13.07.2015 Views

Porosity Aware Buffered Steiner Tree Construction - Computer ...

Porosity Aware Buffered Steiner Tree Construction - Computer ...

Porosity Aware Buffered Steiner Tree Construction - Computer ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS, VOL. XX, NO. Y, MONTH 2003 105SourceSourcev4v1v4v1v5v2v5v2v3v3(a)(b)Fig. 7. An example of local blockage avoidance. Shaded rectangles represent buffer blockages. The <strong>Steiner</strong> tree before blockage avoidance (a) and afterblockage avoidance (b).our algorithm into an industrial design flow to evaluate impacton overall circuit timing performance.A. Experiments Considering <strong>Porosity</strong>76stages 1−4stages 1−3−4stages 1−2−3−4TABLE IBENCHMARK CIRCUITS.THE SLACK UNIT IS ps.Net # sinks Max slack Min slack Grid sizemcu0s5 18 6596.0 6195.3 29 × 31mcu1s9 19 6507.3 6247.7 40 × 30n1071 17 2560.4 1902.4 26 × 42n18905 29 6650.3 609.5 58 × 46n313 19 6704.4 1232.5 52 × 65n7866 32 6703.5 96.9 88 × 36n8692 21 6390.1 1053.9 74 × 26n8702 43 6588.5 739.3 64 × 41n8730 20 6656.2 729.5 58 × 35netbig1 88 159564.7 1974 65 × 90netbig2 79 65837.9 104.2 47 × 31netbig3 63 40643.2 1327.4 42 × 36Total porosity cost54321100 150 200 250 300 350 400 450Slack improvement/psFig. 8.The cost-slack trade-off curves for net mcu0s5.The statistics for the benchmark circuits are shown inTable I. These nets are extracted from industrial circuits withrandomly generated wire congestion map. The maximum slackand the minimum slack among all sinks in the initial C-<strong>Tree</strong>is given for each net. The algorithms are implemented inC++ and the experiments are performed on a SUN Ultra-4workstation of 400MHz and 2Gb memory.In each set of experiments, we compare three variants: (1)1-4 in which only stage 1 and 4 are performed; (2) 1-3-4 inwhich stage 3 of blockage avoidance is run between stage 1and 4; (3) 1-2-3-4 which the complete four-stage algorithm. Instage 2, the weighting factor for porosity cost and wirelengthis α =1and γ =0.5, respectively. The metric we observeincludes:• Slack improvement: the change of the minimum slackwith respect to initial C-<strong>Tree</strong> result in ps.• #bufs: number of buffers inserted.• Average(P ave ) and the maximum(P max ) porosity cost forthe inserted buffers.• Total wirelength.• CPU time in seconds.TABLE IIISTATISTICS FOR THE POROSITY COST ONLY EXPERIMENTAL RESULTS.THE SLACK IMPROVEMENT IS THE DIFFERENCE FROM 1-4 FLOW.POROSITY COST IS NORMALIZED WITH RESPECT TO THE POROSITY COSTFROM 1-4 FLOW. THE MIN-AVE-MAX IS THE MINIMUM, THE AVERAGEAND MAXIMUM VALUE AMONG ALL NETS.1-3-4 1-2-3-4Min Ave Max Min Ave MaxSlack improve(ps) -9.7 81.4 408.3 -19.3 200.7 814.0Avg porosity cost 0.74 1.12 1.66 0.79 0.91 1.04Max porosity cost 0.63 1.68 4.80 0.45 0.79 1.00

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!