A competitive analysis framework for online prefetching and buffer management algorithms in paralleli/osystemsis presented. this framework has widespread applicability to key i/o-bound applications such as external...
详细信息
ISBN:
(纸本)9780897919661
A competitive analysis framework for online prefetching and buffer management algorithms in paralleli/osystemsis presented. this framework has widespread applicability to key i/o-bound applications such as external merging and concurrent playback of multiple video streams. Two realistic lookhead models, global lookhead and local lookhead, are defined. Algorithms NoM and GREED based on these two forms of lookhead are analyzed for shared buffer and distributed buffer configurations, bothof which occur frequently in existing systems. How toimplement the models of lookhead in practice using the simple techniques of forecasting and flushing is presented.
After thoroughly reading the problem statement on the air defense system i asked myself whether the problem statement was clear and unambiguous. in order to test the internal quality and consistency of the problem sta...
详细信息
the main goal of this work is to show how the Formal Description Technique (FDT) Language of Temporal ordering Specification (LoToS), a standard by international organization for Standardization (iSo), can be useful i...
详细信息
ISBN:
(纸本)0818680970
the main goal of this work is to show how the Formal Description Technique (FDT) Language of Temporal ordering Specification (LoToS), a standard by international organization for Standardization (iSo), can be useful in the design of real-time systems. this paper concentrates on the specification phase of a development methodology and uses as an example the parallel real-time kernel RTXC/MP developed by intelligent systemsinternational and being nowadays commercialized as Virtuoso by Eonic systems.
Several distributed applications are characterized by real-time constraints on response times. High-level actions in such distributed applications are modeled by nonatomic events which are collections of atomic events...
ISBN:
(纸本)0818680970
Several distributed applications are characterized by real-time constraints on response times. High-level actions in such distributed applications are modeled by nonatomic events which are collections of atomic events. this paper studies synchronization relations between nonatomic events in such distributed real-time application executions. the relations form a fine-grained hierarchy that can be used to select suitable relations with good properties and clear intuitions, depending on the application. We show the use of the proposed relations in specifying solutions for distributed mutual exclusion and in specifying distributed predicates. As an example of an application, we show how a real-time air defence system that needs to enforce distributed mutual exclusion and specify and evaluate distributed predicates can use the proposed relations.
A cost-effective way to run a parallel application is to use existing workstations connected by a local area network such as Ethernet or FDDi. in this paper, we present an approach for paralleli/oof multidimensional...
详细信息
ISBN:
(纸本)9780897919661
A cost-effective way to run a parallel application is to use existing workstations connected by a local area network such as Ethernet or FDDi. in this paper, we present an approach for paralleli/oof multidimensional arrays on small networks of workstations with a shared-media interconnect, using the Panda i/o library. in such an environment, the message passing throughput per node is lower than the throughput obtainable from a fast disk and it is not easy for users to determine the configuration which will yield the best i/o performance. We introduce an i/o strategy that exploits local data to reduce the amount of data that must be shipped across the network, present experimental results, and analyze the results using an analytical performance model and predict the best choice of i/o parameters. our experiments show that the new strategy results in a factor of 1.2-2.1 speedup in response time compared tothe Panda version originally developed for the iBM SP2, depending on the array sizes, distributions and compute and i/o node meshes. Further, the performance model predicts the results within a 13% margin of error.
this paper extends an earlier out-of-core Fast Fourier Transform (FFT) method for a uniprocessor withthe parallel Disk Model (PDM) to use multiple processors. Four out-of-core multiprocessor methods are examined. ope...
详细信息
this paper extends an earlier out-of-core Fast Fourier Transform (FFT) method for a uniprocessor withthe parallel Disk Model (PDM) to use multiple processors. Four out-of-core multiprocessor methods are examined. operationally, these methods differ in the size of 'mini-butterfly' computed in memory and how the data are organized on the disks and in the distributed memory of the multiprocessor. the methods also perform differing amounts of i/o and communication. Twoof them have the remarkable property that even though they are computing the FFT on a multiprocessor, all interprocessor communication occurs outside the mini-butterfly computations;communication that ordinarily occurs in a butterfly is folded intoother data-movement operations. An analysis program shows that the two methods that use no butterfly communication usually use less communication overall than the other methods. the analysis program is fast enough that it can be invoked at run time to determine which of the four methods uses the least communication. one set of performance results on a small workstation cluster indicates that the methods without butterfly communication are approximately 9.5% faster. Moreover, they are much easier toimplement.
We present a suite of tools for monitoring and analysis of paging and i/o activity of parallel programs. then we discuss results of the measurements for several parallel programs.
We present a suite of tools for monitoring and analysis of paging and i/o activity of parallel programs. then we discuss results of the measurements for several parallel programs.
this paper presents compiler algorithms tooptimize out-of-core programs. these algorithms consider loop and data layout transformations in a unified framework. the performance of an out-of-core loop nest containing m...
详细信息
this paper presents compiler algorithms tooptimize out-of-core programs. these algorithms consider loop and data layout transformations in a unified framework. the performance of an out-of-core loop nest containing many references can be improved by a combination of restructuring the loops and file layouts. this approach considers array references one-by-one and attempts tooptimize each reference for parallelism and locality. When there are references for which parallelism optimizations do not work, communication is vectorized sothat data transfer can be performed before the innermost tiling loop. Preliminary results from hand-compiles on iBM SP-2 and intel Paragon show that this approach reduces the execution time, improves the bandwidth speedup and overall speedup. in addition, we extend the base algorithm to work with file layout constraints and show how it can be used for optimizing programs consisting of multiple loop nests.
Performance analysis of distributedsystems may be performed on different levels of abstraction. An accurate but time-consuming method is the construction of a simulation model which includes the different subsystems,...
详细信息
ISBN:
(纸本)0818680970
Performance analysis of distributedsystems may be performed on different levels of abstraction. An accurate but time-consuming method is the construction of a simulation model which includes the different subsystems, the communication system, and the load profile. in particular, this approach seems to be very powerful for real-time systems because of the inherent possibility of precise calculations of delays and processing times. A VHDL-based approach is presented which supports the performance analysis of mixed discrete-continuous, distributedsystems.
An overview of the evolution of distributed real-time controllers in aerospace manufacturing environments will be given. Programmable logic controllers (PLCs) and projected PLC market revenues will be reviewed, as wel...
详细信息
ISBN:
(纸本)0818680970
An overview of the evolution of distributed real-time controllers in aerospace manufacturing environments will be given. Programmable logic controllers (PLCs) and projected PLC market revenues will be reviewed, as well as manufacturing deployment strategies and customer support. Hidden costs of PLC technology will be identified. Examples of today's state-of-the-art paralleldistributed real-time manufacturing systems will be given. Large hidden support costs associated with such systems will be identified. in conclusion, ideas on how newer real-time distributedsystems can be supported by the current corporate philosophy of speed, quality and cost. Techniques of how to move forward toimplementation of new real-time distributedsystems will be presented.
暂无评论