Technion Israel Institute of Technology LOW-LEAKAGE REPEATERS

Technion – Israel Institute of Technology LOW-LEAKAGE REPEATERS FOR NETWORK-ON-CHIP INTERCONNECTS Arkadiy Morgenshtein, Israel Cidon, Avinoam Kolodny, Ran Ginosar QNo. C Research Group Electrical Engineering Department Technion – Israel Institute of Technology Haifa, Israel

Highlights • Leakage in No. C links with repeaters • Selecting the Repeater Type • Optimizing Repeater Insertion • Utilization-Oriented Analysis 2 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Networks-on-Chip (No. C) No. C characteristics • Packet-based data routing • Multiple Quality-of-Service levels Physical layer of No. C • Low link utilization Most links idle most of the time! Ø Leakage power is important 3 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Leakage Reduction in Logic Subthreshold leakage is dominant at high temperatures Solutions: • Dual Threshold • Sleep Transistors • more… 4 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Leakage Reduction in Repeaters Solutions: large sizes very high wire loads no transistor stack unique characteristics 5 Low-Leakage Repeaters for No. C Communications ? specific solutions needed ISCAS 2005

Existing Repeater Types LVT – Low-Vt Repeaters • fast • high leakage HVT – High-Vt Repeaters • slow • low leakage SVT - Staggered-Vt • fast (In 0 1) • slow (In 1 0) • low leakage (idle) [16] Sylvester et al. 6 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Research Outline Network-on-Chip Low & Varying Utilization Selecting the Repeater Type SR – Sleep Repeaters DTD – Dual-Vt Domino Repeaters & Optimizing Repeater Insertion Utilization-Dependant Optimal Number of Repeaters Utilization-Oriented Analysis 7 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Dual-Threshold Domino (DTD) Repeaters High-Vt Evaluation Transistors Low-Vt Pre-charge Transistors synchronized Clk link 8 Low-Leakage Repeaters for No. C Communications ISCAS 2005

DTD Repeaters Operation X X X X • Precharge transistors disconnected • CLK line is synchronized with Data • Evaluation by HVT transistors – slower but tolerant to Vt fluctuations • Each Evaluation transistor drives only one transistor at next stage – faster and can be down-sized 9 Low-Leakage Repeaters for No. C Communications ISCAS 2005

DTD Repeaters Operation X X X X • Evaluation transistors disconnected • Precharge to low-leakage mode • Precharge by LVT transistors - fast 10 Low-Leakage Repeaters for No. C Communications ISCAS 2005

DTD Repeaters Operation X 1 0 X X X 11 X 0 X 1 0 1 X X 1 0 0 • HVT transistors are “off” – low leakage 1 Low-Leakage Repeaters for No. C Communications ISCAS 2005

DTD Repeaters Operation X X 1 12 0 X X X 0 X 1 X 0 • For Data=‘ 0’ - no transition occurs Low-Leakage Repeaters for No. C Communications ISCAS 2005

DTD Repeaters Operation X X X 13 X X Low-Leakage Repeaters for No. C Communications ISCAS 2005

DTD Highlights ü Application of domino and double-Vt techniques to low-leakage repeaters Benefits + Effective leakage reduction during standby + Reduced load on each repeater allowing downscaling and area reduction + Tolerance to VT fluctuations by using HVT evaluation transistors Drawbacks - Increased dynamic power consumption due to signaling in domino protocol - Overhead of clock line and pre-charge wiring 14 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Sleep Transistors in Repeaters Logic MTCMOS Evolution Repeaters 15 SR Low-Leakage Repeaters for No. C Communications ISCAS 2005

MTCMOS in Repeaters • Common sleep transistors insertion + Both NMOS and PMOS are used - All stages enter and exit “sleep” mode simultaneously - LARGE sleep transistors - High routing complexity and wiring overhead 16 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Repeaters with Per-Stage Sleep Transistor • Distributed sleep transistors along the link • Each stage of repeaters has a separate pair of sleep transistors + One stage of repeaters is active - others are in low-leakage standby - Sleep Transistor is heavily loaded and has to be scaled with link width 17 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Repeaters with Per-Stage Sleep Transistor • Distributed sleep transistors along the link • Each stage of repeaters has a separate pair of sleep transistors + One stage of repeaters is active - others are in low-leakage standby - Sleep Transistor is heavily loaded and has to be scaled with link width active 18 sleep Low-Leakage Repeaters for No. C Communications ISCAS 2005

Repeaters with Per-Stage Sleep Transistor • Distributed sleep transistors along the link • Each stage of repeaters has a separate pair of sleep transistors + One stage of repeaters is active - others are in low-leakage standby - Sleep Transistor is heavily loaded and has to be scaled with link width sleep 19 active sleep Low-Leakage Repeaters for No. C Communications ISCAS 2005

Repeaters with Per-Stage Sleep Transistor • Distributed sleep transistors along the link • Each stage of repeaters has a separate pair of sleep transistors + One stage of repeaters is active - others are in low-leakage standby - Sleep Transistor is heavily loaded and has to be scaled with link width sleep 20 sleep active Low-Leakage Repeaters for No. C Communications ISCAS 2005

SR – Sleep Repeaters Parallel link using individual zigzag sleep transistors: • One sleep transistor per repeater 1 0 0 + Smaller sleep transistors + Simpler routing 0 1 0 • Zigzag connection: • Only to transistors that are off during sleep + Number of sleep transistors is reduced by 50% 21 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Sleep Repeater Highlights ü Novel: Efficient sleep transistors for repeaters Benefits + Effective leakage reduction during standby + Optimized structure according to specifics of repeater insertion Drawbacks - Area overhead - Increased dynamic power consumption due to additional transistors 22 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Simulation Setup • 65 nm BPTM models for transistors and interconnect • 32 -bit link operating at 105°C temperature • LVT design was used as baseline for repeater insertion: Scaling factor was adjusted for SVT, DTD and SR to meet the delay target equal to LVT • Area, delay and energy were obtained for each of the compared techniques 23 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Total Repeater Area + DTD smallest area - SR largest area 24 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Energy vs. Utilization • SVT: Least energy at high utilization + SR: Least energy at low utilization 8 mm link 25 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Utilization-Dependant Optimal Number of Repeaters Set Target Delay (D

Optimal Number of Repeaters example • For each k a suitable sizing factor h is found to meet the target delay • Optimal k for minimal leakage is kleak=4 kleak kdyn • Optimal k for minimal dynamic power is kdyn=6 Power vs. k for target D=309 ps (instead of Dmin=280 ps), L=10 mm 27 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Number of Repeaters vs. Utilization example • Total power as function of utilization for Kdyn and Kleak Prefer Kleak • Power ratio is calculated for Kdyn and Kleak + Break-even point is at 40% utilization Prefer Kdyn + The results of Kleak are up-to 17% better at low utilization rates Power ratio of Kdyn vs. Kleak 28 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Summary ü DTD (Dynamic Dual-Threshold) Repeaters ü SR (Sleep Repeaters) • Zig-zag structure ü SR least power at low utilization • Thanks to low leakage ü Optimal number of repeaters depends on link utilization 29 Low-Leakage Repeaters for No. C Communications ISCAS 2005

Questions? 30 Low-Leakage Repeaters for No. C Communications ISCAS 2005