journal = "Science of Computer Programming", number = 3, year = 1988, pages = "221-45" } @TechReport{sinharoy:szymanski:91, author = "B. Sinharoy and B. K. Szymanski", title = "Complexity Issues of the Alignment Problem and the Closest Vectors in a Lattice", institution = "Dept. of Comp. Sc., Rensselaer Polytechnic Institute", number = "91-10", month = "May", year = 1991 } @Book{szymanski:91, editor = "B. K. Szymanski", title = "Parallel Functional Programming Languages and Environments", publisher = "ACM Press", address = "New York", year = 1991 } @InProceedings{lu:chen:90, author = "Lee-Chung Lu and Marina C. Chen", title = "Subdomain Dependence Test for Massive Parallelism", booktitle = "Proceedings of the Supercomputing '90 Conference", year = 1990, publisher = "IEEE Computer Society Press", address = "Los Alamitos, CA" } @Article{li:chen:91a, author = "Jinke Li and Marina Chen", title = "The Data Alignment Phase in Compiling Programs for Distributed Memory Machines", journal = "J. of Parallel and Distributed Computing", year = 1991, volume = 13, pages = "213--221" } @Article{le:chen:91b, author = "Jinke Li and Marina Chen", title = "Compiling Communication-Efficient Programs for Massively Parallel Machines", journal = "IEEE Transactions on Parallel and Distributed Systems", year = 1991, volume = 2, number = 3, pages = "361--376", month = "July" } @InBook{chen:choo:li, author = "Marina Chen and Young-il Choo and Jingke Li", title = "Parallel Functional Programming Languages and Environments", chapter = "Crystal: Theory and Pragmatics of Compiling Efficient Parallel Code", publisher = "ACM Press", year = 1991, editor = "B. Szymanski" } @InProceedings{yang:choo:91, author = "Allan Yang and Young-il Choo", title = "Parallel-Program Transformation Using a Metalanguage", booktitle = "Proc.~of the Third ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming", year = 1991 } @InProceedings{chen:choo:91, author = "Marina Chen and Young-il Choo", title = "Compiling Crystal for Massively Parallel Machines", booktitle = "Proceedings of the Workshop on Compilation of (Symbolic) Languages for Parallel Computers", year = 1991, publisher = "Argonne National Laboratory Report ANL-91/34", month = "November" } @InProceedings{yang:choo:92a, author = "Allan Yang and Young-il Choo", title = "Formal Derivation of an Efficient Parallel Gauss-Seidel Method on a Mesh of Processors", booktitle = "Proceedings of the 6th International Parallel Processing Symposium", year = 1992, publisher = "IEEE Computer Society Press", month = "March" } @InProceedings{yang:choo:92b, author = "Allan Yang and Young-il Choo", title = "Metalinguistic Features for Formal Parallel-Program Transformation", booktitle = "Proceedings of the 4th IEEE International Conference on Computer Languages", year = 1992, publisher = "IEEE Computer Society Press", month = "April" } @TechReport{li:chen:89, author = "J. Li and M. Chen", title = "Index Domain Alignment: Minimizing Cost of Cross-Referencing Between Distributed Arrays", institution = "Dept. of Comp. Sc., Yale, Univ.", number = "YALEU/DCS/TR-725", month = "Nov.", year = 1989 } @TechReport{gilbert:schreiber:90, author = "J. R. Gilbert and R. Schreiber", title = "Optimal Expression Evaluation for Data parallel Architectures", institution = "RIACS", number = "90.15", month = "April", year = 1990 } @Article{knobe:lukas:steele:90, author = "K. Knobe, J. D. Lukas and G. L. Steele Jr.", title = "Data Optimization: Allocation of Arrays to Reduce Communication on SIMD Machines", journal = "Journal of Parallel and Distributed Computing", volume = 8, year = 1990, pages = "102-118" } ------------------------------------------------------------------------------- >From lorie@cs.rice.edu Fri Nov 20 23:05:24 1992 What you probably need to look for is automatic alignment and distribution for distributed memory architectures. There are a number of parallel compilers that require alignment and distribution information then convert the program to SPMD form. For example the Rice distributed memory compiler does this. If you want a reference on this compiler send email to tlc@cs.rice.edu and ask for the tech reports CRPC-TR91121 and CRPC-TR91132. My dissertation deals with automatic alignment and distribution in Fortran D for a large class of applications. An early description of the work (before we really started into the guts of the research) is available in CRPC-TR91166 from the above source. You will find in these references pointers to other related work. ------------------------------------------------------------------------------- >From jhummel@esp.ICS.UCI.EDU Sun Nov 22 00:35:08 1992 1. Look at Fortran D work by Kennedy, et al., at Rice University. This is fortran for distributed memory machines (now called high-performance fortran or HPF). 2. Rogers and Pingali at Cornell did some compiler work. 3. A system called Kali, authors ?. 4. A system called Pandore, authors ?. You should be able to look in ICPP, Supercomputing, or Sigplan PLDI conferences for this and related stuff. ------------------------------------------------------------------------------- >From Jean-Louis.Pazat@irisa.fr Mon Nov 23 17:06:15 1992 There are many projects about "parallelizing compilers" for distributed memory machines. Ther are two kinds of compilers: Data distribution directed compilers (DDD): FortranD, Pandore, Vienna Fortran, HPF compilers Code distribution directed compilers: (CDD) DDD Compilers parallelize and distribute sequential code according to a data distribution specified by the user; CDD compilers try to distribute independent iterations among processors, most of them use a global memory; but this memory can be a virtual memory (see KOAN for example). At IRISA, we are working on a DDD Compiler PANDORE II (see the bib. part of this message) 2) About profiling information: We are also working in this area. Yves MAHEO (PhD student), (maheo@irisa.fr) is working with me in this field. We have used profiling techniques to find why in some cases the parallelization is not efficient. Informations are collected by the run-time designed for our compiler (PANDORE II -> C (NX/2) ) on an iPSC/2. We also use a sophisticated trace analysis system (ECHIDNA) wich uses the ESTELLE language. (We also have a code generator for the ESTELLE language). Theses informations are of interest to find where are tight synchronizations in the generated code. Cyrille Bareau (PhD student), (bareau@irisa.fr)is working with Claude Jard (jard@irisa.fr) on this subject. We think that it seems very difficult to have a precise enough feed back from theses tools to drive a compiler. Work is in progress to show some global information to the user, driving the compiler is the next step. ------------------------------------------------------------------------------- >From kremer@cs.rice.edu Mon Nov 23 19:00:24 1992 you might want to look at work done here at Rice in the context of the Fortran D programming system. Several other groups are/have been looking at the problem of compiling programs for distributed memory multiprocessors. Just go through the Proceedings of the following conferences: - International Conference of Supercomputing (ICS) - Supercomputing - HCCA4, DMCC5, DMCC6, ... - Sigplan PLDI - Sigplan PPOPP ------------------------------------------------------------------------------- >From kumar@cat.syr.edu Mon Nov 23 20:36:10 1992 In the most recent conference on supercomputing "SUPERCOMPUTING 92" just held in Minnesota, USA there was a paper on compiling fortran programs for distributed memory machines. The paper is from NPAC, Syracuse University, Syracuse NY 13210 USA. We at Syracuse University are developing a Fortran compiler for MIMD machines. There are lot of technical reports, and you should get in touch with Lisa Deo (lsd@nova.npac.syr.edu) to get the technical reports and their listings. Other places which are also working is CRPC, Rice University, Houston, Texax, USA. Contact Ken Kennedy or Charles Koelbel. Also, you can contact Piyush Mehrotra or Joel Saltz at ICASE, NASA Langley Research Center, Hampton, VA 23665, USA. Most of the papers for compiling techniques for MIMD parallel computers are from these people only. ------------------------------------------------------------------------------- >From J.C.Highfield@lut.ac.uk Mon Nov 23 21:36:58 1992 I suggest you try looking at stuff relating to the SUPRENUM (spelling?) machine - I think it is a parallel vector-processor machine. Hence its compilers need just about every technique you can think of. I recall there is a book by H. Zima and B. Chapman on the subject of vector and parallel compilers (Supercompilers for parallel and vector computers??), so you could start searching on their names. The book is heavily into dependence relations etc., covers vector machines, but for parallel machines I think it only considers loop-parallelism. ------------------------------------------------------------------------------- From: paco@cs.rice.edu (Paul Havlak) Parafrase is a vectorizing/parallelizing Fortran-to-Fortran compiler developed at the University of Illinois CSRD. Subsequent projects at CSRD include Parafrase-2 (an interactive programming environment), Cedar (a parallel computer with shared-memory clusters, with its own KAP-based compiler), and several Lisp-oriented projects. Parafrase-2 (at least) has some PDG-based intermediate representations. The PTRAN group at IBM Watson Research Center also uses a PDG-based representation for partitioning. Of the two major projects at Rice, PFC did not use control dependences, but ParaScope uses the PDG and additional information. In the early-to-mid 1980s, Kennedy and Allen's group at Rice developed a vectorizing/parallelizing Fortran-to-Fortran compiler (Parallel Fortan Converter == PFC) and a dependence browser (PTOOL), running on IBM mainframes. PTOOL is now defunct. PFC relied on IF-conversion to convert control dependences to data dependences. The ParaScope programming environment builds control and data dependences for interactive transformation of sequential and parallel Fortran. The ParaScope editor can use dependence graphs from PFC/PSERVE. Recent work at Rice on compilation of Fortran D (a precursor to High Performance Fortran) uses the same ParaScope infrastructure. ------------------------------------------------------------------------------- >From kosuru@point.cs.uwm.edu Fri Dec 4 01:04:44 1992 IEEE transactions on Distributed Systems is a good starting place to look for articles on parallel Comp on Distributed architectures. There are several articles on this topic. There exists good books: Parallel Programming and Compilers (Polychronopolous) SuperCompilers for Parallel and Vector Computers (Hans Zima and Barbara Chapman) Both are good Summary books. ------------------------------------------------------------------------------- From: "Qiang Alex Zhao" Have you tried SR? Here's a description of SR: SR (Synchronizing Resources) is a language for writing concurrent programs. It is described in "The SR Programming Language: Concurrency in Practice" by Gregory R. Andrews and Ronald A. Olsson (Benjamin/Cummings, 1993, ISBN 0-8053-0088-0). An overview of Version 1 of the language and implementation appeared in the January, 1988, issue of TOPLAS (ACM Transactions on Programming Languages and Systems 10,1, 51-86). The main language constructs are resources and operations. Resources encapsulate processes and variables they share; operations provide the primary mechanism for process interaction. SR provides a novel integration of the mechanisms for invoking and servicing operations. Consequently, all of local and remote procedure call, rendezvous, message passing, dynamic process creation, multicast, and semaphores are supported. New features in Version 2 include shared global variables and operations, circular imports, real numbers and math functions, formatted I/O, a more general syntax with fewer special cases, a more robust system with better error checking, true multiprocessing when run on a Silicon Graphics Iris or Sequent Symmetry, and many additional enhancements. SR has been used at a number of universities and labs for course work and research projects involving concurrent programming. It has been used in concurrent programming courses to reinforce concepts with small programming projects and with larger projects such as experiments with parallel algorithms, replicated databases, distributed simulations, and parts of distributed operating systems such as file systems and command interpreters. SR has also been used as a tool in several masters theses and doctoral dissertations to conduct experiments in parallel and distributed programming and to implement larger systems such as a system for mixed language programming, one for distributed implementation of graph algorithms, experiments with load balancing algorithms, and experiments with upcall program structures. Version 2 is tested on Sun4, Sun3, DECstation, SGI Iris, HP RISC and 9000/300, NeXT, and Sequent Symmetry platforms. Code is also included for the DG AViiON, IBM RS/6000, DEC VAX, Encore Multimax, Apollo DN, and others. FTP: cs.arizona.edu: /sr ---------------------------------------------------------------------------- -- Christian.Collberg@dna.lth.se Department of Computer Science, Lund University, BOX 118, S-221 00 LUND, Sweden