CiteSeerX — Table search results

Results 1 - 10 of 28,280

Table 15: Averages of processor assignment compared with mapping for average dilation ratios, cut-weight and partitioning times.

in Multilevel Mesh Partitioning for Heterogeneous Communication Networks

by C. Walshaw, M. Cross 2001

"... In PAGE 20: ... Even so the average dilation figures are still around 2 times worse on average for partitioning/assignment, whilst the cut-weight figures only show approximately 13-19% degradation for the mapping algorithm. Table15 summarises these four previous tables and includes similar comparisons for the 1d amp; 2d arrays and for partitioning times. Here we can clearly see that the mapping algorithm has the greatest effect for the 1d array with its very sparse processor graph, particularly as C8 increases.... ..."

Cited by 18

Table 15: Averages of processor assignment compared with mapping for average dilation ratios, cut-weight and partitioning times.

in Multilevel Mesh Partitioning for Heterogeneous Communication Networks

by C. Walshaw, M. Cross 2001

"... In PAGE 20: ... Even so the average dilation gures are still around 2 times worse on average for partitioning/assignment, whilst the cut-weight gures only show approximately 13-19% degradation for the mapping algorithm. Table15 summarises these four previous tables and includes similar comparisons for the 1d amp; 2d arrays and for partitioning times. Here we can clearly see that the mapping algorithm has the greatest effect for the 1d array with its very sparse processor graph, particularly as a5 increases.... ..."

Cited by 18

Table 4 Timing driven partitioning placement vs. min-cut driven partitioning placement. Ratio between our and min-cut partitioning-based placement algorithms Circuit Delay (best out of 6 runs) Delay (avg. of 6 runs) HPWL CPU(s)

in BY

by Cristinel Ababei 2004

"... In PAGE 11: ...erformed on the same machine as those in Section 3.4. 16 Table 3 Comparison between our methods and pure hMetis. 37 Table4 Timing driven partitioning placement vs. min-cut driven partitioning placement.... ..."

Table 7.9 The isoperimetric ratio obtained by applying the criterion cut method to the output of the isoperimetric and spectral partitioning algorithms (see section 7.4).

in Isoperimetric partitioning: A new algorithm for graph partitioning

by Leo Grady, Eric L. Schwartz 2006

Cited by 3

Table 7.9 The isoperimetric ratio obtained by applying the criterion cut method to the output of the isoperimetric and spectral partitioning algorithms (see section 7.4).

in Isoperimetric partitioning: A new algorithm for graph partitioning

by Leo Grady, Eric, L. Schwartz 2006

Cited by 3

Table 7.9 The isoperimetric ratio obtained by applying the criterion cut method to the output of the isoperimetric and spectral partitioning algorithms (see Section 7.4).

in Isoperimetric partitioning: A new algorithm for graph partitioning

by Leo Grady, Eric, L. Schwartz 2006

Cited by 3

Table 7.11 The mean isoperimetric ratio obtained using a criterion cut on the output of the partitioning algorithms when applied to three graph families of theoretical interest (see Section 7.4). Means are calculated over a range of parameters defining the graph family (see text for details).

in Isoperimetric partitioning: A new algorithm for graph partitioning

by Leo Grady, Eric, L. Schwartz 2006

Cited by 3

Table 5: Min cut and ratio cut ( 10?5) comparisons for MELO derived 2-way partitionings versus Paraboli [15] and EIG1 [11]. Paraboli results are quoted from [15]. All three algorithms require each cluster to contain at least 45% of the total modules. MELO runtimes are given for a Sun Sparc 10 and reported in seconds.

in Spectral Partitioning: The More Eigenvectors, The Better

by Charles J. Alpert, So-Zen Yao 1995

"... In PAGE 5: ...23% improvement over MELO). Table5 also reports the runtimes required for MELO to construct and split orderings using two and ten eigenvectors, after the eigenvectors have been computed. Despite MELO apos;s O(dn2) complexity, the runtimes are quite reasonable because the algorithm is so simple (see [1] for detailed runtimes for eigenvector computations).... ..."

Cited by 56

TABLE III COMPARISON OF VARIOUS CUTSIZE DRIVEN BOTTOM-UP CLUSTERING ALGORITHMS IN TERMS OF MULTILEVEL 7682 BIPARTITIONING RESULTS.ALGORITHMS IN COMPARISON INCLUDE ABSORPTION (656683), DENSITY (686978), RENT PARAMETER (826980), RATIO CUT (828467), CLOSENESS (677679), CONNECTIVITY (677978), FIRST CHOICE (7067), AND OUR EDGE SEPARABILITY BASED METHOD (698367). RATIO SHOWS THE NORMALIZED TOTAL TO 698367 RESULTS. TIME SHOWS THE TOTAL PARTITIONING TIME

in Edge Separability-Based Circuit Clustering with Application to Multilevel Circuit Partitioning

by Jason Cong, Sung Kyu Lim 2004

Cited by 5

TABLE I Comparison of various bottom-up clustering algorithms in terms of bipartitioning results. All algorithms use two-phase bottom-up clustering and top-down partitioning framework explained in Section III-D. Algorithms in comparison include Absorption (ABS) [18], Density (DEN) [9, 11], Rent Parameter (REP) [15], Ratio Cut (RTC) [19], Closeness (CLO) [17], Connectivity (CON) [16], and Edge Separability (ESC) based method. TIME includes total clustering and partitioning time.

in Edge Separability Based Circuit Clustering with Application to Circuit Partitioning

by Jason Cong, Sung Kyu Lim 2000

Cited by 41

Results 1 - 10 of 28,280

Tools

Table 15: Averages of processor assignment compared with mapping for average dilation ratios, cut-weight and partitioning times.

Table 15: Averages of processor assignment compared with mapping for average dilation ratios, cut-weight and partitioning times.

Table 4 Timing driven partitioning placement vs. min-cut driven partitioning placement. Ratio between our and min-cut partitioning-based placement algorithms Circuit Delay (best out of 6 runs) Delay (avg. of 6 runs) HPWL CPU(s)

Table 7.9 The isoperimetric ratio obtained by applying the criterion cut method to the output of the isoperimetric and spectral partitioning algorithms (see section 7.4).

Table 7.9 The isoperimetric ratio obtained by applying the criterion cut method to the output of the isoperimetric and spectral partitioning algorithms (see section 7.4).

Table 7.9 The isoperimetric ratio obtained by applying the criterion cut method to the output of the isoperimetric and spectral partitioning algorithms (see Section 7.4).

Table 7.11 The mean isoperimetric ratio obtained using a criterion cut on the output of the partitioning algorithms when applied to three graph families of theoretical interest (see Section 7.4). Means are calculated over a range of parameters defining the graph family (see text for details).