Tunable Sensors for Process-Aware Voltage Scaling Tuck-Boon Chan‡ and Andrew B. Kahng†‡ CSE† and ECE‡ Departments, UCSD tbchan@ucsd.edu, abk@cs.ucsd.edu http://vlsicad.ucsd.edu 1 Outline • • • • • Intro: Adaptive Voltage Scaling (AVS) Overview of Proposed Method Voltage Scaling Properties Designing the Circuit Results 2 Adaptive Voltage Scaling Maximum frequency reduce voltage meet performance with less power a typical chip margin worst-case scenario (e.g, due to process variation) f target Voltage Vnominal • Circuits are designed to guardband for performance variation • There is margin for typical chips • Adaptive voltage scaling (AVS) adjusts voltage to reduce power 3 Taxonomy of AVS Techniques AVS classes Power Open-Loop AVS approaches Freq. & Vdd LUT AVS Pre-characterize LUT [Martin02] Post-silicon characterization Process-aware AVS Post-silicon characterization [Tschanz03] Generic monitor ClosedLoop AVS Error Tolerance AVS Design dependent replica Process and temperature-aware AVS Generic on-chip monitor [Burd00] Design-dependent monitor [Elgebaly07, Drake08, Chan12] In-situ monitor In-situ performance monitor Measure actual critical paths [Hartman06, Fick10] Error Detection System Error detection and correction system Vdd scaling until error occurs [Das06,Tschanz10] 4 Motivation for Closed-Loop AVS [Hartman06] • Closed-loop AVS saves up to 62% dynamic power 5 Classes of Closed-Loop AVS ClosedLoop AVS Generic monitor Design-dependent replica • Does not capture design-specific performance variation In-situ monitor • Critical path may be difficult to identify (IP from 3rd party) • Calibrating monitors at multiple modes/voltages requires long test time This work: Tunable monitor for closed-loop AVS • Can be applied as a generic monitor • Or tuned to capture design-specific performance 6 Outline • • • • • Intro: Adaptive Voltage Scaling (AVS) Overview of Proposed Method Voltage Scaling Properties Designing the Circuit Results 7 Voltage Scaling Key Concepts Max. freq. f target Process distance k SS f Scaling rate = V Vmin_path (k ) Vnom Voltage • Process distance: process-induced frequency shift relative to target frequency • Scaling rate: frequency shift (f) per unit voltage difference (V) • Vmin= Minimum Vdd to meet target frequency • Calculated from process distance and scaling rate 8 Monitor Design Concept • Use Vmin of ring-oscillator (RO) as a reference • Design ROs with worst-case voltage scaling properties an arbitrary circuit will meet target frequency at Vmin_ro Process corner A Critical paths Freq. Process corner B Critical paths RO Freq. RO f target f target V V Max. Vmin of ROs > Max. Vmin of paths 9 Proposed Method: Tunable Monitor Scenario 1: Without circuit information • Configure RO for worst-case Vmin • Guardband for arbitrary circuits Scenario 2: With chips at process corners • Extract Fmax and Vmin of chips • Tune voltage scaling properties of ROs so that Vmin_ro > Vmin_chip • Recover margin with one calibration Store config. • Our focus is on voltage scaling property analyze worst-case voltage scaling 10 Problems • Goal: Vmin_ro > Vmin_path • Questions: Vmin of arbitrary Given a process critical paths technology, what is the freq. range of the Vmin that is Path A Path B Path C defined by process distance and scaling rate f target for arbitrary critical paths? V What circuit techniques Vmin = ? can “tune” Vmin? Also, V changes at min different process corners 11 Outline • • • • • Intro: Adaptive Voltage Scaling (AVS) Overview of Proposed Method Voltage Scaling Properties Designing the Circuit Results 12 Vmin Analytical Derivation (1) Vmin proce ssdistance Vnom scalingrate f path (k ,Vnom ) ftarget Vnom Scaling f path (k ,Vnom Process V ) f path (k ,Vnom ) rate distance (2) fpath = inverse of average delays of NMOS & PMOS 2 f path (k ,Vnom ) Dnmos (k ,Vnom ) Dpmos (k ,Vnom ) (3) Calculate delays with • Elmore delay model • Effective currents of transistors 13 Normalized Vmin Vmin Sensitivity 1.020 1.010 1.000 0.990 0.980 0.970 fanout Ron NMOS lwire length Ron PMOS beta 0.2 0.6 1.0 1.4 1.8 2.2 2.6 3.0 3.4 3.8 Normalized value of circuit parameters Vmin for PMOS only Vmin for NMOS only • Vmin is not very sensitive to fanout, interconnect load, etc. • Empirically, bounds on Vmin determined by NMOS and PMOS 14 Effects of Fanout and Series Resistance Vmin (V) SS TT FF SF FS 1.00 0.90 0.80 0.70 0.60 FO1 SS FO2 FO4 FO8 Fanout TT FF SF FS Vmin (V) 1.00 0.90 0.80 0.70 0.60 1 • Fanout has little effect on Vmin High series resistance reduces Vmin But, need long wires 100 200 400 800 1600 Series resistance (ohm) 15 Effects of Cell Type SS TT FF SF FS Vmin(V) 1.10 1.00 0.90 0.80 0.70 0.60 0.50 Cell type • Cell type affects Vmin • Maximum Vmin at different corners are determined by different cell types • Stacking causes cell delay biased to PMOS or NMOS changes device characteristics and Vmin 16 Effects of Cell Strength 1.10 1.00 0.90 0.80 0.70 0.60 0.50 TT FF SF FS SS TT FF SF FS 1.10 1.00 0.90 0.80 0.70 0.60 0.50 Vmin (V) Vmin (V) SS X1 layout X2 X3 Cell changes device X0 X1 X2 X3 NAND3 cell strength INV. cell strength characteristics and Vmin X0 • Vmin does not increase from X1 to X3 • But increases from X0 to X1 • X1 to X3 {1,2,3} fingers, same device characteristic • X0 to X1 Both 1 finger but different diffusion area 17 Outline • • • • • Intro: Adaptive Voltage Scaling (AVS) Overview of Proposed Method Voltage Scaling Properties Designing the Circuit Results 18 Design of RO with Tunable Vmin • Identified two circuit knobs to tune Vmin • Series resistance • Cell types (INV, NAND, NOR) • Proposed circuit • ROs with different cell types (worst-case Vmin are determined by different cells at different process corners) • Tune Vmin a configurable series resistance at each stage Control pins 1 bit 1 bit 1 bit High resistance Low resistance 19 Tunability Vmin(V) • Vmin decreases linearly with % high-resistance passgates • ROs with different gate types have similar trend 1.100 1.000 0.900 0.800 0.700 0.600 0.500 INVX3 SS TT FF SF FS 0% 12% 24% 36% 48% 61% 73% 85% 100% High resistance passgates (%) 20 Outline • • • • • Intro: Adaptive Voltage Scaling (AVS) Overview of Proposed Method Voltage Scaling Properties Designing the Circuit Results 21 Experiment Methodology • Goal: Validate PVS ROs in simulation • Check Vmin of ROs vs. Vmin of paths • with arbitrary circuits and process variation • Experiment setup: • 65nm industrial technology • Implement 3 testcases (arbitrary circuits) • Implement 3 tunable ROs (INV, NAND, NOR) Power (mW) Area (mm2) Freq. target FPU 4.1 0.015 710 TLU 438.0 0.098 507 MUL 19.8 0.050 1042 22 Process Variation Setup • Simulate critical paths and ROs with SPICE 200 Monte Carlo samples (global variation) • 4 variation sources, Gaussian distributions Variation sources NMOS Vth PMOS Vth Channel length Gate oxide thickness mean 0 +/- 3 sigma 30mV 0 0 0 30mV 5nm 0.06nm • Difference between slow and fast corners define +/- 3 sigma values of variation sources 23 Vmin Extraction and Comparison • Define ftarget of chip and ROs at “slow-slow” process corner nominal voltage = 1.0V • Vmin_chip = max. Vmin of critical paths of a testcase • Vmin_est = max. Vmin of 3 ROs • For each testcase, calculate Vmin_est - Vmin_chip of every Monte Carlo sample • A chip is safe when Vmin_est - Vmin_chip > 0 24 Scenario 1: Guardband for Arbitrary Circuit • Vmin_est - Vmin_chip > 0 under process variation • Similar results for different testcases • Small difference between normal and tunable ROs due to series passgates TLU testcase FPU testcase MUL testcase 25 Scenario 2: Tune ROs for Margin Reduction • Extract Vmin_chip at different process corners • Configure % high-resistance passgates min. : {V min_est (k ) Vmin_chip (k )} k s.t. : Vmin_est (k ) max {Vmin_ro (k , i, c)} i Vmin_est (k ) Vmin_chip (k ),k Ensures Vmin_est guided by ROs is always safe 26 Experiment Result on Tunability Aggressive config. Vmin_est < Vmin_chip Some chips will fail Optimized config. • Increase % high resistance passgates • Vmin_est ≈ Vmin_chip Default config. • Low resistance passgates • Guardband for worst-case • Vmin_est > Vmin_chip • 13mV margin 27 Experiment Result on Tunability Aggressive config. Vmin_est < Vmin_chip Some chips will fail Optimized config. • Increase % high resistance passgates • Vmin_est ≈ Vmin_chip Default config. • Low resistance passgates • Guardband for worst-case • Vmin_est > Vmin_chip • 13mV margin Benefits of tunability • Recover voltage margin • Compensate for difference between SPICE model vs. silicon • Recover margin when chip performance variation is reduced due to improvements in chip manufacturing 28 Summary • Monitor design based on voltage scaling properties • Estimate the worst-case voltage scaling property across different process corners Does not require information about critical paths Can be used as an IP for arbitrary circuits • Recover margin if fmax of sample silicon is available • Future works Proof of concept silicon Account for performance variation due to layout context 29 Thank you! tbchan@ucsd.edu, abk@cs.ucsd.edu http://vlsicad.ucsd.edu 30 Backup Slides 31 Effects of Pass Gates • Pass gate is equivalent to large resistance • Vmin decreases with fewer parallel pass gates 32 Effects of Cell Type and Strength • Key observations: Vmin is affected by cell types Use NAND, NOR type ROs • Cell strength changes Vmin Use cells with large Vmin 33