| Mattern, F., Mehl, H., Schoone, A., Tel, G.: Global Virtual Time Approximation with Distributed Termination Detection Algorithms. Technical Report RUU-CS-91-32, University of Utrecht, The Netherlands (1991) |
....other are grouped to the same processor to reduce the communication. For objects on a remote workstations, the event messages have to be communicated using the message passing software. Messages sent in a Time Warp simulation (on a network of workstations) include the global virtual time (GVT) [52] estimation messages, anti messages after a rollback and several other kinds of messages. A message communicated between applications on different processors must cross the process address space, the processor boundary, the workstation boundary through the interconnection network to the process ....
Mattern, F., Mehl, H., Schoone, A. A., and Tel, G. Global virtual time approximation with distributed termination detection algorithms. Tech. Rep. RUU-- CS--91--32, Dept. of Computer Science, University of Utrecht, The Netherlands, 1991.
....means follows from Tel s result that every deadlock detection protocol can be converted into a GV T protocol. Perhaps (in Tel s terminology) our results can be described as a search for the distributed infimum algorithms underlying certain deadlock detection schemes. Related work has been done in [4] where the authors propose running several termination detection protocols in parallel to calculate GV T . However, many of the elegant results in [4] are for synchronous protocols (where message transmission is instantaneous) and the extensions to asynchronous protocols are not as elegant or ....
....can be described as a search for the distributed infimum algorithms underlying certain deadlock detection schemes. Related work has been done in [4] where the authors propose running several termination detection protocols in parallel to calculate GV T . However, many of the elegant results in [4] are for synchronous protocols (where message transmission is instantaneous) and the extensions to asynchronous protocols are not as elegant or efficient. We also use a different set of termination detection protocols (that arise in CM simulations) as our point of departure, and so our results are ....
[Article contains additional citation context not shown here]
F. Mattern, H. Mehl, A. Schoone, and G. Tel, Global Virtual Time Approximation with Distributed Termination Detection Algorithms, Tech. Rep. RUU-CS-91-32, Dept. of CS, Univ. of Utrecht, Netherlands, 1991.
....3699 RAID 50000 4 756 1052 3650 PHOLD 10 22 542 111 243 PHOLD 20 18 571 99 250 PHOLD 100 9.8 587 134 333 SMMP 10000 2.2 276 131 628 SMMP 20000 2.7 309 137 588 SMMP 30000 1.7 334 139 680 Table 2. Performance of OFC TW (risk = 0. 1) Performance results for GVT were gathered using Mattern s algorithm [14] for two different GVT periods (100 and 1000) Mattern s algorithm was chosen for comparison because it is the fastest GVT estimation algorithm available in warped. In addition, in warped [16] GVT estimates were required only for fossil collection. The number of cycles between GVT estimations is ....
F. Mattern, H. Mehl, A. A. Schoone, and G. Tel. Global virtual time approximation with distributed termination detection algorithms. Tech. Rep. RUU-- CS--91--32, Dept. of Computer Science, University of Utrecht, The Netherlands, 1991.
....adapt the four counter solution presented in the preceding section to the synchronous model. Here, however, we aim at a different solution which is not based on counting. 6 This section is based on the paper Global Virtual Time Approximation with Distributed Termination Detection Algorithms [21], co authored by H. Mehl, A. Schoone, and G. Tel. Formally, we model the behavior of the basic computation by means of three atomic actions such that rules R1, R2, and R3 are obeyed. We assume that each process P i has a system variable state i with values from factive, passiveg. Rule R3 ....
....and van Gasteren [6] However, whereas in that algorithm a flag is set when a message is sent, our scheme uses a flag (the sticky state indicator) which is set when a message is received. More details and a formal proof of a generalized variant of the sticky flag algorithm may be found in [21]. 4 The Echo Algorithm The echo algorithm is a distributed algorithm which can be used to traverse an arbitrarily connected graph in a parallel way and to distribute data of the initiator to all processes. It was first published by Chang in 1982 [4] Segal found a slightly more efficient version ....
[Article contains additional citation context not shown here]
F. Mattern, H. Mehl, A.A. Schoone, and G. Tel. Global Virtual Time Approximation with Distributed Termination Detection Algorithms. Technical Report RUU--CS--91-32, University of Utrecht, 1991.
....of the termination detection problem are the same as those of other important problems in distributed computing. Indeed, termination detection algorithms are related to algorithms for computing distributed snapshots [3, 19] detecting deadlocks [5, 23] and approximating a distributed infimum [31, 20]. Thus the problem is seen to be important both from a practical, algorithmical, and from a theoretical, methodological point of view. From both points of view we consider it useful to recognize general design paradigms for distributed termination detection algorithms. One such paradigm was ....
....as the one described here have been given between members of this family. It was argued by Natarajan [23] that termination is a special case of communication deadlock. It was shown by Tel [32] that termination detection is a special case of global virtual time approximation, and Mattern et al. [20] presented a derivation from termination detection algorithms to global virtual time approximation algorithms. Even though in some cases a solution to a simple problem is derived from a solution to a more complicated problem, we think transformations of this kind are useful for several ....
[Article contains additional citation context not shown here]
Mattern, F., H. Mehl, A.A. Schoone, and G. Tel, Global Virtual Time Approximation with Distributed Termination Detection Algorithms, Tech. Rep. RUU--CS--91--32, Dept. of Computer Science, University of Utrecht, 1991.
....distributed termination detection is a particular case of GVT approximation, this scheme does also yield an elegant parallel termination detection algorithm for distributed computations. Interestingly, it is also possible to derive GVT approximation algorithms from termination detection algorithms [21] which shows that the two problems are closely related instances of the general snapshot problem. 19 ....
Mattern, F., Mehl, H., Schoone, A., Tel, G. Global Virtual Time Approximation with Distributed Termination Detection Algorithms. Tech. Rep. RUU-CS-91-32, Department of Computer Science, University of Utrecht, The Netherlands, 1991.
No context found.
Mattern, F., Mehl, H., Schoone, A., Tel, G.: Global Virtual Time Approximation with Distributed Termination Detection Algorithms. Technical Report RUU-CS-91-32, University of Utrecht, The Netherlands (1991)
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