A. Feldmann, T. Gross, D. O'Hallaron, md T. M. Stricker. Subset barrier synchronization on a private-memory parallel systems. In Syrup. on Parallel Algorithms and Architectures, pages 209 28, 992.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....very little research has been done to provide architectural supports in multicomputer networks to implement efficient barrier synchronization. Modern multicomputer networks are using advanced switching techniques like worm hole routing and circuitswitching. Barrier synchronization schemes in [6] are tuned towards iwarp system. The unicast based barrier synchronization for hypercubes [14] does not exploit the benefits of path based routing [11] on wormhole routed networks. Virtual channels are increasingly becoming popular to support adaptive routing [3] and increased throughput [5] It ....

A. Feldmann, T. Gross, D. O'Hallaron, md T. M. Stricker. Subset barrier synchronization on a private-memory parallel systems. In Syrup. on Parallel Algorithms and Architectures, pages 209 28, 992.

....[10] Many hardware, software, and hybrid schemes have been proposed in the literature to efficiently implement barrier synchronization on shared memory multiprocessors [1, 15] with bus and multistage interconnections. Topology and routing specific synchronization schemes have been proposed in [8, 19]. However, very little research has been done to implement fast barrier synchronization on popular k ary n cube networks with wormhole routing. One easier way to implement barrier on a k ary n cube network is to use software messagepassing. Traditionally, the wormhole routed systems have supported ....

A. Feldmann, T. Gross, D. O'Hallaron, and T. Stricker. Subset Barrier Synchronization on a Private Memory Parallel System. In Proceedings of the Symposium on Parallel Algorithms and Architectures, pages 209--218, 1992.

....the mapping of processes to processors may not be known at compile time, iWarp barriers cannot rely on the identifiers (IDs) of each processor to identify different subset barriers. We implemented a flexible barrier synchronization package that allows the synchronization of sets of processors [Feldmann, et al. 92] This package embodies two general communication models that capture important capabilities of modern private memory systems: the bounded buffer broadcast model, based on a global broadcast network, and the anonymous destination message passing model, based on a shared, general purpose ....

Feldmann, A., T. Gross, D. O'Hallaron, and T. Stricker. Subset Barrier Synchronization on Private-Memory Machines. In Proc. SPAA 92, pages 209-218. ACM, San Diego, June, 1992.

....Many hardware, software, and hybrid schemes have been proposed in the literature to efficiently implement barrier synchronization on shared memory multiprocessors [1, 13] with bus and multistage interconnections. Topology and routing specific synchronization schemes have been proposed in [7, 16]. However, very little research has been done to provide architectural supports for the popular k ary n cube networks to implement fast barrier synchronization. Systems like Cray T3D use dedicated tree based networks with barrier registers to provide fast synchronization. However, these schemes ....

A. Feldmann, T. Gross, D. O'Hallaron, and T. Stricker. Subset Barrier Synchronization on a Private Memory Parallel System. In Proceedings of the Symposium on Parallel Algorithms and Architectures, pages 209--218, 1992.

....or user defined functions) where there is involvement from all processes of an user defined group. As defined by the standard, the results may be available to only one member of the group or all the members. The operations can be carried on either scalar or vector data. Barrier synchronization [11, 12, 20, 24] is a special case of this class of operation where there is no data (just an event) and the result is available to all members of the group. Hence, in this paper, without loss of generality, we identify both reduction and barrier synchronization as a single class of reduction operation. Many ....

A. Feldmann, T. Gross, D. O'Hallaron, and T. Stricker. Subset Barrier Synchronization on a Private Memory Parallel System. In Proceedings of the Symposium on Parallel Algorithms and Architectures, pages 209--218, 1992.

....processor contains two clock timers (with a resolution of 8 clocks, i.e. 400 ns on a 20 MHz system) and one of these timers is reserved for the user program. Our implementation sets the user timers of all nodes to a common global time in two steps using a synchronization package developed locally[FGOS92] 3.2 Processing queue state data The monitoring program stores the gathered data in a buffer. At the end of execution or when the buffer fills up, the monitoring program sends this data to the host processor. If the data is transferred after the end of the monitored program s execution, the ....

A. Feldmann, T. Gross, D. O'Hallaron, and T. Stricker. Subset Barrier Synchronization on a Private Memory Parallel System. In Proc. Symposium on Parallel Algorithms and Architectures, San Diego, June 1992. ACM.

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