V. Strumpen. A large-scale metacomputer approach for distributedparallel computing. Procs. HPCN-94, Munich, Springer LNCS 797, (1994), 278--285.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....and their current load, executive processes for setting up the application tasks with the corresponding communication links, and load sensor daemons for determining the CPU work load by running small benchmark routines at regular intervals. An extension of the Parform project, named PARC [10, 38], emphasizes on the efficient management of distributed applications in very large WAN clusters. Based on a tree shaped logical interconnection topology,thePARC system provides a programming interface and tools for the administration of large WAN metacomputers with several hundred nodes. HiCon. ....

V. Strumpen. A large-scale metacomputer approach for distributedparallel computing. Procs. HPCN-94, Munich, Springer LNCS 797, (1994), 278--285.

....QuickThreads package [11] which provides this basic functionality in a portable manner. The runtime system is linked to an application written in the C language. Support for process and communication link management is not offered in the runtime layer, and must be provided by systems such as Parc [20]. This work focusses on clusters distributed over large geographical distances. The latency of longhaul networks connecting these clusters is not expected to decrease by several orders of magnitude in the near future. Network properties such as relatively high error rates and the need for flow ....

....applications with large resource requirements with respect to computational power, memory and communication volume as needed for parallel scientific computations, for example. Furthermore, portability is a design constraint that is indispensible in order to work with large networks of machines [20]. The key component enhancing state of the art systems is a runtime layer that extends the TCP IP protocol suite [19] where it is sandwiched between the transport layer and the application layer. The runtime layer integrates multithreading and a protocol for multiplexing and demultiplexing message ....

V. Strumpen. A Large-Scale Metacomputer Approach for Distributed Parallel Computing. In High-Performance Computing and Networking, LNCS 797, pages 278--285. Springer-Verlag, Munich, April 1994.

....the development of a latency hiding protocol as a refinement of the TCP IP suite [9] By utilizing otherwise wasted protocol wait times for computational tasks, latencies are reduced further, and smaller granularities can be supported. The earlier origin of this work was the development of Parc [22], a platform for network computing in the large. There, for the sake of portability, all protocols are implemented on top of TCP IP and the Berkeley Socket Interface [20] Stream sockets abstract from low level data transfer by offering the stream model to the programmer. However, 5 this ....

....such as the Intel Paragon. Communication latency hiding can also be emulated on workstations by refining the send and receive operations implemented on top of the TCP IP protocol. The ideas about latency hiding arose during the design of a message passing kernel, which is the basis of Parc [22], a platform for network computing in the large. To ensure portability, all protocols are implemented on top of TCP IP and the Berkeley socket interface [20] Although stream sockets abstract from low level data transfer by offering a stream model to the programmer, there are still deficiencies ....

[Article contains additional citation context not shown here]

Strumpen V., "A Large-Scale Metacomputer Approach for Distributed Parallel Computing, " In High-Performance Computing and Networking, Vol. II, Springer-Verlag, Munich, April 18--20, 1994, 278--285.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC