Review of Taming The Memory Hogs.

Essay by kalexanderUniversity, Master'sA+, December 2003

download word file, 7 pages 0.0

Angela Brown was a PhD candidate at Carnegie Mellon University until 2002 with Dr. Todd C. Mowry as her advisor in the Computer Science Department. Dr. Angela Brown worked on thesis research in the area of investigating a fully automatic scheme that enables out-of-core applications to explicitly manage their memory requirements by prefetching pages that will be needed in the near future, and informing the operating system about pages that can be replaced [1]. Dr. Mowry currently directs the research of six graduate students at the university in the areas of computer architecture, compilers, operating systems, parallel processing and database performance [2]. While Dr. Brown was pursuing her degree, she published four papers on the topic of her research. In 2000, at the Fourth Symposium on Operating System Design and Implementation in San Diego, California, Angela Brown presented a paper entitled Taming the Memory Hogs: Using Compiler-Inserted Releases to Manage Physical Memory Intelligently [3].

The paper discusses how to improve the performance of out-of-core applications that consume a great deal of resources running on the same machine as an interactive application without harming the performance of the interactive application. This review summarizes the ideas illustrated in Dr. Brown's paper, compares them to alternate methods and presents opinions.

Applications with datasets that exceed the capacity of physical memory and referred to as out-of-core programs are common in many scientific computing applications. These applications require large amounts of I/O throughout their executions, and when schemes are used to prefetch data into memory, they consume even more resources. This creates a problem when a program of this type is running on the same machine as an interactive program that depends on user interaction and response time. The simple solution of virtual memory for managing requirements of an out-of -core application delivers unacceptable performance...