Adaptive body biasing is a powerful technique that allows post-silicon tuning of individual manufactured dies such that each die optimally meets the delay and power constraints. Assigning individual bias control to each gate leads to severe overhead, rendering the method impractical. However, assigning a single bias control to all gates in the circuit prevents the method from compensating for intra-die variation and greatly reduces its effectiveness. In this paper, we propose a new variability-aware method that clusters gates at design time into a handful of carefully chosen independent body-bias groups, which are then individually tuned post-silicon for each die. We show that this allows us to obtain near-optimal performance and power characteristics with minimal overhead. For each gate, we generate the probability distribution of its post-silicon ideal body bias voltage using an efficient sampling method. We then use these distributions and their correlations to drive a statistically aware clustering technique. We study the physical design constraints and show how the area and wirelength overhead can be significantly limited using the proposed method. Compared with a fixed design-time based dual threshold voltage assignment method, we improve leakage power by 38%-68% while simultaneously reducing the standard deviation of delay by two to nine times.

Extensive research has proposed the use of multiple on-die power supplies (VDD) for reducing power consumption in CMOS circuits. We present a detailed study and design techniques for power delivery systems in dual VDD CMOS circuits. We first show that the total current to be delivered by the voltage supplies is significantly reduced (by 27%-46%) in dual VDD circuits. This current reduction prompts various design strategies that can be employed to design the power delivery system. We describe issues that arise at the system, board and package levels and propose a high-level model for the same. We then provide a new placement driven approach for designing on-die dual VDD power grids. Compared to already existing methods, the dual VDD grids generated by our approach reduce the worst case and average voltage drop by up to 12.3% and 6.8% respectively with no area overhead and sometimes improving wire congestion. We also show that dual VDD circuits can afford lower on-die decoupling capacitance budgets

Adaptive body biasing (ABB) is a powerful technique that allows post-silicon tuning of individual manufactured dies such that each die optimally meets the delay and power constraints. Assigning individual bias control to each gate leads to severe overhead, rendering the method impractical. However, assigning a single bias control to all gates in the circuit prevents the method from compensating for intra-die variation and greatly reduces its effectiveness. In this paper, we propose a new variability-aware method that clusters gates at design time into a handful of carefully chosen independent body bias groups, which are then individually tuned post-silicon for each die. We show that this allows us to obtain near-optimal performance and power characteristics with minimal overhead. For each gate, we generate the probability distribution of its post-silicon ideal body bias voltage using an efficient sampling method. We then use these distributions and their correlations to drive a statistically-aware clustering technique. We study the physical design constraints and show how the area and wirelength overhead can be significantly limited using the proposed method. Compared to a fixed design time based dual threshold voltage assignment method, we improve leakage power by 38-71% while simultaneously reducing the standard deviation of delay by 2-9X.

We present the first in-depth study of the two existing algorithms, namely, Clustered Voltage Scaling (CVS) and Extended Clustered Voltage Scaling (ECVS), used for assigning the voltage supply to gates in integrated circuits having dual power supplies. We present a comparison of the achievable power savings using these algorithms on various benchmark circuits and first point out that ECVS does provide appreciably larger power improvements compared to CVS. We then provide a new algorithm based on ECVS that further improves the power savings by efficient assignment of the power supplies to the gates. Our new algorithm provides up to 66% power reduction and improves the power savings by up to 28% and 13% with respect to CVS and ECVS respectively. Furthermore, since level conversion is an essential component of dual power supply systems we also present the first circuit-specific sensitivity study of achievable power savings to the energy and delay penalties imposed by level conversion.

Multi-V/sub DD/ design is an effective way to reduce power consumption, but the need for level conversion imposes delay and energy penalties that limit the potential gains. In this paper, we describe new level converting circuits that provide 10%-61% lower energy consumption at equivalent or better speeds compared to those available in the literature. Furthermore, we make the argument that level converters should be evaluated largely by their maximum speed since slower level converters consume valuable timing slack that can be used to reduce the energy of other gates in the circuit. Based on this criterion, we find the new structures to offer up to a 25% speed improvement over conventional level converters. Using an efficient dual V/sub DD/ voltage assignment algorithm, we show that this speed improvement can yield a reduction of up to 7.3% in total circuit power in small benchmark circuits. We also propose embedding the functionality of logic gates into the level converting circuits. For typical values of the second supply voltage, this technique can reduce delay by 15% at constant energy or lower energy by up to 30% at fixed delay.

Power dissipation is becoming the most challenging design constraint in nanometer technologies. Among various design implementation schemes, standard cell ASICs offer the best power efficiency for high-performance applications. The flexibility of ASICs allow for the use of multiple voltages and multiple thresholds to match the performance of critical regions to their timing constraints, and minimize the power everywhere else. We explore the trade-off between multiple supply voltages and multiple threshold voltages in the optimization of dynamic and static power. The use of multiple supply voltages presents some unique physical and electrical challenges. Level shifters need to be introduced between the various voltage regions. Several level shifter implementations will be shown. The physical layout needs to be designed to ensure the efficient delivery of the correct voltage to various voltage regions. More flexibility can be gained by using appropriate level shifters. We will discuss optimization techniques such as clock skew scheduling which can be effectively used to push performance in a power neutral way.