Simultaneous peak and average power minimization during datapath scheduling for DSP processors

Quantum mechanical principles that govern the basic laws of physics increasingly limit CMOS operation with transistor scaling. Traditional logic based CMOS circuits cannot achieve ultra-low power levels due to heat dissipated for a single bit loss of information as represented by the Landauer barrier. Reversible logic is a promising computing paradigm towards realization of ultra-low power computing circuits. Reducing average and peak power consumption is an effective strategy for mitigation of side-channel attacks, such as Differential Power Analysis. We present designs of Forward Body Biased Adiabatic Logic for reduction of average, peak, and differential power. HSPICE simulations with predictive 22nm technology are used to analyze performance metrics and exhaustive simulation results are presented for various reversible CMOS designs. Average power is improved upon by up to 91%, the peak power by up to 96%, and the differential power is improved by up to a factor of 128.57.

In this paper, we present a combination of basic retiming and multiple voltage scheduling (MVS) techniques in order to optimize dynamic peak power as well as average power consumption in synchronous sequential circuits under timing constraints. First, we devise a mixed-integer linear programming (MILP) formulation for the problem of scheduling for optimal peak and/or average power consumption through a unification of retiming and MVS techniques. Then, to alleviate the problem of variable explosion in MILP, we present a two-stage algorithm for peak and average power optimization. First, power-oriented retiming is proposed to restructure the input SDFG in order to achieve parallelization to favor nodes with high power consumption followed by an MILP formulation for peak and/or average power optimization using MVS technique

Recently, dynamic frequency scaling has been explored at the CPU and system levels for power optimization. Low-power datapath scheduling using multiple supply voltages has been well researched. In this work, we develop new datapath scheduling algorithms that use multiple supply voltages and dynamic frequency clocking in a coordinated manner in order to reduce the energy consumption of datapath circuits. In dynamic frequency clocking, the functional units can be operated at different frequencies depending on the computations occurring within the datapath during a given clock cycle. The strategy is to schedule high-energy units, such as multipliers at lower frequencies, so that they can be operated at lower voltages to reduce energy consumption and the low-energy units, such as adders at higher frequencies, to compensate for speed. The proposed time- and resource-constrained algorithms have been applied to various high-level synthesis benchmark circuits under different time and resource constraints. The experimental results show significant reduction in energy for both the algorithms.

With advancing technology, large dynamic power consumption has significantly limited circuit miniaturization. Minimizing peak power consumption, which is defined as the maximum power consumption among all voltage partitions, is important since it enables energy saving from the voltage island shutdown mechanism. In this paper, we prove that the peak power driven voltage partitioning problem is NP-complete and propose an efficient provably good fully polynomial time approximation scheme for it. The new algorithm can approximate the optimal peak power driven voltage partitioning solution in O(m <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> (mn/ϵ <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sup> )) time within a factor of (1 + ϵ) for sufficiently small positive e, where n is the number of circuit blocks and m is the number of partitions which is a small constant in practice. Our experimental results demonstrate that the dynamic programming cannot finish for even 20 blocks while our new approximation algorithm runs fast. In particular, varying e, orders of magnitude speedup can be obtained with only 0.6% power increase. The tradeoff between the peak power minimization and the total power minimization is also investigated. We demonstrate that the total power minimization algorithm obtains good results in total power but with quite large peak power, while our peak power optimization algorithm can achieve on average 26.5% reduction in peak power with only 0.46% increase in total power. Moreover, our peak power driven voltage partitioning algorithm is integrated into a simulated annealing based floorplanning technique. Experimental results demonstrate that compared to total power driven floorplanning, the peak power driven floorplanning can significantly reduce peak power with only little impact in total power, HPWL, estimated power ground routing cost, level shifter cost and runtime. Further, when the voltage island shutdown is performed, peak power driven voltage partitioning can lead to over 10% more energy saving than a greedy frequency based voltage partitioning when multiple idle block sequences are considered.

We present a high peak and average power laser system with ultrashort pulses at high repetition rates. Pulse shortening and peak power enhancement of a state-of-the-art fiber laser system is achieved by utilizing nonlinearity, namely selfphase modulation and subsequent compression in a chirped mirror compressor. The nonlinear interaction is achieved by propagation in a noble gas filled hollow core fiber with an inner diameter of 200 &mu;m and a length of 0.5 m. A total second dispersion of -7000 fs² is applied by a chirped mirror compressor resulting in ultrashort pulses of 71 fs duration. This is achieved by coupling 400 MW, 800 fs pulses from the CPA system to the Xenon filled hollow core fiber. The average power at the output of the compressor is measured to be 10 W at 50 kHz repetition rate resulting in 200 &mu;J pulse energy. Hence, the compressed pulses have a peak power of more than 2 GW. Consequently, the pulses of the CPA system are shortened by a factor of ten and the peak power is enhanced by a factor of 5. In addition this approach offers further peak and average power scalability.

A new trellis shaping design is proposed for reducing the peak-to-average power ratio of the bandlimited orthogonal frequency-division multiplexing (OFDM) signals. The approach is based on recursive minimization of the autocorrelation sidelobes of an OFDM data sequence. A novel metric in conjunction with the Viterbi algorithm is devised. The performance of the trellis shaping depends on signal mapping strategy, and the two types of mapping, referred to as Type-I and Type-II, are proposed. The Type-I mapping has no capability of reducing the average power, but it can achieve a significant reduction of the peak-to-average power ratio. On the other hand, the Type-II mapping is designed to achieve both peak and average power reduction. The bit error probability of the system over an AWGN channel is evaluated based on the simulations, which confirms the effectiveness of the proposed scheme.

We present an algorithm called MOVER (Multiple Operating Voltage Energy Reduction) to minimize datapath energy dissipation through use of multiple supply voltages. In a single voltage design, the critical path length, clock period, and number of control steps limit minimization of voltage and power. Multiple supply voltages permit localized voltage reductions to take up remaining schedule slack. MOVER initially finds one minimum voltage for an entire datapath. It then determines a second voltage for operations where there is still schedule slack. New voltages con be introduced and minimized until no schedule slack remains. MOVER was exercised for a variety of DSP datapath examples. Energy savings ranged from 0% to 50% when comparing dual to single voltage results. The benefit of going from two to three voltages never exceeded 15%. Power supply costs are not reflected in these savings, but a simple analysis shows that energy savings can be achieved even with relatively inefficient DC-DC converters. Datapath resource requirements were found to vary greatly with respect to number of supplies. Area penalties ranged from 0% to 170%. Implications of multiple voltage design for IC layout and power supply requirements are discussed.

Multiple Dynamic Supply Voltage (MDSV) is a technique that focuses on reducing the dynamic power. This technique is an evolution of the Multiple Supply Voltage (MSV). MSV and MDSV introduce some difference on traditional physical synthesis due to the different voltage operations of each region in the design. To convert the voltage among regions supplied by different voltages, these techniques insert a Level Shifter (LS) circuit. As a penalty consequence, traditional LS circuits insert a delay and extra power consumption in the design. The dynamical behavior of MDSV designs has brought a new condition, where LS inserted in the circuit can be in an idle state temporally. This work presents a new architecture to reduce the power consumption and delay, bypassing the LS. The architecture explores an alternative path to current flow in the cases that LS is idle. With this new approach we reduce up to 15% of power consumption and up to 75% and 15% of delay.

Previous research has demonstrated a reduction in peak power via isokinetic dynamometry and power testing. The purpose of this study was to determine the effects of static stretching (SS) on the Wingate test (WAnT). Thirteen recreationally active males (mean age: 22.5±1.9 years, height: 172.5±8.4 cm, body mass: 83.4±24.6 kg) were tested for peak (PP), average (AP), and minimum (MP) power output using the WAnT. Two WAnT trials were completed by each subject on a Monark 894-E ergometer that was interfaced with a desktop computer. Participants completed a 5-minute warm-up on the cycle ergometer and then performed the pre-WAnT. Upon completion, the stretching protocol four static stretches were performed: one was participant self-administered and the remaining three by the investigator. Each stretch was held for 30 seconds and was repeated 4 times on each leg and repeated after 4 minutes of rest. Significant differences (P<0.05) were found for PP only for the pre- versus post-WAnT (802.6±207.6 vs. 768.8±199.7 W). Static stretching before performance of the WAnT demonstrated a significant decrease in peak power output. Based upon these results, it appears that SS decreases peak power output when performing an anaerobic capacity test on a cycle ergometer.

Objective: The aim of this study was to determine the effects of sprint interval exercises (SIT) conducted under different conditions (hypoxia and blood flow restriction [BFR]) on mechanical, cardiorespiratory, and muscular O2 extraction responses.Methods: For this purpose, 13 healthy moderately trained men completed five bouts of 30 s all-out exercises interspaced by 4 min resting periods with lower limb bilateral BFR at 60% of the femoral artery occlusive pressure (BFR60) during the first 2 min of recovery, with gravity-induced BFR (pedaling in supine position; G-BFR), in a hypoxic chamber (FiO2≈13%; HYP) or without additional stress (NOR). Peak and average power, time to achieve peak power, rating of perceived exertion (RPE), and a fatigue index (FI) were analyzed. Gas exchanges and muscular oxygenation were measured by metabolic cart and NIRS, respectively. Heart rate (HR) and peripheral oxygen saturation (SpO2) were continuously recorded.Results: Regarding mechanical responses, peak and average power decreased after each sprint (p < 0.001) excepting between sprints four and five. Time to reach peak power increased between the three first sprints and sprint number five (p < 0.001). RPE increased throughout the exercises (p < 0.001). Of note, peak and average power, time to achieve peak power and RPE were lower in G-BFR (p < 0.001). Results also showed that SpO2 decreased in the last sprints for all the conditions and was lower for HYP (p < 0.001). In addition, Δ[O2Hb] increased in the last two sprints (p < 0.001). Concerning cardiorespiratory parameters, BFR60 application induced a decrease in gas exchange rates, which increased after its release compared to the other conditions (p < 0.001). Moreover, muscle blood concentration was higher for BFR60 (p < 0.001). Importantly, average and peak oxygen consumption and muscular oxyhemoglobin availability during sprints decreased for HYP (p < 0.001). Finally, the tissue saturation index was lower in G-BFR.Conclusions: Thus, SIT associated with G-BFR displayed lower mechanical, cardiorespiratory responses, and skeletal muscle oxygenation than the other conditions. Exercise with BFR60 promotes higher blood accumulation within working muscles, suggesting that BFR60 may additionally affect cellular stress. In addition, HYP and G-BFR induced local hypoxia with higher levels for G-BFR when considering both exercise bouts and recovery periods.

Mitchell, JA, Pumpa, KL, and Pyne, DB. Responses of lowerbody power and match running demands after long-haul travel in international rugby sevens players. J Strength Cond Res 31(3): 686-695, 2017-This study determined the effect of long-haul (>5 hours) travel on lower-body power and match running demands in international rugby sevens players. Lower-body power was assessed in 22 male international rugby sevens players (age 21.7 ± 2.7 years, mass 89.0 ± 6.7 kg, stature 180.5 ± 6.2 cm; mean ± SD) monitored over 17 rugby sevens tournaments. A countermovement jump was used to monitor lower-body power (peak and mean power) over repeated three week travel and competition periods (pretravel, posttravel, and posttournament). Small decreases were evident in peak power after both short and long-haul travel (-4.0%, ±3.2%; mean, ±90% confidence limits) with further reductions in peak and mean power posttournament (-4.5%, ±2.3% and -3.8%, ±1.5%) culminating in a moderate decrease in peak power overall (-7.4%, ±4.0%). A subset of 12 players (completing a minimum of 8 tournaments) had the effects of match running demands assessed with lower-body power. In this subset, long-haul travel elicited a large decrease in lower-body peak (-9.4%, ±3.5%) and mean power (-5.6%, ±2.9%) over the monitoring period, with a small decrease (-4.3%, ±3.0% and -2.2%, ±1.7%) posttravel and moderate decrease (-5.4%, ±2.5% and -3.5%, ±1.9%) posttournament, respectively. Match running demands were monitored through global positioning system. In long-haul tournaments, the 12 players covered ∼13%, ±13% greater total distance (meter) and ∼11%, ±10% higher average game meters >5 m·s when compared with short-haul (<5 hours) travel. Effective pretravel and posttravel player management strategies are indicated to reduce neuromuscular fatigue and running load demands in rugby sevens tournaments after long-haul travel.

The declines in power during the performance of daily activities are associated with a reduced ability to perform basic and instrumented activities of daily living. The Gallon Jug Shelf Transfer Test (GJST) is a validated upper body muscle function test that can be used to quantify power and monitor functional changes in older persons. PURPOSE: Although the validity of the GJST and its predictive equations for transfer power has been established, the reliabilities of the GJST power equations need to be assessed if they are to be used in clinical or research environments. Therefore, this study aims to evaluate the intra-rater and test-retest reliabilities of the GJST. METHODS: 40 independently-living seniors between the ages of 50-90 performed two sets of 3 repetitions of the GJST separated by 48 hours on Day 1 and the same pattern of testing on Day 2. Using the predictive equations developed in our laboratory, the values for peak power (PP) and average power (AP) were then computed. Statistical analyses to assess various forms of reliability included calculating intraclass correlation coefficient (ICC), confidence intervals (CI), coefficient of variation (CV), standard error of measurement (SEM), and minimal detectable change (MDC) values. Further Bland-Altman plots evaluate the agreement among the tests. RESULTS: Intraclass correlation coefficients (ICC > .90, p < .0001), Cronbach’s alpha, and Bland-Altman analyses of between-trial and between-day test-retest reliability for peak and average power indicated excellent reliability. Cronbach’s α and ICC for trial 2 (days 1 & 2) and day 2 (trials 1 & 2) slightly increased, compared to trial 1 (days 1 & 2), and day 1 (trials 1 & 2). The lines of equality for all Bland-Altman plots, whether showing between-day or between-trial comparisons, fell within the LOA showing that there was no significance for the difference scores (p < .05). Bias, SD, LOA, CV, SEM, and MDC, slightly decreased for trial 2 (days 1 & 2) and day 2 (trials 1 & 2) compared to trial 1 (days 1 & 2), and day 1 (trials 1 & 2) for peak and average power. CONCLUSIONS: Based on our statistical analyses, the GJST has proven to be a safe, easy to administer, ADL- specific, and highly reliable assessment for determining upper body power for healthy older adults.

In this paper, we propose a method to solve the multiple supply voltage scheduling problem which is to assign the operational nodes of a control/data flow graph to a voltage level to minimize the average power consumption within a given computation time. Different from the previous researches focused on the operational nodes in the critical path and utilized the slack time to change the voltage of other nodes, our method can deal with all nodes without considering whether the node is in the critical path or not, and the benefit is that the voltage assignment of each node becomes more flexible. The proposed method consists of two phases, the scheduling phase and the adjusting phase, and considers both the power (delay) of the computational components and the power (delay) of the level shifters. Experimental result shows that using three voltages on a number of standard benchmarks, an average power saving of 34.23% can be obtained if the delay overhead is set as 0, and 48.07% can be obtained if the total delay is set as 1.6 times of the original delay.

In this paper, a tabu search-based behavior level synthesis scheme is proposed to minimize power consumption with resource operating at multiple voltages under the timing and the resource constraints. Unlike the conventional methods where only scheduling is considered, our synthesis scheme considers both scheduling and partitioning simultaneously to reduce power consumption due to the functional units as well as the interconnects among them. More importantly, our approach tends to efficiently address a few practical layout problems inherent to multiple voltage designs. In particular, we have configured our solutions as a three-tuple vector to account for both the timing and the partition. Cycling of the same solutions is prevented by applying a tabu list with an update mechanism enhanced with an aspiration function. In this way, the algorithm can search a large solution space with modest computation effort and fast convergence rate. Experiments with a number of DSP benchmarks show that the proposed algorithms achieve an average power reduction by 49.6%.

Physical training in heat or hypoxia can improve physical performance. The purpose of this parallel group study was to investigate the concurrent effect of training performed simultaneously in heat (31 °C) and hypoxia (FIO2 = 14.4%) on anaerobic capacity in young men. For the study, 80 non-trained men were recruited and divided into 5 groups (16 participants per group): control, non-training (CTRL); training in normoxia and thermoneutral conditions (NT: 21 °C, FIO2 = 20.95%); training in normoxia and heat (H: 31 °C, FIO2 = 20.95%); training in hypoxia and thermoneutral conditions (IHT: 21 °C, FIO2 = 14.4%), and training in hypoxia and heat (IHT + H: 31 °C, FIO2 = 14.4%). Before and after physical training, the participants performed the Wingate Test, in which peak power and mean power were measured. Physical training lasted 4 weeks and the participants exercised 3 times a week for 60 min, performing interval training. Only the IHT and IHT + H groups showed significant increases in absolute peak power (p < 0.001, ES = 0.36 and p = 0.02, ES = 0.26, respectively). There were no significant changes (p = 0.18) after training in mean power. Hypoxia appeared to be an environmental factor that significantly improved peak power, but not mean power. Heat, added to hypoxia, did not increase cycling anaerobic power. Also, training only in heat did not significantly affect anaerobic power. The inclusion of heat and/or hypoxia in training did not induce negative effects, i.e., a reduction in peak and mean power as measured in the Wingate Test.