1 / 10

The BDS Circuit Synthesis System

The BDS Circuit Synthesis System. What it Does and Doesn’t Do. What it Is. BDS (BDD Based Logic Optimization System?) is a multi-level circuit optimization system that uses local BDD representations for circuit optimization

lloyd
Télécharger la présentation

The BDS Circuit Synthesis System

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. The BDS Circuit Synthesis System What it Does and Doesn’t Do

  2. What it Is • BDS (BDD Based Logic Optimization System?) is a multi-level circuit optimization system that uses local BDD representations for circuit optimization • BDS is written by Congguang Yang and Maciej Ciesielski at the University of Massachusets

  3. How it Works (very briefly) • Builds local BDDs to represent portions of a boolean network (and therefore can handle arbitrarily large circuits without memory blowup) • Looks for dominators and generalized dominators to factor and simplify the BDD representation

  4. What’s a Dominator, Anyway? • A “1-dominator” is a node in a BDD that is situated on every path from the root to the “1” terminal. Similarly, “0-dominators” are defined. It can be shown that the presence of one of these leads to a certain algebraic decomposition of the BDD. • “Generalized dominators” are nodes with similar properties with respect to other decompositions.

  5. What it Does • Takes input as a .blif (Berkeley Logic Interchange Format) file • Outputs a circuit in terms of not, and, or, xnor, & implication gates • Can output in either .blif or .dot (for GraphVis) • Beats SIS (the standard multilevel circuit minimization tool) on a handful of benchmarks

  6. .BLIF files • Might look something like this .model model_name .inputs i1 i2 i3 i4 .outputs o1 .names i1 i2 i3 i4 o1 1-1- 1 # i am a comment 10-1 1 110- 1 .end

  7. What it Doesn’t Do • Guarantee optimal circuits (even in small cases) – but it does find them for some large cases • Use an arbitrary gate library • Do any internal don’t care based optimization

  8. Circuit Optimization - starring don’t cares • One paradigm of circuit optimization is to iteratively look at relatively small subcircuits, and replace them with optimized versions. • It is almost inevitable that the subcircuits will never recieve certain input configurations, or alternatively that certain output configurations are completely equivalent. • This yields more freedom in the choice of a replacement subcircuit – which of course can lead to smaller circuits.

  9. Don’t Cares -in two flavors • Satisfiability: The set of inputs that will never occur to a given subcircuit. • Observability: The set of inputs for which the subcircuit does not affect the circuit’s value • The union of these sets form the don’t care set of a subcircuit

  10. Don’t Cares -in quantum circuits • Quantum computers have don’t care’s of their own • The “work bits” ubiquitous in quantum algorithms are the reversible equivalent of classical don’t cares (why?) • Because of the probabilistic measurement step, a new sort of don’t cares (with no classical analogue) is introduced.

More Related