S. Sardesai, Chime: A Versatile Distributed Parallel Processing Environment, PhD thesis, Arizona State University,  Home/Search   Document Not in Database   Summary   Related Articles   Check

....the tasks are more dynamic. Tasks can be suspended, or new ones can be created dynamically in a nested parallel environment. We did not implement bunching in our system. We proposed a way to tolerate failures of the centralized managers, which is not available in Calypso. 102 7.2. 4 Chime Chime[51, 50] has the same goal as our system, namely, to allow reliable distributed computation with nested parallelism and synchronization capabilities on a network of workstations. The di#erence between the two systems is the approach towards handling nested parallelism. Instead of dealing with jobs ....

S. Sardesai. Chime: A Versatile Distributed Parallel Processing Environment. PhD thesis, Arizona State University, July 1997.

....Fabian Monrose, Peter Wycko# and Roman Yangarber; but especially Arash Baratloo, who advised me to seek my summer employment, but also knew when I just needed to take it easy and told me so. Anina Karmen has had the patience of a saint in handling my academic a#airs. I have benefited from the wisdom of Richard Cole, Partha Dasgupta, Art Goldberg, Ben Goldberg, Bob Hummel, Krishna Palem, Jack Schwartz and many others here; but I am most grateful to Vijay Karamcheti, who I could always count on for extensive, reasoned and thoughtful feedback on my papers, which he was always prepared to ....

....over the course of this work. My father s high expectations have been a source of continuing inspiration, and I am thankful that he could witness my ultimate triumph; I know that my mother, z l, is watching and beaming as well. Most of all, I thank my wife, whose kindness, patience, support and wisdom granted me the strength and time to finish. v This research was sponsored by the Defense Advanced Research Projects Agency and Rome Laboratory, Air Force Materiel Command, USAF, under agreement numbers F30602 96 1 0320 and F30602 99 1 0517; by the National Science Foundation under grant number ....

[Article contains additional citation context not shown here]

S. Sardesai and P. Dasgupta. Chime: A Versatile Distributed Parallel Processing Environment. Available as http://www.eas.asu.edu/~calypso/ frames/research_papers/chime/.

....memory environment, even when the underlying parallel and distributed system resources may incur faults and exhibit wide variations in processing speeds. This support is exposed to the programmer in the form of a number of programming systems: Calypso [2] described in further detail below, Chime [24] which supports distributed execution of CC [6] programs, and Charlotte [4] which provides a web based metacomputing infrastructure. In addition, the MILAN system consists of supporting infrastructure such as the ResourceBroker [3] a system for dynamically managing the association and ....

S. Sardesai and P. Dasgupta. Chime : A versatile distributed parallel processing environment. ##### ######################.

....memory environment, even when the underlying parallel and distributed system resources may incur faults and exhibit wide variations in processing speeds. This support is exposed to the programmer in the form of a number of programming systems: Calypso [2] described in further detail below, Chime [24] which supports distributed execution of CC [6] programs, and Charlotte [4] which provides a Web based metacomputing infrastructure. In addition, the MILAN system consists of supporting infrastructure such as the ResourceBroker [3] a system for dynamically managing the association and 1422 ....

S. Sardesai and P. Dasgupta, Chime: A versatile distributed parallel processing environment, available at http:##www.cs.nyu.edu#milan.

....view of a fault free virtual shared memory environment, even when the underlying resources may incur faults and exhibit wide variations in processing speeds. This support is exposed to the programmer in the form of several programming systems: Calypso [1] described in further detail below, Chime [15] which supports distributed execution of CC [3] programs, and Charlotte [2] which provides a web based metacomputing infrastructure. In addition, the MILAN system consists of supporting infrastructure such as ResourceBroker (a system for dynamically managing the association and integration of ....

S. Sardesai and P. Dasgupta. Chime : A Versatile Distributed Parallel Processing Environment. http://cs.nyu.edu/milan/milan/index.html.

.... principal outcomes of the MILAN project are (1) a core set of fundamental resource management techniques [21, 3, 23, 9] enabling construction of efficient, reliable, predictable virtual machines, and (2) the realization of these techniques in three complete programming systems: Calypso [3] Chime [27], and Charlotte [6] Calypso extends C with parallel steps interleaved into a sequential program. Each parallel step specifies the independent execution of multiple concurrent tasks or a family of such tasks. Chime extends Calypso to provide nested parallel steps and inter thread communication ....

....any over bunching that may occur. 3.2 Chime Chime is a parallel processing system that retains the salient features of Calypso, but supports a far richer set of programming features. The internals of Chime are significantly different from Calypso, and it runs on the Windows NT operating system [27]. Chime is the first system that provides a true general shared memory multiprocessor environment on a network of machines. It achieves this by implementing the CC [8] language (shared memory part) on a distributed system. Thus in addition to Calypso features of fault tolerance and load ....

[Article contains additional citation context not shown here]

S. Sardesai. Chime: A Versatile Distributed Parallel Processing Environment. PhD thesis, Arizona State University, July 1997.

....view of a fault free virtual shared memory environment, even when the underlying resources may incur faults and exhibit wide variations in processing speeds. This support is exposed to the programmer in the form of several programming systems: Calypso [1] described in further detail below, Chime [16] which supports distributed execution of CC [3] programs, and Charlotte [2] which provides a web based metacomputing infrastructure. In addition, the MILAN system consists of supporting infrastructure such as ResourceBroker (a system for dynamically managing the association and integration of ....

S. Sardesai and P. Dasgupta. Chime : A versatile distributed parallel processing environment. http://cs.nyu.edu/milan/milan/index.html.

....computation loop, the latter of which can be parallelized. Even loop carried flow dependences can sometimes be broken in this manner. 9 Related Work This work unifies the fields of distributed systems and automatic parallelization. Previous distributed systems, such as Calypso[Bar99] and Chime[SD] require explicit programmer annotations for effective parallelization, making the porting of large existing codes impractical. On the other hand, Coelho[Coe94] has demonstrated that naive application of automatic parallelization techniques to the NOW setting leads to disappointing performance, ....

S. Sardesai and P. Dasgupta. Chime: A Versatile Distributed Parallel Processing Environment. Available as http://www.eas.asu.edu/ calypso/frames/research papers/chime/.

.... principal outcomes of the MILAN project are (1) a core set of fundamental resource management techniques [21, 3, 23, 9] enabling construction of efficient, reliable, predictable virtual machines, and (2) the realization of these techniques in three complete programming systems: Calypso [3] Chime [27], and Charlotte [6] Calypso extends C with parallel steps interleaved into a sequential program. Each parallel step specifies the independent execution of multiple concurrent tasks or a family of such tasks. Chime extends Calypso to provide nested parallel steps and inter thread communication ....

....compensates any over bunching that may occur. Chime is a parallel processing system that retains the salient features of Calypso, but supports a far richer set of programming features. The internals of Chime are significantly different from Calypso, and it runs on the Windows NT operating system [27]. Chime is the first system that provides a true general shared memory multiprocessor environment on a network of machines. It achieves this by implementing the CC [8] language (shared memory part) on a distributed system. Thus in addition to Calypso features of fault tolerance and load ....

[Article contains additional citation context not shown here]

S. Sardesai. Chime: A Versatile Distributed Parallel Processing Environment. PhD thesis, Arizona State University, July 1997.

....view of a fault free virtual shared memory environment, even when the underlying resources may incur faults and exhibit wide variations in processing speeds. This support is exposed to the programmer in the form of several programming systems: Calypso [1] described in further detail below, Chime [15] which supports distributed execution of CC [3] programs, and Charlotte [2] which provides a web based metacomputing infrastructure. In addition, the MILAN system consists of supporting infrastructure such as ResourceBroker (a system for dynamically managing the association and integration of ....

S. Sardesai and P. Dasgupta. Chime : A Versatile Distributed Parallel Processing Environment. http://cs.nyu.edu/milan/milan/index.html.

....computations. 1. Introduction This paper describes a series of experiments to test the implementation, features and performance of a parallel processing system called Chime. The experiments include runs of various scientific applications. Chime is a system developed at Arizona State University [1, 2] for running parallel processing applications on a Network Of Workstations (the NOW approach) Chime is a full implementation of the parallel part of Compositional C (or CC ) 3] running on Windows NT. CC is a language developed at Caltech and is essentially two languages in one. It has two ....

....not designed to run on NOWs. Chime solves this problem, by implementing parallel CC on the NOW architecture. Chime uses a set of innovative techniques called two phase idempotent execution strategy [4] distributed cactus stacks [1] eager scheduling [4] dependency preserving execution [2] and the well known technique called distributed shared memory [5] to implement parallel C . In addition, it has the extra features of load balancing, fault tolerance and high performance. Chime is the first (and of this writing, the only) parallel processing system that provides the above ....

[Article contains additional citation context not shown here]

Shantanu Sardesai, CHIME: A Versatile Distributed Parallel Processing System, Doctoral Dissertation, Arizona State University, Tempe, May 1997.

....Figure 2 shows the result of running a similar (but larger) RayTrace program under Pentium Pro 200MHz systems We performed many other tests, including tests for load balancing (mixture of slow and fast machines) and fault tolerance (transient machines. Many of these results can be found in [MSD97, SD97] and on the Web site. A particular test of Chime is interesting. We ran a trivial program under Chime that takes a 1024 element array and initializes each element in parallel. To make the test rigorous, we did thread creation recursively, that is at the top level two tasks are created, which in ....

S. Sardesai and P. Dasgupta, Chime: A Versatile Distributed Parallel Processing Environmen, Technical Report, TR-97-002, Department of Computer Science and Engineering, Arizona State University, 1997. Also available via www.eas.asu.edu/~calypso

....memory multiprocessor environment on a network of machines. It achieves this by implementing the CC [CK92] language (shared memory part) on a distributed system. In addition to shared memory, parallelism and synchronization features of CC , Chime also provides faulttolerance and load balancing [Sar97, Das97]. 2. Related Work Shared memory parallel processing in distributed systems is limited to a handful of DSM systems 2 that provide quite similar functions (Midway [BZS93] Quarks [Kha96] TreadMarks [ACD 96] Munin [Car95] DSM systems do not provide a uniform view of memory i.e. some global ....

....utilization of unreliable shared resources available over the network. The resource utilization and handling of unreliable computing resources is provided by dynamic load balancing and fault tolerance features. Measurements show that there is almost no additional overhead for these features [MSD97, Sar97]. Chime has been implemented in C , using Microsoft Visual C 4.2 and was developed on Pentium based machines running Windows NT 4.0. 4. Distributed Cactus Stacks When different nested parallel tasks start their execution from within the context of their parent task on different workstations, ....

[Article contains additional citation context not shown here]

Shantanu Sardesai, CHIME: A Versatile Distributed Parallel Processing System, Doctoral Dissertation, Arizona State University, Tempe, May 1997.

No context found.

S. Sardesai, Chime: A Versatile Distributed Parallel Processing Environment, PhD thesis, Arizona State University,

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