PVR%bgerug51.earn@NSS.CS.UCL.AC.UK (Patrick Van Renterghem / Transputer Lab) (01/16/89)
In response to all the queries about Parasoft's Express: Parasoft Corp., 27415 Trabuco Circle, Mission Viejo Ca, 92692 contact: John Salmon, (818) 356-6788, johns@tybalt.caltech.edu or: Larry Lesser, (714) 380-9739, 'their sales manager' Express is a parallel processing toolkit, not only for transputers, but also for the Intel IPSC hypercube, the BBN and other distributed memory machines. This makes applications somewhat portable from one loosely coupled machine to another, what is a great thing as these machines are upgraded (diff. comp. to comm. ratio, diff. topology, #links, ...) or replaced by entirely different models. Express is not an operating system as Helios tries to be for the transputer. It is more a sort of environment to make the programming of loosely coupled machines easier. It consists of: - primitives to communicate data between processors - a transparent I/O system, so that every node in the system can do I/O - a device independent graphics package (NO, not X-Windows) - a source level debugger, at the moment only for C - performance monitors, profilers, ... As I already said some time ago, their overview documentation is pretty philosophical, with not a lot of information on the actual implementation. But now that I have seen it running (at Caltech), I am convinced it has some advantages over Helios. The most important one is that it incorporates some ideas of dynamic load balancing, that have been developed at Caltech. Dynamic load balancing on loosely coupled machines is one of the main factors that may lead to a breakthrough for these systems. As fully automatic tools are still a long way off, semi-automatic tools that are independent of the architecture of the machine are the best we have at the moment. The second thing that makes the Parasoft Express differ from others is that it is being used by the Parasoft people to port other applications to these machines. As a matter of fact, they just completed the port of the well-known and widely used neural network program from Neuralware Inc. They are working on other applications as well (graphics, IC design, ...). My personal opinion is that it is a very interesting product and after using Parsytec's Helios for a couple of months, we may decide to use this one in stead. It costs less than $1000, call +1 (714) 380-9739 for more information. They do not have a European distributor yet, but I am willing to give more information about it. Just invite me to a conference and I will give a more elaborate discussion of Express and the ideas behind it. ***************************************************************************** * Patrick Van Renterghem, BITNET: pvr@bgerug51.bitnet * * R&D Assistant, EDU: pvr%bgerug51.bitnet@cunyvm.cuny.edu * * State University of Ghent UUCP: mcvax!bgerug51.bitnet!pvr * * Belgium JANET: PVR%earn.bgerug51@NSS.CS.UCL.AC.UK * * * * Automatic Control Lab/The Transputer Lab, Tel: +32 91 22 57 55 ext. 313 * * State University of Ghent, Fax: +32 91 22 85 91 * * Grotesteenweg Noord 2, * * B-9710 Ghent-Zwijnaarde, Belgium * *****************************************************************************
PVR@BGERUG51.BITNET (Patrick Van Renterghem / Transputer Lab) (01/16/89)
In response to all the queries about Parasoft's Express: Parasoft Corp., 27415 Trabuco Circle, Mission Viejo Ca, 92692 contact: John Salmon, (818) 356-6788, johns@tybalt.caltech.edu or: Larry Lesser, (714) 380-9739, 'their sales manager' Express is a parallel processing toolkit, not only for transputers, but also for the Intel IPSC hypercube, the BBN and other distributed memory machines. This makes applications somewhat portable from one loosely coupled machine to another, what is a great thing as these machines are upgraded (diff. comp. to comm. ratio, diff. topology, #links, ...) or replaced by entirely different models. Express is not an operating system as Helios tries to be for the transputer. It is more a sort of environment to make the programming of loosely coupled machines easier. It consists of: - primitives to communicate data between processors - a transparent I/O system, so that every node in the system can do I/O - a device independent graphics package (NO, not X-Windows) - a source level debugger, at the moment only for C - performance monitors, profilers, ... As I already said some time ago, their overview documentation is pretty philosophical, with not a lot of information on the actual implementation. But now that I have seen it running (at Caltech), I am convinced it has some advantages over Helios. The most important one is that it incorporates some ideas of dynamic load balancing, that have been developed at Caltech. Dynamic load balancing on loosely coupled machines is one of the main factors that may lead to a breakthrough for these systems. As fully automatic tools are still a long way off, semi-automatic tools that are independent of the architecture of the machine are the best we have at the moment. The second thing that makes the Parasoft Express differ from others is that it is being used by the Parasoft people to port other applications to these machines. As a matter of fact, they just completed the port of the well-known and widely used neural network program from Neuralware Inc. They are working on other applications as well (graphics, IC design, ...). My personal opinion is that it is a very interesting product and after using Parsytec's Helios for a couple of months, we may decide to use this one in stead. It costs less than $1000, call +1 (714) 380-9739 for more information. They do not have a European distributor yet, but I am willing to give more information about it. Just invite me to a conference and I will give a more elaborate discussion of Express and the ideas behind it. ***************************************************************************** * Patrick Van Renterghem, BITNET: pvr@bgerug51.bitnet * * R&D Assistant, EDU: pvr%bgerug51.bitnet@cunyvm.cuny.edu * * State University of Ghent UUCP: mcvax!bgerug51.bitnet!pvr * * Belgium JANET: PVR%earn.bgerug51@NSS.CS.UCL.AC.UK * * * * Automatic Control Lab/The Transputer Lab, Tel: +32 91 22 57 55 ext. 313 * * State University of Ghent, Fax: +32 91 22 85 91 * * Grotesteenweg Noord 2, * * B-9710 Ghent-Zwijnaarde, Belgium * *****************************************************************************