Title: TRIO2 Interconnect Modeling and Synthesis for Signal Integrity
1TRIO2Interconnect Modeling and Synthesis for
Signal Integrity
- Project Leader Prof. Lei He, LHE_at_EE.UCLA.EDU
- Authors Jun Chen, Kevin Lepak, Min Xu
- UCLA Electrical Engineering Department
- Los Angeles, CA 90095
- (Tools and references available at
http//eda.ee.ucla.edu)
2Interconnect Design Closure ?Interconnect
Synthesis Global Routing
- Device locations and constraints
- Delay
- Power
- Signal integrity
- Skew
- ...
- Interconnect synthesis should be integrated with
global routing for design closure
3TRIO PackageInterconnect Synthesis under RC
model
- Technology advances lead to the need for
interconnect-driven design - Interconnect optimization techniques for
performance - Topology optimization
- Buffer (repeater) insertion
- Device sizing, wire sizing and spacing
- TRIO Tree, Repeater, and Interconnect
Optimization - Reduce interconnect delay by up to 7x
- Installed at Intel, IBM, HP, Fujitsu, and
TRIO Performance Optimization under RC model
4Is TRIO Sufficient for DSM Design?
- Coupling capacitance becomes the dominant
capacitance component - Inductive coupling gains more significance
- Interconnect impedance is more than resistance
- Z ? R j?L where ? ? 1/tr
- On-chip inductance should be considered
- when ?L ? R due to smaller and quicker devices
- Coupling capacitance and inductance affect both
delay and signal integrity - Signal integrity is one (or THE) upcoming barrier
of design closure
5Impact of Coupling Capacitance
- Coupling capacitance introduces significant delay
variations, especially for local and intermediate
interconnects
6Resistance vs Inductance
Length 2000, Width 0.8 Thickness 2.0,
Space 0.8
Ls and Lx for two parallel wires
R and ?L for a single wire
- Reactance due to inductance becomes comparable to
resistance for global interconnects
7Impact of Inductance
no shielding
coplanar wave guide for inductance reduction
- Inductance introduces significant noise,
especially for global interconnects
8TRIO2
- TRIO2 integrated toolset for interconnect
modeling, analysis and synthesis for signal
integrity under RC and RLC models - RC and RLC extraction
- Table-based RC model (with process variations)
DAC97, CICC99 - Table-based R and L model CICC99, DATE00,
GVLSI01 - RC and RLC circuit model generation
- 3D full model based on PEEC and VPEC submission
- 2D cascade model and normalized model
DATE00GVLSI01 - Worst-case noise/delay model
- Keff model, and Length scaled Keff model
ISPD00, ICCAD02 - Five-pole model, Decoupling model, WCN model, and
t-line model DAC01, GVLSI02, ASP-DAC03 - Interconnect synthesis
- Shield insertion for buses and clocks CICC99
- Simultaneous shield insertion and net ordering
ISPD00, DAC01 - Power net and shielding co-design ICCAD01
- Clock synthesis (buffering and shielding to
control noise and rising time)
9SINO ISPD00, DAC00, ICCAD01Simultaneous
Shield Insertion and Net Ordering
- Given A set of signal nets, sensitivity, and
noise bound - Assume Shields can be assigned to arbitrary
tracks or uniform shielding with minimal
perturbation - Find An SINO solution such that
- Capacitive noise free
- No sensitive wires are adjacent
- No neighboring wires switch simultaneously
- Inductive noise less than the given noise bound
- Minimal area
10SINO Experiment
Noise Bound Average number of shields Average number of shields Average number of shields Average number of shields
Noise Bound NOSI SINO NOSI SINO
Noise Bound Sensitivity Rate 30 Sensitivity Rate 30 Sensitivity Rate 60 Sensitivity Rate 60
0.15V 7.0 5.4 (-23) 10.8 7.85 (-30)
0.20V 6.15 5.05( -14) 10.6 7.45 ( -30)
0.25V 6.00 4.2(-30) 9.75 7.4(-33)
- SINO achieves up to 33 shield reduction compared
to best alternative (NOSI) - 32-bit signal bus, 2000 um long wire, in 70nm
technology
11PWL Model for CPW
- Piece-Wise Linear model computes the far end
response for a CPW with capacitive loading and
ramp input - CPW coplanar wave guide
Transform CPW to transmission line without
loading by moment matching
Construct the waveform for step input of
transmission line by 3-piece linear approximation
for each round trip of waveform
Construct the waveform for ramp input by over
3-piece approximation per round trip
12PWL versus SPICE
setting runtime runtime runtime runtime
model SPICE PWL work3 work4
1 88.1 0.01 0.01 0.18
2 148.1 0.01 0.01 0.18
3 368.23 0.01 0.01 0.12
4 23.23 0.01 0.01 0.73
5 121.39 0.01 0.01 0.2
6 344.7 0.01 0.01 0.02
- PWL obtains highly accurate waveform in 1000x
less time - 4 and 5 (Meindl00 and Eo02) have much
bigger errors
13CPW Synthesis for Buffered Clock Tree
a x1 w1 s1 g1 x2 w2 s2 g2 power
0 237 1.5 4.4 8.2 149 1.1 5.0 10.7 1769ff
0.3 288 1.7 2.9 3.9 166 1.2 2.4 6.3 1899ff
1 500 4.8 3.2 2.1 324 1.4 1.5 2.1 2952ff
- Tradeoff between area and power
- Primarily decided by buffer size
- Min-power solution costs 3x area than min-area
solution, but 2x less power - Desired solution with a0.3
- Save 42 area with 4 more power compared to
min-power solution - a weighting factor to decide area/power
tradeoff
14Overall Design Closure for Signal Integrity ?
TRIO2 Gsino VPEC
- Iterative ECO flow
- Circuit modeling and analysis of critical nets
(TRIO2) - Correction by net-based TRIO2/Gsino
- One-pass correctness by construction
- Chip-level Gsino w/ embedded TRIO2 and VPEC
- Gsino global routing with simultaneous shielding
and net ordering - VPEC vector potential equivalent circuit model,
provably passive and 1000x faster than PEEC