Erich Schikuta and Heinz Stockinger. High Performance Cluster Computing: Architectures and Systems, chapter Parallel I/O For Clusters: Methodologies and Systems, pages 439-462. Prentice-Hall International, 1999.  Home/Search   Document Not in Database   Summary   Related Articles

....provided by ViPIOS. Conclusions and prospects for future work are given in section 6. 2 State of the Art In the last few years we saw a strong stimulus on research in the area of parallel I O. The lessons learned can be summarized by the following characteristic goals for ecient I O operations[22]: Maximize the use of available parallel I O devices to increase the bandwidth. 1 Minimize the number of disk read and write operations per device. Minimize the number of I O speci c messages between processes to avoid unnecessary costly communication. Maximize the hit ratio (the ....

....to avoid unnecessary data accesses. This led to the development of both abstract abstract methods and real systems. We will give a survey of both separately. 2. 1 Parallel I O Methods The parallel I O methods can be grouped into application level, I O level and access anticipation methods (see [22]) 2.1.1 Application Level Methods These methods try to organize the main memory objects by mapping the disk space (e.g. bu er) to make disk accesses ef cient. Therefore, these methods are also known as bu ering algorithms. Commonly these methods are realized by runtime libraries, which are ....

Erich Schikuta and Heinz Stockinger. High Performance Cluster Computing: Architectures and Systems, chapter Parallel I/O For Clusters: Methodologies and Systems, pages 439-462. Prentice-Hall International, 1999.

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