### Table 36 Implementation-Dependent Constants

"... In PAGE 63: ...329 LuminanceWhite PEXFLOAT 1.0 Table36 Implementation-Dependent Constants (Continued) Name Type... In PAGE 64: ...October 1993 Implementation-Dependent Constants Table36 lists the values of the Implementation-Dependent Constants in Solaris PEX 2.1.... ..."

### Table 71 Differences between Server and Direct PEXlib reported values for certain implementation-dependent constants

### Table 6: Memory required per processor by the packed solver and ScaLAPACK (Mbytes).

"... In PAGE 10: ... Remark 2. Since the memory required by the packed implementation depends on the chosen value for b, it is interesting to compare in detail in Table6... ..."

### Table 2: PROFIBUS parameters for a network length of 500 m (station delay is implementation dependent) The function of relevant PROFIBUS frames, as well as their duration, is described in Table3. Frame Symbol Duration ( s)

1997

Cited by 1

### Table 2.1: GMTK can implement dependencies between two parent and child random variables depending on the type of the variables. This table shows which forms are currently implemented in GMTK. The squares marked with an X are currently supported.

in Contents

2002

### Table 2.1: GMTK can implement dependencies between two parent and child random variables depending on the type of the variables. This table shows which forms are currently implemented in GMTK. The squares marked with an X are currently supported.

### Table 2: Complexity of formation of matrix objects or matrix-vector actions using forward or hybrid modes of modern automatic di erentiation software. The asterisk signi es that the reverse mode con- sumes memory, in a carefully drawn time-space trade-o , so r is implementation dependent. Object

in Using Automatic Differentiation for Second-Order Matrix-free Methods in PDE-constrained Optimization

2000

Cited by 2

### Table I gives details and further references. The e ectiveness of the tools generated by the ASF+SDF Meta-Environment is critically dependent on the quality of the rewriting implementation. The original in- terpretive implementation left room for improvement. Its author, inspired by earlier rewrite compilation work of Kaplan [1987], sketched a more e cient compilational

2002

Cited by 22

### Table 1: Total number of bytes transmitted during global combining. All runs are with 64 processors.

1995

"... In PAGE 21: ... Timing results are instructive, but depend on implementation details and machine ar- chitecture. To evaluate the optimizations with a more objective measure, Table1 shows the total number of bytes transmitted in the course of the Quad algorithm. Since the amount of work to merge clusters is directly proportional to the length of messages, these numbers give a good indication of how successfully the optimizations exploit various cluster con gu- rations.... ..."

Cited by 4

### Table 5: Token-Ring Inaccessibility Times (ladd = 48) Analytic Results In order to complete our study of networking accessibility we now evaluate the inaccessibility time bounds, for a given token-ring network. The network length is Cl = 500m and the total number of stations NST = 32. We have assigned a typical value of 200 s to all the processing times, with an overhead of 25% for their worst-case gures. These values are estimates, supported by our personal experience on the evaluation of communication protocols performance. However, an experimental measurement of these parameters is a mandatory condition when a highest degree of accuracy is required. To all the remaining implementation dependent parameters we have assigned the set of values which characterise the behaviour of [23] (latAM = 27, latst = 2:5, NClm = 2, NBcn = 7, tCIFS = tBIFS = 20ms, tdect signloss = 200ms, tPhyOut = 8:7ms and tWireFault = 5s ). Further 16Upstream and downstream.

1992

Cited by 6