Results 1 - 10
of
284,949
Table 4: Comparisons for two-level service hierarchy by adjusting the hit ratios between the two levels.
1996
"... In PAGE 16: ... meet the scalability criteria. By varying the hit ratios in Equation 1 for DHM to determine the optimal bandwidth combinations of the entire system, Table4 shows the scalability by varying the hit ratios. It is assumed that there is no write-through data streams from higher to lower level servers.... In PAGE 19: ... Suppose that the hit ratio to the second level storage is approximately 90%. From Table4 , the rst and second level servers should equip with ve CD jukeboxes and at least four 20GB RAIDs respectively, to satisfy the storage requirement as shown in Figure 9. In addition, there are ve PC-based machines to attach storage, and two types of Fast Ethernet switch hubs for network interconnection.... In PAGE 21: ... 6 7 3 2 S23 S24 S21 S22 SW1 CD ROM Jukebox 20 GB RAID 100 Mbps Connection Twelve 100 Mbps Ports One 100 Mbps Port and Thirty-Two 10 Mbps Ports Fast Ethernet Switch Hub with Thirty-Two 10 Mbps Connections 32 32 32 32 32 32 32 8 24 Not Used 5 8 4 1 Fast Ethernet Switch Hub with S1 SW2 Figure 9: The KOD system diagram for a Karaoke house In this scenario, each RAID only supports 50 accesses, although its bandwidth can sustain 106 concurrent streams through SCSI II, which adds up to a total of 444 potential concurrent accesses (Nmax A ) for the channels. By Table4 , if we add 5 more 20 GB RAIDs in the system (N2 = 9), the storage system can maximize its usage and support up to 974 customers with a much lower cost factor for storage. In the mean while, the hit ratio of the second level server is increased to 98% from 90%.... ..."
Cited by 3
Table 2: Number of clusters and their average size (in parentheses) generated in a two-level architecture.
"... In PAGE 6: ...75. Table2 shows the number of clusters and their average size generated in the two-level architecture using di erent clustering methods.... ..."
Table 2: Number of clusters and their average size (in parentheses) generated in a two-level architecture.
"... In PAGE 6: ...75. Table2 shows the number of clusters and their average size generated in the two-level architecture using di erent clustering methods.... ..."
Table 3.7: Value of B for minimum delay in a two-level bus hierarchy
1989
Cited by 2
Table 1: Two-level aggregation examples.
2006
"... In PAGE 3: ... XPORT customizes its function- ality and optimization framework to respect these constraints. Table1 shows the different aggregation function combinations along with some example metrics. 2.... ..."
Cited by 5
Table 2: CPU cycles per operation (Unit: CPU cycles) Multi-Hash-Function Two-Level LRU List [2]
"... In PAGE 5: ... In the experiments, the multi-hash-function framework adopted a 2KB hash table with 4096 entries. Table2 shows the run-time overheads for each operation of the experi- mented methods, where \Checkup quot; means the veriflcation of whether an LBA is for hot data, \Status-Update quot; means the updating of the status of an LBA, and \Decay quot; means the decaying of all counters. The \Status-Update quot; opera- tion of the two-level LRU list method was the insertion of the LBA into the two lists.... ..."
Table 3: Ratios of speedups of a two-level HPAM
"... In PAGE 7: ...1 and actual hardware costs indicates the two-level organization is best in almost all the cases. Table3 shows the ratios of speedups of a two- level HPAM to a one-level HPAM. The second level of the two-level HPAM and the one-level HPAM have the same number and type of processors.... In PAGE 7: ...01 for values of MPD = 10, 100 and 1000 respec- tively. The results of Table3 show that these condi- tions are met for almost all the benchmarks and values of MDP. In some cases the gain factor is as high as 8.... ..."
Table 2: Two-level implementation of MCNC benchmark FSMs
"... In PAGE 3: ... It may be better not to reduce the FSM before decomposing it. Two-level implementation ( Table2 ): For 5 of 7 ex- amples the decomposition obtains a smaller area than the lumped machine (29% - 63%). For all ex- amples the component machines are smaller than the lumped machine.... ..."
Table 2 Experimental results for two-level caching.
"... In PAGE 11: ... 5.2 Experiments for the two-level cache policies In Table2 , the first three columns show the model names, the corresponding parameters used and the total memory consumed for the evaluation respec-... In PAGE 12: ... To perform a fair comparison among these cache policies, we use fixed-size unbounded hashing with init = 100 for all experiments. In Table2 , we can see that our two-level cache policies do significantly improve the runtime in some cases, especially the models with a constant or linear number of MDD levels. When there is an improvement, MoveFirst al- ways outperforms MoveForward and DoNothing.... In PAGE 12: ... In the worst case, neither MoveFirst nor MoveForward slow down the runtime with respect to DoNoth- ing. In a nutshell, by comparing column 5 in Table 1 with column 4 in Table 2 (for runtime) and column 8 in Table 1 with column 3 in Table2 (for memory consumption), we observe that the overhead of the two-level cache is not negligible in comparison to the one-level cache. However, for those benchmarks where the cache policies can improve the runtime, this is a good tradeoff.... ..."
Table 1: Testability and area comparisons of two-level implementations
1992
"... In PAGE 27: ...heorems 7.1 and 7.2 may easily be generalized. 8 Experimental Results We present experimental results using an ENF-based program for analyzing the delay-fault testability of multilevel circuits, and preliminary experimental results obtained via algebraic factorization to retain robust path-delay-fault testability. We chose several examples from the MCNC Benchmark set, whose statistics and path-delay- fault testabilities under a two-level implementation are summarized in Table1 . The testabilities for single-output minimized and multiple-output minimized two-level representations were determined using an ENF-based analysis program.... ..."
Cited by 11
Results 1 - 10
of
284,949