Slide 1Low-Power Design Rules – Anno 1997Adding Leakage to the EquationLow-Power Design Rules – Anno 2007Some Concepts Worth WatchingNovel Switching DevicesExample: Nano-Mechanical RelaysRelay Circuit Design and ComparisonAdiabatic Logic and Energy RecoverySelf-Timed and Asynchronous LogicExploring the Unknown – Alternative Computational ModelsExample: Collaborative NetworksLearning from Sensor Network Concept“Sensor Networks on a Chip”Example: PN code acquisition for CDMABook SummaryInteresting References for Further ContemplationJan M. RabaeyLow Power Design Essentials ©2008Chapter 13SummaryandPerspectivesLow Power Design Essentials ©200813.2 Low-Power Design Rules – Anno 1997Minimize waste (or reduce switching capacitance)–Match computation and architecture–Preserve locality inherent in algorithm–Exploit signal statistics–Energy (performance) on demandVoltage as a design variable–Match voltage and frequency to required performanceMore easily accomplished in application-specific than programmable devices[Ref: J. Rabaey, Intel’97]Low Power Design Essentials ©200813.3 Adding Leakage to the EquationThe emergence of power domainsLeakage not necessarily a bad thing–Optimal designs have high leakage (ELk/ESw ≈ 0.5)Leakage management requires runtime optimization–Activity sets dynamic/static power ratioMemories dominate standby power–Logic blocks should not consume power in standby[Emerged in late 1990s]Low Power Design Essentials ©200813.4 Low-Power Design Rules – Anno 2007Concurrency Galore–Many simple things are way better than one complex oneAlways-Optimal Design–Aware of operational, manufacture and environmental variationsBetter-than-worst-case Design–Go beyond the acceptable and recoupThe Continuation of Voltage Scaling–Descending into ultra-low voltages–How close can we get to the limits?Explore the Unknown[Ref: J. Rabaey, SOC’07]Low Power Design Essentials ©200813.5 Some Concepts Worth WatchingNovel switching devicesAdiabatic logic and energy-recoverySelf-timed and asynchronous designEmbracing non-conventional computational paradigms–Towards massive parallelism?Low Power Design Essentials ©200813.6 Novel Switching DevicesNanotechnology brings promise of broad range of novel devices–Carbon-nanotube transistors, NEMS, spintronics, molecular, quantum, etc–Potential is there – long-term impact unclear–Will most probably need revisiting of logic design technologyOut-of-the-box thinking essentialLow Power Design Essentials ©200813.7 Example: Nano-Mechanical RelaysMinimum energy in CMOS limited by leakage–Even if had a perfect (zero leakage) power gateHow about a nano-scale mechanical switch?–Infinite Roff, low Ron[Ref: H. Kam, UCB’08] SourceDrainLow Power Design Essentials ©200813.8 Relay Circuit Design and Comparison90nm CMOSRelay FA Cell[Ref: E. Alon, UCB’08]Low Power Design Essentials ©200813.9 Adiabatic Logic and Energy Recovery Concept explored in the 1990’s-Proven to be ineffective at that time With voltage scaling getting harder, may become attractive again Example: Resonant AdiabaticMixed-Signal Processor Array forCharge-Based Pattern Recognition [Ref: R. Karakiewicz, JSSC’07] Adiabatic mixed-signal multiply-accumulation (MAC). Charge-coupled MOS pair represents variable capacitive load.Adiabatic logic modeled as transmission gate driving capacitive load from resonant clock© IEEE 2007Low Power Design Essentials ©200813.10 Self-Timed and Asynchronous LogicSynchronicity performs best under deterministic conditions and when duty cycle is highHowever, worst-case model does not fair well when variability is highIn ideal case, self-timed logic operates at “average conditions”delaynumberDelay distribution as a function of variabilityProtocol and signaling overhead of self-timed made it unattractive when delay distributions where narrowThis is not longer the case, especially under ultra low-voltage conditionsEffective “synchronous island” size is shrinkingThe “design flow” argument does not really hold either−Example: Handshake Solutions [Ref: Handshake]Low Power Design Essentials ©200813.11 Exploring the Unknown –Alternative Computational Models•10-15% of terrestrial animal biomass109 Neurons/”node”Since 105 years agoHumans•10-15% of terrestrial animal biomass105 Neurons/”node”Since 108 years agoAntsEasier to make ants than humans“Small, simple, swarm”Easier to make ants than humans“Small, simple, swarm”[Courtesy: D. Petrovic, UCB]Low Power Design Essentials ©200813.12 Example: Collaborative Networks Networks are intrinsically robust → exploit it! Massive ensemble of cheap, unreliable components Network Properties:–Local information exchange → global resiliency–Randomized topology & functionality → fits nano properties–Distributed nature → lacks an “Achilles heel”Bio-inspiredMetcalfe’s Law to the rescue ofMoore’s Law!Low Power Design Essentials ©200813.13 Learning from Sensor Network Concept[Ref: J. Rabaey, I&C’04]Low Power Design Essentials ©200813.14 “Sensor Networks on a Chip” “Large” number of very simple unreliable components provide estimates of result Fusion block combines estimates exploiting the statistics Fusion block only “reliable” componentEstimators need to be independentfor this scheme to be effective[Ref: S. Narayanan, Asilomar’07]Sensor NOC© IEEE 2007Low Power Design Essentials ©200813.15 Example: PN code acquisition for CDMAStatistically similar decomposition of function for distributed sensor-based computation.Robust statistics framework for design of fusion block. Creates better result with power savings of up to 40% for 8 sensors in PN-code acquisition in CDMA systemsNew applications in filtering, ME, DCT, FFT and others[Ref: S. Narayanan, Asilomar’07]Prob (Detection)© IEEE 2007Low Power Design Essentials ©200813.16 Book SummaryEnergy-Efficient one of (if not the most) compelling issues in Integrated Circuit Design today and in the coming decadesThe field has matured substantially–From “getting rid of the fat” and reducing waste–To “truly energy-lean” design technologiesStill plenty of opportunities to move beyond what can be done todayThere is plenty of room at the bottomLow Power Design Essentials ©200813.17 Interesting References for Further
View Full Document