1 / 20

Fast Timing-Model Independent Buffered Clock-Tree Synthesis

Fast Timing-Model Independent Buffered Clock-Tree Synthesis. Xin -Wei Shih and Yao- Wen Chang. Outline. Introduction Problem formulation Algorithms Experimental results Conclusions. Introduction.

vala
Télécharger la présentation

Fast Timing-Model Independent Buffered Clock-Tree Synthesis

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Fast Timing-Model Independent Buffered Clock-TreeSynthesis Xin-Wei Shih and Yao-WenChang

  2. Outline • Introduction • Problem formulation • Algorithms • Experimental results • Conclusions

  3. Introduction • Skew-minimized buffered clock-tree synthesis plays an important role in high-performance VLSI designs for synchronous circuits. • Due to the insufficient accuracy of existing timing models for modern chip design, embedding simulation process into a clock-tree synthesis flow becomes inevitable.

  4. Introduction • A possible way to improve the speed is performing the clock construction by structure optimization. • Mesh • In this paper, a novel timing-model independent buffered clock treesynthesis method is proposed. • Buffering and wiring structures of all paths from the clock source to its sinks are almost the same.

  5. Problem formulation • Problem: Buffered Clock-Tree Synthesis (BCTS) • Instance: Given a set of clock sinks, a slew-rate constraint, and a library of buffers. • Question: Construct a buffered clock tree to minimize its skew, subject to no slew-rate violation.

  6. Algorithm

  7. Branch-Number Planning • The number of leaves (sinks) can be treatedas a multiplication sequence of branching. • This multiplication sequenceexactly forms a factorization. • Then, the BNP is arranged in the non-increasing order1 of thefactorization list.

  8. Branch-Number Planning

  9. Tree Construction-Partitioning • A top-down manner like [10]or a bottom-up one like [7, 11], they can hardly apply to nonbinarytree structures. • Therefore, we propose a novel partitioningmethod, which can not only handle non-binary tree structures, butalso achieve good quality in terms of the cluster diameter. • cluster diameter : the maximum distance among sub-trees withinthe same cluster.

  10. Partitioning • We borrow the idea of cake cutting, i.e., slicinga cake into pieces from the center of the cake.

  11. Embedding-Region Construction

  12. Node Embedding

  13. Pseudo Sink Handling • Since the identical branch numbers at the same levelare required in the symmetrical structure, a pseudo sink should betransformed into a dangling wire to maintain the symmetry. • For partitioning,we relax that the sizes of clusters in a partition can differ by atmost one for the first recursion. • For nodeembedding, we let the embedding regions of pseudo sinks cover theentire chip.

  14. Buffer Insertion • A top-down manner • By tracing along the tree edges, oncethe slew rate is about to violate the constraint, identical buffersare inserted for all branches. • Insert identical buffers in terms of the typeand the size at the same level. • The slew rate is approximated byaccumulated capacitance starting from the latest inserted buffer.

  15. Buffer Insertion

  16. Experimental results • Implemented in the C++ programming language on a 2.6 GHz AMD-64 workstation. • Four ISPD’09 Clock Network Synthesis Contest benchmarks with no blockages [17] and the IBM benchmarks [19]. • Use ngspice [13] simulation based on the 45nm process technology [14] to evaluate the quality.

  17. Experimental results • clock skew (skew) • clock-latency range(CLR) • total resource usage (usage)

  18. Experimental results

  19. Experimental results

  20. Conclusions • We have presented a fast timing-model independent buffered clock tree synthesis method to construct a symmetrical clock tree with little wiring overhead. • By symmetrically constructing a clock tree, the clock skew can be minimized without referring to simulation information.

More Related