gda@creare.creare.UUCP (Gray Abbott) (10/09/89)
With respect to the discussion on parallelism in OOP, is anyone doing work on using OOP languages (e.g. C++) for computationally-intensive applications, such as computational fluid dynamics (CFD), for high-performance hardware (e.g. Cray)? I'm especially interested in the possibility of using C++ to "hide" the vectorization of computations, so that the programs are more transparent, without sacrificing too much in performance (or even improving performance by increasing the amount of code that is effectively vectorized). Know of anything like that?
wayne@schaefer.MATH.WISC.EDU (Rick Wayne) (10/10/89)
In article <GDA.89Oct9114712@creare.creare.UUCP> gda@creare.creare.UUCP (Gray Abbott) writes: > >I'm especially interested >in the possibility of using C++ to "hide" the vectorization of >computations, so that the programs are more transparent, without >sacrificing too much in performance (or even improving performance >by increasing the amount of code that is effectively vectorized). this is what i was trying to get at in the original posting. (thanks for the education on the way, folks!) i was confusing distribution with parallelism; sloppy thinking on my part. however, i keep wondering that if you solve the synchronization problem inherent in distributed code, might you not get parallelism for free? actually, i guess it would probably be the other way around -- make your code parallel, and the rendezvous-ing involved would obviate explicit synchronization. so the programmer still has to sweat decomposition and vectorizing if the compiler doesn't. rw
mccalpin@masig3.ocean.fsu.edu (John D. McCalpin) (10/12/89)
In article <GDA.89Oct9114712@creare.creare.UUCP>, gda@creare.creare.UUCP (Gray Abbott) writes: >With respect to the discussion on parallelism in OOP, is anyone >doing work on using OOP languages (e.g. C++) for computationally-intensive >applications, such as computational fluid dynamics (CFD), for >high-performance hardware (e.g. Cray)? I'm especially interested >in the possibility of using C++ to "hide" the vectorization of >computations, so that the programs are more transparent, without >sacrificing too much in performance (or even improving performance >by increasing the amount of code that is effectively vectorized). >Know of anything like that? I know of two projects along those lines. First, the GIBBS project at Cornell (under Kenneth Wilson) is making use of C++ to write scientific number-crunching codes in a much higher-level syntax. I believe that they are using cfront to generate C code to run on the Cray machines (through Cray's new vectorizing C compiler). Second, there is a project funded by the Institute for Naval Oceanography (in Bay St. Louis, Mississippi), which is writing a large ocean model in C++. I believe that the emphasis is on modularity, so that, for example, the turbulent diffusion calculations could be easily replaced without effecting the rest of the code. Sorry I don't have the names of people involved in these projects.... -- John D. McCalpin - mccalpin@masig1.ocean.fsu.edu mccalpin@scri1.scri.fsu.edu mccalpin@delocn.udel.edu
dixon@sagittarius.crd.ge.com (walt dixon) (10/12/89)
In article MCCALPIN.89Oct11154351@masig3.ocean.fsu.edu John D. McCalpin writes >>With respect to the discussion on parallelism in OOP, is anyone >>doing work on using OOP languages (e.g. C++) for computationally-intensive ... >Second, there is a project funded by the Institute for Naval >Oceanography (in Bay St. Louis, Mississippi), which is writing a large >ocean model in C++. I believe that the emphasis is on modularity, so >that, for example, the turbulent diffusion calculations could be >easily replaced without effecting the rest of the code. >Sorry I don't have the names of people involved in these projects.... The second project is named DAIMS (Data Analysis and Interactive Modeling System). The principal investigators are Tom Keffer and Dale Haidvogel. The other name mentioned in the documentation I have seen is Bruce Eckel. The documentation I've seen gives the following addresses for Bruce Eckel and Tom Keffer eckel@sperm.ocean.washington.edu uw-beaver!sperm.ocean.washington.edu!eckel keffer@sperm.ocean.wasington.edu uw-beaver!sperm.ocean.washington.edu!keffer Walt Dixon {arpa: dixon@crd.ge.com } {us mail: ge-crd } { po box 8 } { schenectady, ny 12301 } {phone: 518-387-5798 } Walt Dixon dixon@crd.ge.com
steve@hubcap.clemson.edu ("Steve" Stevenson) (10/12/89)
From article <MCCALPIN.89Oct11154351@masig3.ocean.fsu.edu>, by mccalpin@masig3.ocean.fsu.edu (John D. McCalpin): > In article <GDA.89Oct9114712@creare.creare.UUCP>, gda@creare.creare.UUCP > (Gray Abbott) writes: > > I know of two projects along those lines. > > First, the GIBBS project at Cornell (under Kenneth Wilson) is making > use of C++ to write scientific number-crunching codes in a much > higher-level syntax. GIBBS has moved to ohio state when Ken Wilson and wife Allyson Brown moved. Allyson is the computer scientist. -- Steve (really "D. E.") Stevenson steve@hubcap.clemson.edu Department of Computer Science, (803)656-5880.mabell Clemson University, Clemson, SC 29634-1906