This position paper proposes the use of an object-oriented database to model the high-level aspects of distributed systems. Components of the system that are modeled by objects are applications, services, libraries, f...
详细信息
ISBN:
(纸本)0818622652
This position paper proposes the use of an object-oriented database to model the high-level aspects of distributed systems. Components of the system that are modeled by objects are applications, services, libraries, files, IPC channels, hosts, peripherals and users. In this approach a database schema defines the possible interactions among the objects in the system, and the contents of the database model the current state of the system. Queries on the database can be used to answer questions about the system, to construct communication paths between objects, and to choose the appropriate location and medium for the display of information. There are also general applications to system administration, maintenance, performance tuning, and debugging.
The author presents novel designs of optimal VLSI sorters which combine rotate-sort with enumeration-sort. One major contribution of the present work is the development of an efficient index-mapping methodology to con...
详细信息
ISBN:
(纸本)0818621583
The author presents novel designs of optimal VLSI sorters which combine rotate-sort with enumeration-sort. One major contribution of the present work is the development of an efficient index-mapping methodology to construct simple reduced-area shuffle networks which are used to permute data between sorting stages. Moreover, the proposed networks which are used to permute data work are amenable to simple partitioning schemes resulting in multiple chip networks in which each chip has a small number of I/O (input/output) pins.
The authors introduce several implementations of delayed consistency for cache-based systems in the framework of a weakly ordered consistency model. A performance comparison of the delayed protocols with the correspon...
详细信息
ISBN:
(纸本)0818621583
The authors introduce several implementations of delayed consistency for cache-based systems in the framework of a weakly ordered consistency model. A performance comparison of the delayed protocols with the corresponding on-the-fly (non-delayed) consistency protocol is made through execution-driven simulations of four parallel algorithms. The results show that, for parallel programs in which false sharing is a problem, significant reductions in the data miss rate of parallel programs can be obtained with just a small increase in the cost and complexity of the cache system.
The authors present a novel approach that enable compiler optimization of procedure calls. They introduce two interprocedural transformations that move loops across procedure boundaries, exposing them to traditional o...
详细信息
ISBN:
(纸本)0818621583
The authors present a novel approach that enable compiler optimization of procedure calls. They introduce two interprocedural transformations that move loops across procedure boundaries, exposing them to traditional optimizations on loop nests. These transformations are incorporated into a code generation algorithm for a shared-memory multiprocessor. The code generator relies on a machine model to estimate the expected benefits of loop parallelization and parallelism-enhancing transformations. Several transformation strategies are explored, and one that minimizes total execution time is selected. Efficient support of this strategy is provided by an existing interprocedural compilation system. Significant performance improvements due to interprocedural transformation are demonstrated on two scientific programs taken form the Perfect Benchmarks.
The authors describe MTOOL, a software tool for analyzing performance losses in shared memory parallel programs. MTOOL augments a program with low overhead instrumentation which perturbs the program's execution as...
详细信息
ISBN:
(纸本)0818621583
The authors describe MTOOL, a software tool for analyzing performance losses in shared memory parallel programs. MTOOL augments a program with low overhead instrumentation which perturbs the program's execution as little as possible while generating enough information to isolate memory and synchronization bottlenecks. After running the instrumented version of the parallel program, the programmer can use MTOOL's window-based user interface to view memory, synchronization, and compute time bottlenecks at increasing levels of detail from a whole program level down to the level of individual procedures, loops, and synchronization objects. An initial implementation of MTOOL runs on Silicon Graphics multiprocessors and is in use by several groups at Stanford.
This paper presents a variational formulation for real-time simulation of multibody mechanical systems. Static scheduling algorithms are employed to evenly distribute computation on parallel processors. A modified Lag...
详细信息
ISBN:
(纸本)0791808874
This paper presents a variational formulation for real-time simulation of multibody mechanical systems. Static scheduling algorithms are employed to evenly distribute computation on parallel processors. A modified Lagrange multipller elimination method is developed to minimize the computational effort in reducing the dimension of differential-algebraic equations of motion. A general-purpose dynamic simulation program based on the methods developed is shown to simulate a variety of multibody systems faster than real-time on shared memory multiprocessors, enabling real-time operator-in-the-loop simulation of ground vehicles.
Consideration is given to the problem of optimally merging two sets of ordered data such that the mean absolute distance (discrete l1 norm) that a merged element must move in order to properly order the combined set i...
详细信息
ISBN:
(纸本)0780300041
Consideration is given to the problem of optimally merging two sets of ordered data such that the mean absolute distance (discrete l1 norm) that a merged element must move in order to properly order the combined set is minimized. Such a problem is important in the implementation of multidimensional order statistics filters and database applications. A powerful result obtained under the assumption that the elements of both sets are independent and identically distributed and are derived from the same continuous parent distribution is that the optimal merging (using any lp norm) is independent of the parent distribution.
The authors evaluated two automatic tools and one interactive tool using 26 typical NAS (Numerical Aerodynamics Simulation) applications on a Cray Y-MP. It was found that automatic tools produce insufficient performan...
详细信息
ISBN:
(纸本)0818621583
The authors evaluated two automatic tools and one interactive tool using 26 typical NAS (Numerical Aerodynamics Simulation) applications on a Cray Y-MP. It was found that automatic tools produce insufficient performance improvement. Interactive tools can produce better performance because they help users find and eliminate false dependencies. However, simple-minded code transformation has resulted in a significant performance degradation which cancels the speedup obtainable by parallelization. Therefore, tools must perform machine-specific optimizations. The benchmarks contain a large number of small to medium size loops, which limits the performance achievable by parallelizing only loops. Features to assess whether a section of code should be parallelized, vectorized, or left sequential are also necessary.
The authors present parallel implementations of two methods for computing turbulent flow in complex geometries. Both methods are based on the random vortex method, designed to accurately simulate such fluid phenomena ...
详细信息
ISBN:
(纸本)0818621583
The authors present parallel implementations of two methods for computing turbulent flow in complex geometries. Both methods are based on the random vortex method, designed to accurately simulate such fluid phenomena as vortex shedding, merger, and rollup, as well as quantitative features of the flow. The second method, based on a vortex-in-cell method, is an extremely fast version which can offer qualitative portrayal of the dominant fluid structures and mechanisms useful in the design stage. Both methods are nonstandard, containing few of the positive attributes commonly associated with methods that easily lend themselves to massively parallel implementations. They are Lagrangian schemes, in which the position of each computational element is affected by all others at each time step. The efficient execution of these methods on a Connection Machine CM-2 requires parallel N-body solvers, parallel elliptic solvers, and parallel data structures for the adaptive creation of computational elements on the boundary the confining region.
The author describes the formal verification of a portion of an existing mainframe operating system using the UNITY methodology proposed by K. M. Chandy and J. Misra (1988). The UNITY methodology consists of a program...
详细信息
ISBN:
(纸本)0818621338
The author describes the formal verification of a portion of an existing mainframe operating system using the UNITY methodology proposed by K. M. Chandy and J. Misra (1988). The UNITY methodology consists of a programming notation and a logic for specifying and reasoning about properties of parallel programs. The author discusses the steps involved in specifying the operating system, deriving from the specification a correct UNITY program, and using the latter to drive the assembly language implementation of the system. He also comments on his experience in applying a formal methodology in the context of a large software project.
暂无评论