### Table 2. Two and Three Dimensional Results

in SUMMARY

"... In PAGE 8: ... Under these conditions the difference in the effectiveness of the learning algorithm with targetsfandf can be attributed directly to the additional input dimension. Table2 provides the results of approximatingAx,y) = (x+y)/2 and f(xy,z) = (x*)/2. For the 5/25 system cotiguration, the three-dimensional system generated with 50,000 training examples was less accurate than the two- dimensional approximation produced with 5000 examples.... In PAGE 9: ... As in the case of the propagation model, the localized FAMs may not require all of the dimensions of the input space to produce an appropriate response. In fact, the target function and experimental data shown in Table2 is an example of this type of behavior. The test for the contribution of the ith input dimension begins by constructing the n-dimensional FAM for the region.... ..."

### Table 9: Performance data for two and three-dimensional, unordered, CCFFT on a 2048 processor CM-200.

1992

"... In PAGE 18: ... The latter uses only radix-8 kernels, which are the most e cient. Timings for two- and three-dimensional CCFFT are given in Table9 , and shown in Figure 7. The signi cant increase in performance for the two-dimensional CCFFT between the 1024 1024 array and the 2048 2048 array is due to one of the axis being local to a processor for the larger array.... ..."

Cited by 1

### Table 10: Performance data for two and three-dimensional, ordered, CCFFT on a 2048 processor CM-200.

1992

"... In PAGE 19: ... This part of the axis requires a radix-2 kernel, which is less e cient than the radix-4, and the radix-8 kernels normally used. For reference, performance data for ordered two and three-dimensional transforms are given in Table10 . The execution time increases by 50 - 100% for our examples, considerably more than for entirely local transforms.... ..."

Cited by 1

### Table 9. Execution time (seconds) for Haskell versions with three-dimensional intermediate forms

1997

"... In PAGE 18: ... Peak heap use (in K bytes) for strict Haskell versions (Chalmers) 5.4 Time analysis Table9 shows the time performance of the codes that use three-dimensional interme- diate structures. Both Glasgow and Chalmers favor the selection/recursion code, and both show the copy versions performing worse than the selection versions.... ..."

Cited by 6

### Table 6.2 Performance of Algencan, Ipopt and Lancelot in the three-dimensional Bratu-based optimization problem.

2005

Cited by 12

### Table 8: Performance data for two and three-dimensional, ordered, CCFFT on a 2048 pro- cessor CM-200.

1992

### Table VI. Three-dimensional 6-block contact problem: performance results of the FETI-C solver

in A Numerically Scalable Domain Decomposition Method For The Solution Of Frictionless Contact Problems

### TABLE I1 PERFORMANCE OF THREE-DIMENSIONAL CONNECTED COMPONENT LABELING (CCL)

### Table 2. I/O Characteristics of the ROMIO three-dimensional block test

2003

"... In PAGE 7: ...locks. The memory datatype is contiguous. 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 Aggregate read bandwidth (Mbytes/sec) Compute processes (2 per node) DatatypeList Two-Phase Data SievingPOSIX 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 Aggregate write bandwidth (Mbytes/sec) Compute processes (2 per node) Datatype List Two-PhasePOSIX Figure 10. 3D block read and write perfor- mance Table2 characterizes the resulting I/O patterns using our tested optimizations, and Figure 10 shows the results of our tests. Again, datatype I/O is the clear performance winner; peak performance is more than double that of the next-best approach.... ..."

Cited by 10

### Table 2. I/O Characteristics of the ROMIO three-dimensional block test

2003

"... In PAGE 7: ...locks. The memory datatype is contiguous. 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 Aggregate read bandwidth (Mbytes/sec) Compute processes (2 per node) DatatypeList Two-Phase Data Sieving POSIX 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 Aggregate write bandwidth (Mbytes/sec) Compute processes (2 per node) DatatypeList Two-Phase POSIX Figure 10. 3D block read and write perfor- mance Table2 characterizes the resulting I/O patterns using our tested optimizations, and Figure 10 shows the results of our tests. Again, datatype I/O is the clear performance winner; peak performance is more than double that of the next-best approach.... ..."

Cited by 10