Tools
Sorted by:
Try your query at:
 |
 |
 |
 |
 |
 |
Results 1 - 10
of
50,026
Table 1: Scalability Metric The results achieved are encouraging. Mutuality Preservation is shown to be adaptive, give good performance results while incurring a small overhead on the system in terms of computation time, communication and storage. It is also general, stable (exhibiting a high hit-ratio), scalable ( high scalability measure) and can tolerate message loss. Dependence on system size and latencies is minimized. A node retains nodes of interest in its cache and replaces others by random selection from the rest of the system (biased-random selection). Cache membership thus changes dynamically as nodes are independently and mutually included or excluded from their respective caches. This allows Mutuality Preservation to attain very good results with minimal extra overhead. With Mutuality Preservation large scale and complex networked systems can be addressed and ubiquitous resource access supported. In the next section we demonstrate the suitability of the algorithm to a networked or distributed environment, applying it to provide Ubiquitous Resource Access in a complex and large networked system like the Internet.
"... In PAGE 22: ... It can be seen that system size has a minimal affect on the performance of the algorithm: there are minor differences between the results for different system sizes. This is also shown by the values of the scalability metric, presented in Table1 . Scaling system size from 20 to 50 nodes, the results ... ..."
Table 1. Mean values of the service time (in seconds) and speedup for the high scalable workload, LU
Cited by 2
Table 1. Mean values of the service time (in seconds) and speedup for the high scalable workload, LU
Cited by 2
Table Growth by Filtering Based on Address Allocation Policies. http://www.research.att.com/ jrex/papers/filter.pdf [26] E. Kohler, J. Li, V. Paxson, and S. Shenker. Observed structure of ad- dresses in IP traffic. 2nd Internet Measurement Workshop, November 2002. [27] T. Bates and Y. Rekhter. Scalable Support for Multi-homed Multi-provider Connectivity. RFC 2260. [28] V. Srinivasan, G. Varghese. Fast address lookups using controlled prefix expansion. ACM Transactions on Computer Systems. [29] M. Waldvogel, G. Varghese, J. Turner, B. Plattner. Scalable High-speed prefix matching. ACM Transactions of Computer Systems. [30] M. Degermark, A. Brodnik, S. Pink. Small Forwarding Table for Fast Routing Lookups. SIGCOMM 1997.
Table 4 Speedups for the FFT application 1M data set 4M data set
"... In PAGE 8: ...and MPI versions are presented in Table4 for 1M and 4M data sets. Neither MPI nor SAS show high scalability for our test cases.... ..."
Table 3: Comparison of Existing Modes of Operation with proposed Compaction/Randomization Mode. independent, (vi) there is no cell-to-cell dependency (no feedback from previous cells), and (vii) it is highly scalable (i.e., cells from the same stream can be ciphered and deciphered in paral- lel). Table 3 summarizes a comparison of the proposed randomization/compaction mode with some well-known existing modes, namely, ECB, CBC, CFB, OFB, and Counter. The presented scheme is seen to be superior to known existing modes of operation such as ECB, CBC, CFB, OFB, and Counter, each of which has only some of the above-listed attractive features. Based on the proposed technique of compression/randomization for encryption, the paper also presented a secure mechanism for in-band synchronization of encryption/decryption key updates. The mechanism used a marker cell within the data channel, whose original data payload that contains the old and new keys is pseudo-randomized and subsequently block ciphered before transmission. An important aspect of the solution is the ability to distinguish the marker cell from the other encrypted cells by encoding the information in the data eld as to whether the 16
Table 1. The ASSIST operational architectural requirements. Availability Requirements ProvideaWebfrontendtotheASSISTsystem Need to interact with supplier and other systems across firewalls Need for the ASSIST system/user interfaces to interact with multiple servers Need to interact with legacy systems Business logic independence from plumbing details and user interface Scalability for large amounts of users, data, and projects High availability and easy fail recovery Integration Requirements Eliminate or reduce vendor-lock in for COTS tools Need to interact with common COTS tools (Excel, IDEAS, Microsoft Project, etc.)
1992
"... In PAGE 3: ... Thus the primary requirement for ASSIST is to provide an integrated view and use of the different ASSIST information and tools (whether commercial, legacy, or internal). Table1 lists the operational architectural requirements that have been distilled for application to the ASSIST architecture. These include availability requirements, integration requirements, and security requirements.... ..."
Cited by 1
Table 6: Runtimes (in seconds) for the original and SAW orderings on the Tera MTA.
2000
"... In PAGE 7: ...Table 6: Runtimes (in seconds) for the original and SAW orderings on the Tera MTA. Results using 60 streams per processor are pre- sented in Table6 . Both CG and the underlying SPMV achieve high scalability of over 90% using the ORIG ordering.... ..."
Cited by 3
Table 2: Execution time (sec) of sparse displacement field computation with static work distribution on WM-SciClone cluster.
"... In PAGE 7: ... The computation is highly scalable. As evident from Table2 , we see a significant reduction in the runtime as we increase the number of processors involved. However, the same Table shows that the reduction is not proportional to the number of processors we use, because the scalability of our application is hindered by load imbalances.... ..."
Table 1 Performanceresults in seconds for the routines DGGHRD and DHGEQZ on a single P2SC processor.
"... In PAGE 2: ... The measured results are obtained by using up to 64 Thin Nodes #28P2SC, 120 MHz#29 on the IBM Scalable POWERParallel System at High Performance Computing Center North #28HPC2N#29. Table1 shows uni-processor results for the blocked two-stage implementation of DGGHRD, when computing the #28H; T#29 form, #28H; T#29 form and Q, and #28H; T#29 form and Q and Z. The table also shows the performance results for DHGEQZ when computing the eigenvalues, Shur form, and Shur form and Shur vectors.... ..."
Results 1 - 10
of
50,026
