K. A. Yelick. Programming Models for Irregular Applications. In Proceedings of the Workshop on Languages, Compilers and Run-Time Environments for Distributed Memory Multiprocessors, volume 28(1). ACM SIGPLAN Notices, January 1993. 10  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... that similar annotations are required when using a data parallel language (e.g. HPF) For other classes of applications, specifically highly irregular applications, this approach may not be feasible since these applications require the use of dynamic data structures and dynamic load balancing [23]. For such applications, it is necessary to extend the runtime system to support dynamic reconfiguration. By 4 building these reconfiguration procedures into the runtime system, they can be made transparent to the programmer who uses the run time system library. Application level Checkpointing ....

K. A. Yelick. Programming Models for Irregular Applications. In Proceedings of the Workshop on Languages, Compilers and Run-Time Environments for Distributed Memory Multiprocessors, volume 28(1). ACM SIGPLAN Notices, January 1993. 10

.... that similar annotations are required when using a data parallel language (e.g. HPF) For other classes of applications, specifically highly irregular applications, this approach may not be feasible since these applications require the use of dynamic data structures and dynamic load balancing [33]. For such applications, it is necessary to extend the run time system to support dynamic reconfiguration. By building these reconfiguration procedures into the run time system, they can be made transparent to the programmer who uses the run time system library. 3.1.1 Support for Efficient ....

Kathleen A. Yelick. Programming Models for Irregular Applications. In Proceedings of the Workshop on Languages, Compilers and Run-Time Environments for Distributed Memory Multiprocessors, volume 28(1). ACM SIGPLAN Notices, January 1993.

....performance and optimising tools can access all levels ( Rapunzel s hair ) Parts are available from ftp: ftp.cs.berkeley.edu ucb CASTLE. Currently implemented are Active messages (see separate entry for CMAM) and a parallel extension to the C language (see separate entry for Split C) See [75, 123, 45, 46, 120] for more details. 8.1 pC Name: pC Description: pC is an extension to C which gives a form of data parallelism. It contains many of the ideas found in HPF (High Performance Fortran) Compilers are available for a number of parallel machines, which take a pC program and convert it to a ....

K Yelick. Programming models for irregular applications. In Proceedings of the Workshop on Languages, Compilers and Run-Time Environments for Distributed Memory Multiprocessors, Boulder, CO, October 1992. To appear in SIGPLAN notices.

....queue for pairs of polynomials. The priority corresponds to the heuristic merit of a pair. Since this is only a heuristic, it need not be adhered to exactly. In the parallel algorithm, gpq is as much a scheduling control as a data structure. The distributed task queue by Wen et al., described in [10], is designed for applications that can be broken down into tasks. The suggested paradigm is that the task queue starts with some number of tasks, processors dequeue tasks and perform them, possibly generating more tasks. Our parallel algorithm naturally fits into this model. 4.2.1 ....

Katherine Yelick. Programming models for irregular applications. In Proceedings of the Workshop on Languages, Compilers and Run-Time Environments for Distributed Memory Multiprocessors, October 1992. To appear in SIGPLAN notices.

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