T. Kamada, S. Matsuoka, and A. Yonezawa. Efficient Parallel Global Garbage Collection on Massively Parallel Computers. In Supercomputing Conference, pages 79--88, 1994.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Instead of using the terms touched, suspended and untouched 17 for different possible states of actors, we use the terms black, gray and white respectively to remain consistent with Kafura s terminology. A garbage collector for active objects in an MPP environment has been described by Kamada [18]. In this work, all active objects are considered live. An interesting feature of their algorithm is the use of a centralized agent which handles many problems related to synchronization and detection of termination of various phases of the garbage collection. Puaut [29, 30] presents an algorithm ....

Tomio Kamada, Satoshi Matsuoka, and Akinori Yonezawa. Efficient parallel global garbage collection on massively parallel computers. In Eliot Moss, Paul R. Wilson, and Benjamin Zorn, editors, OOPSLA/ECOOP '93 Workshop on Garbage Collection in Object-Oriented Systems, October 1993.

....systems such as disks. Execution can be resumed by restoring those contexts. No additional mechanism is required to have a consistent state in the distributed processes. Global garbage collection: Global garbage collection is difficult because of the references included in transmitted messages[KMY94] As in the case of consistent checkpointing, network preemption gives a clear consistent state timing and a chance to investigate the messages in transmission. Here, channel contexts are added to the root set of a global garbage collection. Thus network preemption makes the marking of live ....

Tomio Kamada, Satoshi Matsuoka, and Akinori Yonezawa. Efficient Parallel Global Garbage Collection on Massively Parallel Computers. In Supercomputing Conference, pages 79--88, 1994.

....for determining whether some set of objects form a cycle of garbage. A heuristic method is used to select candidate garbage upon which the back tracing mechanism is invoked. This mechanism is expensive and relies on the heuristic selection method to carefully select candidates for investigation. Kamada, Matsuoka, and Yonezawa [1994] describe a weighted reference counting method which uses a central node to collect reference information in order to detect inter node cycles. 1.3 Paper Outline The problems identified and addressed by this paper lie deep within the algorithms that constitute the DMOS collector. Section 2 ....

KAMADA, T., MATSUOKA, S., AND YONEZAWA, A. 1994. Efficient parallel global garbage collection on massively parallel computers. In Supercomputing '94 (1994). IEEE Computer Society.

....or tight coordination between processors. Most of them are complex in order to overcome problems in distributed environments such as faulty processes or lost messages, and have not been implemented. There has been little work in the context of parallel highperformance computing. One study [41] focuses on superficial aspects such as the number of messages or the total overhead of global collections. How they affect overall application performance has not yet been studied. Compared to the many proposed algorithms, our global collector is rather simple. Our global collection assumes ....

Tomio Kamada, Satoshi Matsuoka, and Akinori Yonezawa. Efficient parallel global garbage collection on massively parallel computers. In Proceedings of Supercomputing '94, pages 79--88, 1994.

....have to periodically inform servers that they are still interested in an object reference. Once all remote references have expired, normal local collection can be used to reclaim objects. Yonezawa s group have undertaken one of the few investigations into garbage collection for a parallel system [6, 9]. Our work is similar to theirs in using a parallel marking scheme and buffering of mark messages. Their system supports the ABCL parallel object oriented language. Their most recent system supports: asynchronous local, synchronous local and parallel global garbage collection. All collections ....

....buffers, avoiding any further copying. However, under buffer pressure it may be necessary to copy pointers out of buffers into a more compact bitmap to avoid network deadlock. Our current implementation uses a termination detection algorithm similar to the one used originally by Yonezawa s group [6]. An elected leader keeps records of pending mark messages. Once all processors have informed the leader that they have completed their local cycle and, as indicated by a bit vector in the message to the leader, that they have not sent further mark messages, the leader detects termination and ....

T Kamada, S Matsuoka, and A Yonezawa. Efficient parallel global garbage collection on massively paralle computers. In Proceedings of Supercomputing'94, pages 79--88, Washington DC, November 1994. IEEE.

....worse as heap became larger in Bin tree and Puzzle. 5 Related Work A number of algorithms for reference counting and global mark and sweep have been proposed in the literature[9, 13] Performance studies are rare, presumably because most of them are not implemented. Existing performance studies [10, 14, 16, 19] only show performance of either a global mark and sweep or a reference counting and do not compare them. Most of previous works about distributed garbage collection schemes have been saying that: Reference counting can do local collections independently and can incrementally reclaim global ....

....to make it latency tolerant, because communication latency is naturally masked by the application. Notice that while the global mark and sweep presented in this paper may be sensitive to latency, it is certainly possible to make a global mark and sweep latency tolerant, in ways described in [7] or [10], for example. Hence, the real benefit is that the latency tolerance comes free for reference counting. To summarize, it will be a choice in environments where communication latency is very long, or we know that most distributed garbage is shallow. They will be the case, for example, in ....

Tomio Kamada, Satoshi Matsuoka, and Akinori Yonezawa. Efficient parallel global garbage collection on massively parallel computers. In Proceedings of SuperComputing, pages 79--88, 1994.

....at the RWCP. Our approach is based on the concurrent object oriented paradigm. Last year we designed a concurrent object oriented programming (COOP) language called ABCL ST and this year we developed its language systems environments including the optimizing compilers[6, 7, 4, 5] runtime systems[2, 3] and debuggers[1] 2 Language Systems Improvements of the ABCL ST compiler for EM 4 The main target machine of the ABCL ST compiler is a fine grained data driven parallel computer, EM 4, developed at Electrotechnical Laboratories, which is regarded as an archi prototype of the RWC 1 machine. We ....

T. Kamada, S. Matsuoka, and A. Yonezawa. Efficient parallel global garbage collection on massively parallel computers. In Proc. of Joint Symposium on Parallel Processing (JSPP), pages 33--40, May 1994. (in Japanese).

....at the RWCP. Our approach is based on the concurrent object oriented paradigm. Last year we designed a concurrent object oriented programming (COOP) language called ABCL ST and this year we developed its language systems environments including the optimizing compilers[6, 7, 4, 5] runtime systems[2, 3] and debuggers[1] 2 Language Systems Improvements of the ABCL ST compiler for EM 4 The main target machine of the ABCL ST compiler is a fine grained data driven parallel computer, EM 4, developed at Electrotechnical Laboratories, which is regarded as an archi prototype of the RWC 1 machine. We ....

T. Kamada, S. Matsuoka, and A. Yonezawa. Efficient parallel global garbage collection on massively parallel computers. In Proc. of Supercomputing, pages 79--88. IEEE, 1994.

....or tight coordination between processors. Most of them are complex in order to overcome problems in distributed environments such as faulty processes or lost messages, and have not been implemented. There has been little work in the context of parallel high performance computing. One study [19] focuses on superficial aspects such as the number of messages or the total overhead of global collections. How they affect overall application performance has not yet been studied. Compared to the many proposed algorithms, our global collector is rather simple. Our global collection assumes ....

Tomio Kamada, Satoshi Matsuoka, and Akinori Yonezawa. Efficient parallel global garbage collection on massively parallel computers. In Proceedings of Supercomputing '94, pages 79--88, 1994.

No context found.

T. Kamada, S. Matsuoka, and A. Yonezawa. Efficient Parallel Global Garbage Collection on Massively Parallel Computers. In Supercomputing Conference, pages 79--88, 1994.

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