### Table 10: Parameters used in P-LogP model.

2006

"... In PAGE 29: ... Both this models do not handle overhead for medium sized to long messages correctly and do not model hierarchical networks. The P-LogP model uses different sets of parameters for both networks, and consists of five parameters as shown in Table10 . This model uses parameters as a function of message size and uses measured values as input.... ..."

### TABLE 3: LOG PRICES (P) AND LOG WAGES (W), n = 360

2001

"... In PAGE 35: ...TABLE3 0: MODEL A dx = dy = .8 , de = .... In PAGE 35: ...212 .146 TABLE3 1: MODEL A, dx = dy = 1.2 , de = .... In PAGE 35: ...005 .002 TABLE3 2: MODEL A, dx = dy = 1.2 , de = .... In PAGE 36: ...TABLE3 3: MODEL B, dx = dy = .5 , de = .... In PAGE 36: ...010 .002 TABLE3 4: MODEL B, dx = dy = .8 , de = .... In PAGE 36: ...018 .002 TABLE3 5: MODEL B, dx = dy = .5 , de = .... In PAGE 37: ...TABLE3 6: MODEL B, dx = dy = .5 , de = .... In PAGE 37: ...024 .006 TABLE3 7: MODEL B, dx = dy = .8 , de = .... In PAGE 38: ...TABLE3 9: MODEL B, dx = dy = .8 , de = .... ..."

Cited by 1

### TABLE 3: LOG PRICES (P) AND LOG WAGES (W), n = 360

1999

"... In PAGE 35: ...TABLE3 0: MODEL A dx = dy = .8 , de = .... In PAGE 35: ...212 .146 TABLE3 1: MODEL A, dx = dy = 1.2 , de = .... In PAGE 35: ...005 .002 TABLE3 2: MODEL A, dx = dy = 1.2 , de = .... In PAGE 36: ...TABLE3 3: MODEL B, dx = dy = .5 , de = .... In PAGE 36: ...010 .002 TABLE3 4: MODEL B, dx = dy = .8 , de = .... In PAGE 36: ...018 .002 TABLE3 5: MODEL B, dx = dy = .5 , de = .... In PAGE 37: ...TABLE3 6: MODEL B, dx = dy = .5 , de = .... In PAGE 37: ...024 .006 TABLE3 7: MODEL B, dx = dy = .8 , de = .... In PAGE 38: ...TABLE3 9: MODEL B, dx = dy = .8 , de = .... ..."

### Table 4: LogP/LogGP model parameters (MPICH-2).

2005

Cited by 16

### Table 1. Performance Levels and Ranges for Speedup12 (logP log2P)

"... In PAGE 6: ... Kuck12 presents a comprehensive discussion on the issues that limit the practical speedup. He speci es acceptable performance levels and ranges for speedup based on the current state of technology, as illustrated in Table1... In PAGE 14: ...22 to 1.32, placing it in the intermediate performance range ( Table1 ). This performance is obtained despite distributing a very small fraction of the algorithmic code.... In PAGE 17: ...55 when compared to the sequential version. From Table1 , this represents a high performance range speedup indicating substantial computational gains. Thus, even discounting a distributed perspective for the algorithmic logic in developing on-line strategies, a distributed computational environment seems bene cial for on-line data processing under ITS applications.... ..."

### Table 1: Analysis of different Barrier algorithms The predictions for LogP and LogGP are identical since Barrier calls exchange zero-length messages.

2005

"... In PAGE 8: ... It is trivial to obtain LogP predictions from the LogGP formulas by setting the gap-per-byte parameter G to zero. Table1 summarizes performance equations for four different Barrier implementations. Table 2 summarizes equations used for predicting Reduce performance.... ..."

Cited by 16