Conference Paper

A new paradigm for low-power, variation-tolerant circuit synthesis using critical path isolation.

DOI: 10.1145/1233501.1233628 Conference: 2006 International Conference on Computer-Aided Design (ICCAD'06), November 5-9, 2006, San Jose, CA, USA
Source: DBLP

ABSTRACT Design considerations for robustness with respect to variations and low power operations typically impose contradictory design requirements. Low power design techniques such as voltage scaling, dual-Vth etc. can have a large negative impact on parametric yield. In this paper, we propose a novel paradigm for low-power variation- tolerant circuit design, which allows aggressive voltage scaling. The principal idea is to (a) isolate and predict the set of possible paths that may become critical under process variations, (b) ensure that they are activated rarely, and (c) avoid possible delay failures in the critical paths by dynamically switching to two-cycle operation (assuming all standard operations are single cycle), when they are activated. This allows us to operate the circuit at reduced supply voltage while achieving the required yield. Simulation results on a set of benchmark circuits at 70nm process technology show average power reduction of 60% with less than 10% performance overhead and 18% overhead in die-area compared to conventional synthesis. Application of the proposed methodology to pipelined design is also investigated.

  • [Show abstract] [Hide abstract]
    ABSTRACT: A novel circuit architecture for variable latency adder based on present and transitional states prediction (PTSP) method is presented in this paper, for taking the low power benefits of voltage-over-scaling. With the scaling down of CMOS technology, failure from process variation and high power consumption has become severe problem in VLSI design and the traditional conservative methodology is about to reach its limit. The technique of adaptive clocking has been proved promising to jointly address the mentioned two issues above. Previous works have focused on two or multi-stage predictions of present input data with error recovery but ignored the data correlation, which could result plenty of redundant cycles. In this work, along with the present data, sequence dependence between successive data is also introduced into function speculation and realized by a simple feedback strategy. Analytical energy saving and performance models have been deduced and validated by simulation using Hspice with 65nm CMOS technology, where the redundant cycles are eliminated up to 16% and the maximum energy saving is 15% with 3% area overhead, being compared with conventional adaptive clocking adder. Furthermore, the new adder with PTSP is applied to the domain of approximate computation and gets a decrement in error deviation of up to 50% in an accumulator.
    Power and Timing Modeling, Optimization and Simulation (PATMOS), 2013 23rd International Workshop on; 01/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: The ever-increasing parametric variations in the latest nanometer technologies pose a severe reliability challenge for VLSI design. Specifically, technology scaling leads to an increasing performance variation even when the average performance improves. The traditional VLSI design methodology requires that all computation in a logic stage complete within one clock cycle. This has hindered further performance improvement. Alternatively, allowing computation in a logic stage to complete in a variable number of clock cycles leads to average performance improvement and enables further power reduction. In this paper, we present a generic variable-latency design methodology, which includes timing analysis, delay test input generation, design of a completion prediction unit for logic computation latency, and a clock gating scheme. Our experiments based on on the 45nm Nangate open cell library and the des MCNC benchmark circuit show that, for a clock gating occurrence probability 6.25%, our technique leads to maximum 8.29%, 9.96%, and 9.18% area reduction, and 27.54%, 28.08%, and 29.93% power reduction with a prediction unit of 4, 5, and 6 inputs, predicting the top 1, 2, and 4 timing-critical paths, respectively.
    Circuits and Systems (MWSCAS), 2013 IEEE 56th International Midwest Symposium on; 01/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: As CMOS technology driven by Moore's law has approached device sizes in the range of 5-20 nm, noise immunity of such future technology nodes is predicted to decrease considerably, eventually affecting the reliability of computations through them. A shift in the design paradigm is expected from 100% accurate computations to probabilistic computing with accuracy dependent on the target application or circuit specifications. One model developed for CMOS technology that emulates the erroneous behavior predicted is termed probabilistic CMOS (PCMOS). In this paper, we propose a PCMOS-based architecture implementation for traditional motion estimation algorithms and show that up to 57% energy savings are possible for different existing motion estimation algorithms. Furthermore, algorithmic modifications are proposed that can enhance the energy savings to 70% with a PCMOS architectural implementation. About 1.8-5 dB improvement in peak signal-to-noise ratio under energy savings of 57% to 70% for two different motion estimation algorithms is shown, establishing the resilience of the proposed algorithm to probabilistic computing over the comparable conventional algorithm.
    IEEE Transactions on Circuits and Systems for Video Technology 01/2014; 24(1):1-14. DOI:10.1109/TCSVT.2013.2273627 · 2.26 Impact Factor


Available from