580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How can architectural choices in Winner-Take-All (WTA) and Loser-Take-All (LTA) circuits optimize power, speed, and precision in analog neuromorphic systems?
This theme investigates different circuit architectures for WTA/LTA implementations, critical building blocks in analog neural networks and signal processing. Understanding how architectures impact power consumption, speed (frequency response), accuracy, and silicon area is crucial for advancing neuromorphic hardware capable of real-time computations and low power operation in CMOS technology.
Key finding: This comprehensive review identifies three primary architectures for WTA/LTA circuits: current conveyors, binary tree, and time-domain WTAs. Current conveyor architectures, while simple, suffer from exponential complexity... Read more
2. What optimization techniques can automate transistor sizing to improve performance and reduce design iteration in custom MOS circuits?
Transistor sizing critically influences delay, power, and area in custom MOS analog and digital circuits. Traditional sizing methods are manual, relying heavily on designer intuition and iterative simulations, which are time-consuming and error-prone. Automating the sizing process via optimization enables rapid design space exploration, improved design reuse, and performance guarantees. This theme focuses on algorithmic approaches and software toolchains that integrate simulation, sensitivity analysis, and nonlinear optimization for transistor sizing.
Key finding: The paper introduces JiffyTune, an automated transistor sizing tool leveraging dynamic circuit simulation with gradient-based nonlinear optimization via LANCELOT and SPECS simulator. The tool handles multi-objective... Read more
3. How can quantum computing paradigms revolutionize the design of combinational and sequential logic circuits?
Quantum computing leverages qubits’ superposition and entanglement to perform computations more efficiently than classical digital binary logic in certain contexts. Designing classical logic elements like flip-flops and combinational circuits using quantum gates introduces new opportunities for parallelism, reduced delay, and potentially lower power use. This theme explores methodologies for mapping classical circuit functionality onto quantum architectures, evaluating tradeoffs in gate counts, delays, and error rates.
Key finding: This study demonstrates the design and simulation of flip-flops and combinational logic circuits using quantum computing frameworks (Qiskit). By employing reversible quantum gates such as Toffoli, Fredkin, and Hadamard, the... Read more
4. What are the complexities and computational capabilities of monotone planar Boolean circuits and their implications for circuit design and learning?
Monotone planar circuits (circuits with only AND, OR gates arranged in a plane without wire crossings) impose topological restrictions that affect computational power and complexity. Understanding these tradeoffs informs logic circuit layout optimizations, complexity theory, and efficient hardware realizations. Furthermore, exploring their learnability with membership queries has implications for automated circuit design and verification.
Key finding: This paper establishes that monotone planar circuits with bounded width and negated inputs exactly characterize non-uniform AC0, contrasting with non-planar circuits that characterize NC1. It also improves algorithmic results... Read more
5. How can open-source EDA toolkits support scalable, automated physical implementation flows for modern chip design?
EDA software suites are critical in converting RTL code and netlists into manufacturable chip layouts. The rise of open-source EDA platforms enables collaborative innovation, accelerates research, and reduces barriers to entry by providing accessible, scalable tools covering floorplanning, placement, routing, and timing analysis. This theme investigates infrastructure design, modular tool architectures, AI integration for optimization, and validation through real chip tape-outs.
Key finding: The iEDA project presents an open-source physical design platform covering full flow steps including floorplan, placement, CTS, routing, and timing optimizations with integrated static timing and power analysis. iEDA... Read more