### Table 1: Description of conditions where the SB algorithm outperforms the combining algorithm on a single switch.

1999

"... In PAGE 8: ... For instances of all-to-all broadcast where the startup cost dominates, the combining algorithm clearly outperforms the SB and ring algorithms. However, for cases where the startup cost and transmission latency are comparable (or neither is negligible), Table1 de nes which algorithm requires less time to complete in terms of the number of participating nodes and the ratio of transmission latency to send/receive overhead. 3.... ..."

Cited by 8

### Table 1: Description of conditions where the SB algorithm outperforms the combining algorithm on a single switch.

1999

"... In PAGE 8: ... For instances of all-to-all broadcast where the startup cost dominates, the combining algorithm clearly outperforms the SB and ring algorithms. However, for cases where the startup cost and transmission latency are comparable (or neither is negligible), Table1 de nes which algorithm requires less time to complete in terms of the number of participating nodes and the ratio of transmission latency to send/receive overhead. 3.... ..."

Cited by 8

### Table 5: Total costs for the improved combined algorithm as a function of the number of

2004

"... In PAGE 33: ... Note that the fuel-e cient algorithm is essentially a special case of the improved combined algorithm with m =1. Table5 provides total costs for the improved combined algorithm as a function of the number of agents for various w t =w f ratios. The number of participating agents was p =14 (i.... ..."

Cited by 8

### Table 5 provides total costs for the improved combined algorithm as a function of the

2004

"... In PAGE 34: ...51 10.48 Table5 : Total costs for the improved combined algorithm as a function of the number of moving agents (out of 14 participating agents), for various w t =w f values.... ..."

Cited by 8

### Table 3. Combined Throughput with Switch Conges- tion in an ATM WAN

1995

"... In PAGE 3: ... The experimental conditions included 128 kB TCP windows, 64 kB write buffers in ttcp, and ATM pacing at 35 Mb/s for each source. The results of Experiment 3 are shown in Table3 . These indicate that pacing is very effective in improving through- put.... ..."

Cited by 3

### Table 1: Comparative switch count.

"... In PAGE 5: ... In section 3 we present a constructive way of building an arbitrary size permutation network and we give the associated routing algorithm. This construction is the best known at this time in terms of switch count and improves previous results as shown by Table1 (section 2) and by computation (section 4). Finally, in section 5 we extend our results in case of one faulty switch.... In PAGE 6: .... W. Green, no publication being known though. Table1 presents the values obtained by the di erent constructions. 3 Network Construction Three binary switches can be used to construct a 3 3 permutation network as shown in Figure 6.... ..."

### Table 4. This means that the Fast DSOM is able to combine improvements

709

"... In PAGE 22: ...Table4 summarizes the results obtained by Algorithm 1 used with Scheme 2. As expected, the improvements appear with high values of M.... In PAGE 22: ... If this term is dominated by the pre-calculation phase (O(N2)) the improvement will remain unnoticed. This is why in Table4 the speed up is roughly increasing with M and decreasing with N. While in some extreme cases, that is when M is low compared to N (e.... ..."

### Table 3: The addressed combinations of switch and workload models

1995

Cited by 3

### Table 6: RAMpage (L1 associativity). No context switches on misses.

"... In PAGE 5: ... The improvement in going from a direct- mapped 16KB cache to an 8-way associative 256KB cache (combined capacity of L1i and L1d) is 19%. The comparable figure for RAMpage without context switches on misses ( Table6 ) is again under 12%, while that for the conventional hierarchy (Table 7) is almost 14%. 5 Conclusions In summary, improving L1 can reduce the importance of L2, but cannot in itself remove the memory wall problem.... ..."

### Table 8: Improved coverage estimates for combinational benchmarks. exact cov. (%) Stafan (short) Stafan (long) new algorithm (both sequences)

Cited by 1