wg@opal.cs.tu-berlin.de (Wolfgang Grieskamp) (02/14/91)
Hello, What is the current state of GRIP? Is it still alive? I've access only to some older papers: [1] Simon L. Peyton Jones, Chris Clack, Jon Salkild, Mark Hardie: "GRIP - a high performance architecture for parallel graph reduction", Proc IFIP Conference on Funct. Prog. Lang., Portland, Ed. G.Kahn, LNCS 274 (1987). [2] Simon L. Peyton Jones, Chris Clack, Jon Salkild: "High-performance parallel graph reduction", PARLE '89, Vol. 1, Ed. E.Odjik, M.Rem, J.-C.Syre, Springer 89. Does the machine tends to a product-quality fixpoint? Is there a user community? Are there other applications then classical graph reduction? Hints are very appreciated. -- Wolfgang Grieskamp wg@opal.cs.tu-berlin.de tub!tubopal!wg wg%opal@DB0TUI11.BITNET
kh@cs.glasgow.ac.uk (Kevin Hammond) (02/17/91)
In article <2636@opal.cs.tu-berlin.de> wg@opal.cs.tu-berlin.de writes: >What is the current state of GRIP? Is it still alive? GRIP is up and running LML and Haskell programs, after a hiatus when it was being moved to Glasgow. We've recently been getting some performance results which are quite exciting, e.g.: Fine-grained parallelism can be very nearly as good as coarse-grained parallelism [Nijmegen]; GRIP is insensitive to LIFO/FIFO spark pool throttling strategies [PDCS]. >Does the machine tends to a product-quality fixpoint? We use the Spineless Tagless G-Machine for *fast* graph reduction. The hardware is generally very sound (no machine faults detected for over 6 months). The software is still under development, but is improving in quality. I have run decent-sized functional programs (1000-2000 line programs) on the machine [but no large-scale results have been published yet, maybe FPCA?]. >Is there a user community? Yes, there is a project (FLARE) which is aimed at exploiting GRIP for real applications, and we have a (small) number of local users. But it's really not yet in a state where we could throw it open to all comers. > Are there other applications then classical graph >reduction? Hints are very appreciated. The BRAVE project at Essex is using the GRIP hardware to implement a parallel logic programming language. The hardware and OS are essentially general purpose (though you might have to remicrocode the IMUs for a new application area). Kevin [Nijmegen] Hammond, K. and Peyton Jones, S.L., "Some Early Experiments with the GRIP Parallel Reducer", Proc. Workshop on Implementation of Functional Languages on Parallel Architectures, Nijmegen, The Netherlands, June 1990. Technical Report no 90-16, October 1990, Dept of Informatics, University of Nijmegen. [PDCS] Hammond, K. and Peyton Jones, S.L., "Profiling Scheduling Strategies on GRIP", submitted to the Journal of Parallel and Distributed Computing, 1991. I can make PostScript versions of either of these papers available if you're interested. -- This Signature Intentionally Left Blank. E-mail: kh@cs.glasgow.ac.uk