ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 37: December 9, 2011 Repeaters in Wiring Uncorrelated Noise Sources 1 Penn ESE370 Fall2011 -- DeHon Previously • Unbuffered wire delay scales as L2 – 0.5 Rwire Cwire – 0.5 L2 Ru Cu • Correlated Noise – Crosstalk – Inductance 2 Penn ESE370 Fall2011 -- DeHon Today • Interconnect Buffering • Uncorrelated Noise sources – Ionizing particles, thermal, shot • Final 3 Penn ESE370 Fall2011 -- DeHon Delay of Wire • • • • Long Wire: 1mm Rwire = 60K W (for the 1mm) Cwire = 0.16 pF (for the 1mm) Driven by inverter – R0 = 25K W – C0 = 0.01 fF – Assume velocity saturated, sized Wp=Wn=1 • Loaded by identical inverter Penn ESE370 Fall2011 -- DeHon 4 Should be able to do these calculations on final. Formulate Delay Delay of inverter driving wire? ( Rbuf Cself Cwire Cload 0.5Rwire Cwire Rwire Cload 5 Penn ESE370 Fall2011 -- DeHon Calculate Delay • Cload = 2 C0 • Rbuf = R0 • Cself = g 2 C0 = 2 C0 ( Rbuf Cself Cwire Cload 0.5Rwire Cwire Rwire Cload 6 Penn ESE370 Fall2011 -- DeHon Should be able to do these calculations on final. Buffer Middle • Delay if add buffer to middle of wire? 7 Penn ESE370 Fall2011 -- DeHon Formulate and Calculate Delay Rwire Cwire Rwire Cwire 2 Rbuf Cself Cload 0.5 Cload 2 2 2 2 8 Penn ESE370 Fall2011 -- DeHon N Buffers • Delay for N buffers? Rwire Cwire Rwire Cwire N Rbuf Cself Cload 0.5 Cload N N N N ( N Rbuf Cself Cload Rbuf Rwire Cwire Cwire 0.5 Rwire Cload N 9 Penn ESE370 Fall2011 -- DeHon Minimize Delay • How minimize delay? • Differentiate & R C wire wire Solve for N: N 0.5 R C C self load buf ( ( N Rbuf Cself Cload Rbuf ( Rwire Cwire Cwire 0.5 Rwire Cload N Rbuf Cself Cload Rbuf Cwire Rwire Cload 2 0.5Rwire Cwire Penn ESE370 Fall2011 -- DeHon Equalizes delay in buffer and wire 10 Calculate: Delay at Optimum Stages for Example • • • • Rwire = 60K W (for the 1mm) Cwire = 0.16 pF (for the 1mm) Rbuf=R0 = 25K W Cself=Cload=2(C0 = 0.01 fF)=0.02fF ( 2 0.5Rwire Cwire Rbuf Cself Cload Rbuf Cwire Rwire Cload 11 Penn ESE370 Fall2011 -- DeHon Segment Length • Rwire = L×Runit • Cwire = L×Cunit Rwire Cwire N 0.5 R C C self load buf ( Ru Cu N L 0.5 R C C self load buf ( 12 Penn ESE370 Fall2011 -- DeHon Optimal Segment Length • Delay scales linearly with distance once optimally buffered * seg L ( R C C L buf self load 2 N Ru Cu Ru Cu N L 0.5 R C C self load buf Penn ESE370 Fall2011 -- DeHon ( 13 Buffer Size? • How big should buffer be? – Rbuf = R0/W – Cload = 2 W C0 (assuming velocity saturation) – Cself = g 2 W C0 ( 2 0.5Rwire Cwire Rbuf Cself Cload Rbuf Cwire Rwire Cload R0 R0 2 0.5Rwire Cwire (1 g 2WC0 Cwire Rwire 2WC0 W W 14 Penn ESE370 Fall2011 -- DeHon Implication W • Rwire = L×Runit • Cwire = L×Cunit • W independent of Length – Depends on technology R0 Cwire W Rwire 2C0 15 Penn ESE370 Fall2011 -- DeHon Delay at Optimum W 2 0.5Rwire Cwire R0 (1 g 2C0 2 R0 Cwire Rwire 2C0 • With g=1, 1+g=2 • Same size as first term 4 R0 Cwire Rwire 2C0 16 Penn ESE370 Fall2011 -- DeHon Ideas • Wire delay linear once buffered • Optimal buffering matches – Buffer delay – Delay on wire between buffers 17 Penn ESE370 Fall2011 -- DeHon Uncorrelated Noise Sources Ionizing Particles Thermal Noise Shot Noise 18 Penn ESE370 Fall2011 -- DeHon Ionizing Particles • Alpha Particles (He nucleus=He2+) • Impact with mega-electron-volts of energy (3—10MeV) 19 Penn ESE370 Fall2010 -- DeHon Ionizing Particles • Alpha Particles (He nucleus=He2+) • Can penetrate microns into Si • Creating 2×106 electron-hole pairs 20 Penn ESE370 Fall2010 -- DeHon Ionizing Particles • Alpha Particles (He nucleus=He2+) • Can be generated by decay in packaging materials – Lead common one including some fraction of radioactive isotopes 210Pb 21 Penn ESE370 Fall2010 -- DeHon Src: http://en.wikipedia.org/wiki/File:Wirebonding2.svg Comparisons • How many electrons in: – Capacitor: 1fF charges to 1V – e = 1.6×10-19 Coulombs • Recall C0 = 0.01fF, – typical load around 10-20C0 • How large a capacitor to withstand loss of 2×106 electrons? 22 Penn ESE370 Fall2010 -- DeHon Disrupt • Alpha particle will disrupt DRAM Cell • Can disrupt undriven nodes – Latch – Dynamic node – Memory bit 23 Penn ESE370 Fall2010 -- DeHon Ionizing Particles • There are other particles with different energies – Neutrons from cosmic rays • 10x energy of alpha particles 24 Penn ESE370 Fall2010 -- DeHon Particle Flux • Differs with location – Altitude • Denver vs. Philadelphia • Ground vs. aircraft at 30,000 feet • Space (outside atmosphere) – Near poles • Changes upset rate seen by chips – By orders of magnitude 25 Penn ESE370 Fall2010 -- DeHon SEU/bit Norm to 130nm Scaling and Error Rates Increasing Error Rates 10 2X bit/latch count increase per generation logic cache arrays 1 180 130 90 65 45 32 Technology (nm) Penn ESE370 Fall2010 -- DeHon Source: Carter/Intel 26 Thermal Noise 27 Penn ESE370 Fall2010 -- DeHon Thermal Background • Except at absolute 0 (Temperature) – Particles are moving around randomly • Thermal bath means free energy around • Electron can be borrow the thermal energy to hop over barrier – Out of an energy well, bond cite – …Out of a capacitor 28 Penn ESE370 Fall2010 -- DeHon Day 8 Doping with P • End up with extra electrons – Donor electrons • Not tightly bound to atom – Low energy to displace – Easy for these electrons to move 29 Penn ESE370 Fall2010 -- DeHon Day 8 Doped Band Gaps • Addition of donor electrons makes more metallic – Easier to conduct Semiconductor 0.045ev 1.1ev Ec ED Ev 30 Penn ESE370 Fall2010 -- DeHon Day 8 Electron Conduction 31 Penn ESE370 Fall2010 -- DeHon Thermal Background • Except at absolute 0 (Temperature) – Particles are moving around randomly • Thermal bath means free energy around • Electron can be borrow the thermal energy to hop over barrier • Are doing it all the time to give us our semiconductors 32 Penn ESE370 Fall2010 -- DeHon Rising above the Thermal Noise • Must apply more energy than background noise to – Hold electron in place – Move an electron from place to place • Charge/discharge a node with some reliability 33 Penn ESE370 Fall2010 -- DeHon Probability of Noise Error • Probability exponential in energy – Not exactly this…but basic dependence • To keep error rate sufficiently low – Need energy of operation (of storage) to be some multiple of kT Perror e E kT 34 Penn ESE370 Fall2010 -- DeHon Where are we today? • How does kT compare to switching 10C0 at 1V? – k=1.4×10-23 J/K – T=300K (Room Temperature) 35 Penn ESE370 Fall2010 -- DeHon Where are we today? • How does kT compare to switching 10C0 at 1V? – k=1.4×10-23 J/K – T=300K (Room Temperature) • kT=4.2×10-21 J • Eswitch=CV2 = 0.1fJ=10-16J • Eswitch~=2×104 kT 36 Penn ESE370 Fall2010 -- DeHon Scaling • What happens as we scale? – 0.5V 4.5nm (versus 1V 45nm) • Eswitch~=2×104 kT • 45nm to 4.5nm impact on Eswitch? – reduce capacitance by 10x – reduce voltage by 2x – Eswitch~=500 kT 37 Penn ESE370 Fall2010 -- DeHon Shot Noise 38 Penn ESE370 Fall2010 -- DeHon Shot Noise • Actual electron transport is probabilistic • Current is a statement about average rate of electron flow • For large numbers of electrons – Law of large numbers convergence to mean s ~= Sqrt(N) – Large N sqrt(N)/N small • Small percentage variation Penn ESE370 Fall2010 -- DeHon 39 Shot Noise • For small number of electrons (N) s ~= Sqrt(N) – sqrt(N)/N not so small – Higher variation – Noise in switching time 40 Penn ESE370 Fall2010 -- DeHon Electron Counts • How many electrons (N) – in 0.1fF, 1V switching event? – in 0.01fF, 0.5V switching event? s? • How many s out to only get 50% of electrons moving? 41 Penn ESE370 Fall2010 -- DeHon Will we see? • Large chips, fast clock rates many events….samples far out on curve From: http://en.wikipedia.org/wiki/File:Standard_deviation_diagram.svg 42 Penn ESE534 Spring2010 -- DeHon Gaussian Distrubution Number Sigma 1 in How many 1 2 3 4 5 6 3.2 22 370 16K 1.7 M 510M 43 Penn ESE370 Fall2010 -- DeHon Idea • Many sources cause upsets – Ionizing particles, thermal, shot noise • Tend to depend on charge – Of node, of switching even • Scaling decreases charge – Lower voltage, lower capacitance – Also increases susceptible nodes 44 Penn ESE370 Fall2010 -- DeHon Final 45 Penn ESE370 Fall2011 -- DeHon Final • Everything – New stuff: (see last year’s final) • • • • Clocking and dynamic logic Memories Crosstalk Transmission lines – Midterm 1 and 2 questions all fair game • Esp: Delay of circuit, optimize delay, Energy • Noise Margins (interaction with noise effects?) 46 Penn ESE370 Fall2011 -- DeHon Admin • Project Due at 11:59pm – 20% per day late penalty • Final – Noon—2pm – Dec. 20 – Moore 212 47 Penn ESE370 Fall2010 -- DeHon
© Copyright 2026 Paperzz