R. Thakur, W. Gropp, and E. Lusk. A Case for Using MPI's Derived Datatypes to Improve I/O Performance. In Proc. of SC98: High Performance Networking and Computing, 1998.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....exactly the opposite happens for the SRB transfers, the forecaster is able in both instances to make accurate predictions. Regarding related work, performance analysis and scheduling of data intensive applications are described by the ADR group from University of Maryland in [23] and by Thakur in [21]. However, they focus on parallel data servers running over local area networks. Performance monitoring and forecasting of wide area networks is discussed in works such as the NWS [25, 26] GloPerf [13] and [7, 4] The Netlogger system [22] presents a pro ling framework for distributed storage ....

R. Thakur, W. Gropp, and E. Lusk. A Case for Using MPI's Derived Datatypes to Improve I/O Performance. In Proc. of SC98: High Performance Networking and Computing, 1998.

....performance. We demonstrate the performance and portability with performance results for three applications on five different parallel machines. We note that ROMIO can perform the optimizations described in this paper only if users provide complete access information in a single function call. In [14] we explained how users can do so by using MPI s derived datatypes to create file views and by using MPI IO s collective I O functions whenever possible. In this paper we describe the optimizations in detail and provide extensive performance results. The rest of this paper is organized as follows. ....

R. Thakur, W. Gropp, and E. Lusk. A Case for Using MPI's Derived Datatypes to Improve I/O Performance. In Proceedings of SC98: High Performance Networking and Computing, November 1998.

....on its I O needs. The same I O access pattern, however, can be presented to the I O system in different ways, depending on which I O functions the application uses and how. The different ways of expressing I O access patterns in MPIIO can be classified into four levels, level 0 through level 3 [57]. We explain this classification with the help of the same example we considered in previous sections, namely, accessing a distributed array from a file (Figure 13.3) The principle applies to other access patterns as well. Recall that in this example the local array of each process is not ....

Rajeev Thakur, William Gropp, and Ewing Lusk. A case for using MPI's derived datatypes to improve I/O performance. In Proceedings of SC98: High Performance Networking and Computing, November 1998.

....either MPI IO or databases. The user can specify the data with a high level description, together with annotations, and use a similar API for data retrieval. SDM internally translates the user s request into appropriate MPI IO calls, including creating MPI derived datatypes for noncontiguous data [33]. SDM also interacts with the database when necessary, by using embedded SQL functions. ffl Convenient Data Retrieval Capability. SDM allows the user to specify names and other attributes to be associated with a data set. SDM internally selects a file name into which the data will be stored; the ....

Rajeev Thakur, William Gropp, and Ewing Lusk. A Case for Using MPI's Derived Datatypes to Improve I/O Performance. In Proceedings of SC98: High Performance Networking and Computing, November 1998.

No context found.

R. Thakur, W. Gropp, and E. Lusk. A case for using MPI's derived data types to improve I/O performance. In Proc. SC98: High Performance Networking and Computing, Nov 1998.

....benchmark [5] and 3. an unstructured code (which we call UNSTRUC) written by Larry Schoof and Wilbur Johnson of Sandia National Laboratories. We note that ROMIO can perform the optimizations described in this paper only if users provide complete access information in a single function call. In [17], we explained how users can do so by using MPI s derived datatypes to create file views and by using the collectiveI O functions whenever possible. In this paper, we describe the optimizations in detail and provide extensive performance results. The rest of this paper is organized as follows. ....

R. Thakur, W. Gropp, and E. Lusk. A Case for Using MPI's Derived Datatypes to Improve I/O Performance. In Proceedings of SC98: High Performance Networking and Computing, November 1998. To appear.

....astrophysics application template (DIST3D) the NAS BTIO benchmark, and an unstructured code (UNSTRUC) on five different parallel machines: HP Exemplar, IBM SP, Intel Paragon, NEC SX 4, and SGI Origin2000. Preliminary versions of portions of this work were published in two conference papers [36, 37]. 1 Introduction I O is a major bottleneck in many parallel applications. Although the I O subsystems of parallel machines may be designed for high performance, a large number of applications achieve only about a tenth or less of the peak I O bandwidth. The main reason for poor ....

R. Thakur, W. Gropp, and E. Lusk. A Case for Using MPI's Derived Datatypes to Improve I/O Performance. In Proceedings of SC98: High Performance Networking and Computing, November 1998.

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