Chip-Package Co-design - University of California, Los Angeles

Download Report

Transcript Chip-Package Co-design - University of California, Los Angeles

Package-Chip Co-Design
Prof. Lei He
Electrical Engineering Department, UCLA
[email protected]
http://eda.ee.ucla.edu
ASPDAC 06
Lei HE, UCLA
1
Outline
• Overview of Chip Package Co-design
• IO planning and placement
• Power integrity in package
• Reading: package tutorial at ee201c wiki
ASPDAC 06
Lei HE, UCLA
2
Wire-bond vs Flip-chip
• Wire bonding
– Cheap Implementation
– Difficult to design
– IO signals are at
boundary
– High inductance (~1nH)
– More worry on core and
IO power distribution
during design and
analysis
ASPDAC 06
Lei HE, UCLA
3
Wire-bond vs Flip-chip
• Flip-chip
– IO cells can be over entire
of chip area
– Low inductance (~0.1nH)
– High pin count, high cost
– Less worry on power
delivery
ASPDAC 06
Lei HE, UCLA
4
Silicon
Package
Board
(Cadence)
ASPDAC 06
Lei HE, UCLA
5
Connection from die to board
• Die (IO cells -> RTL routing -> bumps)
• -> package (bumps -> escape routing -> package
routing -> balls)
• -> board
ASPDAC 06
Lei HE, UCLA
6
VLSI-Centric Design (Problematic)
• IC and package tools very separated:
IC Physical Design
Package Physical Design
I/O Locations
IBIS Models
Package Modeling/Simulation
IC Modeling/Simulation
(From P. Franzon)
ASPDAC 06
Lei HE, UCLA
7
On-Chip Design Concerns
 Physical Concerns
 Die Netlist Connectivity (logic cells to IO cells)
 System Connectivity (IO cells to package)
 Power Network (power planes to power IO cells to logic cells)
 Electrical Concerns




Core Timing Constraints
System Timing Constraints
Power Budget
Signal Integrity and Reliability Constraints
 Supply voltage scaling imposes very tight noise margins on chip
and package designs
 Significant noise contribution from core switching
 But greater on-chip exposure to package-side (i.e., IO) SSN
ASPDAC 06
Lei HE, UCLA
8
Package Design Concerns
• Physical Concerns
 Reduce stack-up layers and package cost
 Place IO cells and decoupling capacitance
 Complete escaping and package routing
• Electrical Concerns
 Reduce SSN noise
 Lower impedance of power distribution
 Meet timing constraints, especially for bus and
differential pairs
ASPDAC 06
Lei HE, UCLA
9
Typical Package Design Cycle
Pad/package Iteration: P&R of IO/Pad cells,
Pins; Pwr/gnd and inter-cell connections;
PCB pin locations (x,y);
Floorplanning of IO/Pad/Pins;
Define Netlist hierarchy/manipulations
Manufacturing and NRE Costs;
Die, Substrate, Package
Package/Pad/IO Rule checking (PRC):
SI, timing, clocks, IO voltages,
assembly rules, special regions
Verify user specified requirements and rules;
PCB pins, Power grid, # VSS/VDD,
decoupling caps, EMI, ESD, Vias
Defining Interfaces, Signals,
PLL, Power, Clock, # pins, # IOs
Package/Substrate Architecture Exploration
(start ~4/5 months before Tapeout)
ASPDAC 06
Finalize IOs/Pads/Pins;
Package Tapeout
Lei HE, UCLA
10
Needs of Chip-Package Co-Design
•
High system frequency
–
–
400 MHz buses becoming common
On-chip exposure to package noise
• Simultaneous switching noise
• Package resonance
•
High density packaging and high pin count
• Difficult to layout and escape-route
• Again, more SSN for on-die circuit
•
Tight time to market
–
ASPDAC 06
Convergence of package and IO becomes a bottleneck if chip
and package handled by separated flows
Lei HE, UCLA
11
Keys Problems to Solve
• Chip and package co-extraction and co-simulation
– Difficult to obtain accuracy for sign-off
– More difficult to achieve efficiency with accuracy or fidelity for
planning and design
– Challenging to handle mutual inductance and large number of
ports
• Co-design focuses on important links between chip and
package
– Chip side: IO buffer design, noise isolation circuitry, P/G
network, IO pad macro-placement, RDL estimation,
– Package side: Package stack-up, P/G plane design, macroplacement of balls and pins, and estimation of escape routing
– key issues:
• IO planning and placement, power delivery system
ASPDAC 06
Lei HE, UCLA
12
Outline
• Overview of Chip Package Co-design
• IO planning and placement
– Design constraints
– Multi-stage solutions
• Power integrity in package
ASPDAC 06
Lei HE, UCLA
13
Design Constraints
for IO Planning and Placement
•
•
•
•
Power integrity
Timing
I/O standards
Core and board floorplanning
ASPDAC 06
Lei HE, UCLA
14
Power Integrity Constraints
• Power domain constraint
– I/O cell voltage specification
– Cells from same domain prefer physically closer
• Minimize power plane cut lines in the package
– Provide proper power reference plane for traces
– Depend on physical locations of I/O cells
• Proper signal-power-ground (SPG) ratio
– Primary and secondary P/G driver cells
– Minimize voltage drop and Ldi/dt noise
ASPDAC 06
Lei HE, UCLA
15
Timing Constraints
• Substrate routes in package varies significantly
– Length spans from 1mm to 21mm
– Timing varies more than 70ps for SSTL_2
• I/O cells with critical timing constraints shall take
this into account
– Differential pairs and bus prefer to escape in parallel
and in same layer
ASPDAC 06
Lei HE, UCLA
16
I/O Standard Related Constraints
• High-speed design  high-speed I/O
• I/O standard requirements
– Relative timing requirements on signals
– Likely to be connected to the same interface at other
chips, so prefer to keep relative order to ease routing
• Closeness constraint
– Less process variation
• Bump assignment feasibility constraint
ASPDAC 06
Lei HE, UCLA
17
Floorplan Induced Region Constraints
• Top-down design flow
– PCB floorplan
• Bottom-up design
– Chip floorplan
• I/O cells have region preference
– Which side?
– What location?
ASPDAC 06
Lei HE, UCLA
18
Connection from die to board
• Die (IO cells -> RTL routing -> bumps)
• -> package (bumps -> escape routing -> package
routing -> balls)
• -> board
ASPDAC 06
Lei HE, UCLA
19
Flow of IO planning and placement
I/O Planning
Global I/O and
core co-placement
(1) Wire length driven
(2) Even distribution of I/O
and core
(3) Power domain floorplan
Bump array
placement
(1) Escapability
(2) Planar routability
I/O site array
definition
(1) RDL routability
Detailed I/O
placement
(1) Constraint-driven
(2) Legalization
Chip finishing
(e.g., RDL routing)
ASPDAC 06
• Global I/O and Core
co-placement
• Bump array Placement
• I/O site definition
Package finishing
(e.g., substrate routing)
• Constraint driven
detailed I/O placement
Lei HE, UCLA
20
Global I/O and Core Co-placement
Domain Domain
(2.5v)
(1.8v)
• Minimize both wire
length and power
domain slicing
• Power domain plans
I/O cells location, and
becomes region
constraints for I/O cells
for the following steps
Domain(3.3v)
ASPDAC 06
Lei HE, UCLA
21
Bump and Site Definition
• Regular bump pattern is
preferred
Domain Domain
(2.5v)
(1.8v)
Bump
Super site
I/O site
Region
Constraint
Domain(3.3v)
ASPDAC 06
– Escapability analysis
• Regular I/O site is
preferred
– I/O proximity
– RDL planar routability
analysis
• I/O sites more than I/O
cells
– SPG ratio consideration
– Flexibility for later bump
assignment
• I/O super site: a cluster of
I/O sites
Lei HE, UCLA
22
Assign I/O Cells to Super I/O Sites
• A set of region constraints (Ri, CiR)
– A rectangular restricted area Ri for I/O cells CiR
– E.g., floorplan, power domain definition, wire length minimization
• A set of clustering constraints (Li, CiL)
– The spread of I/O cells should be less than a bound
– E.g., I/O standard const., floorplan, timing
• A set of differential pair constraints
– Different pairs should be connected to bumps with similar
characteristics
– E.g., timing
• Solve by ILP or LP followed by netflow-based
legalization
– Paper 2C-4, Wednesday afternoon at this conference
ASPDAC 06
Lei HE, UCLA
23
Experiment Setting
• Real industrial designs
• Constraints not include the ones that are
generated internally
ASPDAC 06
Lei HE, UCLA
24
Experiment Result
• Obtain 100% CSR (constraint satisfaction ratio)
in short runtime
ASPDAC 06
Lei HE, UCLA
25
Power Plane Cuts
Core
Domain
Plane
Cut
Island
IO
Domain
ASPDAC 06
Lei HE, UCLA
26
Power Domain Routing
Domain
Routing
ASPDAC 06
Lei HE, UCLA
27
Outline
• Chip Package Co-design Flow
• IO planning and placement
• Power integrity in package
– Overview and modeling
– Decap insertion
• Impedance based
• Noise-based
ASPDAC 06
Lei HE, UCLA
28
Power Integrity
• Time domain
• Frequency
power and
domain analysis
signal integrity
of Power Planes
• Signal Noise Analysis
Impedance
• Return Path Modelling
for EMI and SSN
analysis
• EMI Analysis
• Package Plane
Resonance
ASPDAC 06
coupled with power
plane models
• Superposition of
Power Noise on Signal
Noise
• IBIS, SPICE and PEEC
models are employed
Lei HE, UCLA
29
PDS: Power Distribution System
Detailed Network Modeling is needed for
accurate analysis of Core and IO Power
ASPDAC 06
Lei HE, UCLA
30
Ideal Package Power Planes
Early Package Design Exploration
Planes have no holes or perforations
Perfect Microstrip or Stripline Patterns
Impedance is well conditioned
ASPDAC 06
Lei HE, UCLA
31
Non-ideal Package Power Planes
Detailed Plane Modeling
Planes are split for different voltage domains
Planes could have any number of holes / perforations
Microstrip or Stripline Patterns: imperfect
ASPDAC 06
Lei HE, UCLA
32
PDS Modelling
• Wire capacitance can be extraction using 2.5D model [Heet al, DAC’97]
– With extension to arbitrary routing angle
• Plane capacitance needs to consider impact of wires in
between
• Inductance is must and can be formula based
– Bonding wires have well controlled shapes
• Susceptance (L-1) makes sparsification easier
• But sign-off often needs 3D field solver
ASPDAC 06
Lei HE, UCLA
33
PDS Design
• Assign power planes in package stackup
• Assign power domains: V18, V25, Vanalog,…
• Decide via stapling
• Improve power delivery
• Reduce current loop and eliminate noise
• Assign P/G balls
ASPDAC 06
Lei HE, UCLA
34
PDS Concerns
• DC Concerns
 On-Chip IR Drop
 Not a big concern in Flip-chip Designs
 In-Package IR Drop
 Important but still very small
 In-PCB IR Drop
 Can be ignored
• AC Concerns
 Low impedance Network across a broad frequency spectrum
 Reduce inductive effective to reduce SSN
 Control Chip/Package resonance
ASPDAC 06
Lei HE, UCLA
35
Power Plane Noise (AC vs DC)
ASPDAC 06
Lei HE, UCLA
36
PDS Design
•PDS Impedance
•Smaller Zo  larger current
0.05  Vdd
Zo 
I transient
•PDS Bandwidth
•Maintain Zo from 0 to fmax
•Decide on Decap Allocation
•High speed drivers draw current from nearby decoupling
capacitors
•Decoupling capacitors reduce the size of the current loop
ASPDAC 06
Lei HE, UCLA
37
Chip-Package Plane Resonance
Resonances are produced due to inductance and capacitance
Z
Capacitor becomes inductive
beyond its self resonant
frequency, f(SR)
f SR 
1
2 LESLC
frequency
Resonant frequency is
f max
1

2 2 L pkg C pkg
Need a set of capacitors to cover small,
medium, and high frequency ranges
ASPDAC 06
Lei HE, UCLA
38
Decoupling capacitors optimization
• Needs for power integrity
– Reduce resonance.
– Reduce effective inductance and resistance.
• Different levels of decoupling capacitors
– Board, package, chip
– Different effective frequency range.
• Decoupling capacitors is not perfect capacitor
– ESL
– ESR
– Lower ESL and ESR, higher cost
• Designing of decoupling capacitors needs to determine
– Values
– Location
– Decoupling capacitor type
ASPDAC 06
Lei HE, UCLA
39
Impact of decoupling capacitors
ASPDAC 06
Lei HE, UCLA
40
Existing Solutions
• Manual trial-and-error approaches
– [Chen et al., ECTC ’96]
– [Yang et al., EPEP 2002]
• Automatic optimization
– [Kamo et al., EPEP 2000], [Hattori et al., EPEP 2002]
• Ignore ESL and ESR.
– [Zheng et al., CICC 2003]
• Use impedance as noise metric
– [Chen et al., ISPD 2006]
• Noise driven decap insertion
ASPDAC 06
Lei HE, UCLA
41
Limit of Impedance Metric
• Can not capture noise accurately
• Will Lead to large over-design
ASPDAC 06
Lei HE, UCLA
42
Incremental impedance computation
• When adding one decoupling capacitor Zd at port k
– the new impedance from port j to port i is
Zij  Zij 
Zik Z kj
Z kk  Z d
• When removing one decoupling capacitor Zd at port k
– the new impedance from port j to port i is
Zij  Zij 
ASPDAC 06
Zik Z kj
Z kk  Z d
Lei HE, UCLA
43
Time complexity
• With one or a few decoupling capacitors inserted
– O(np2): np is the number of ports
– Existing work: O(np3)
• Especially suitable for trial-and-error or iterative
methods
– Only a few decoupling capacitors changed in each
iteration
– Able to compute only impedance or I/O ports before
updating rest ports
ASPDAC 06
Lei HE, UCLA
44
Noise Calculation
• FFT methods
– Frequency components of noise from port j to port i
Vij ( fk )  Zij ( fk )  I j ( fk )
• Worst case noise from all ports
– Superposition
ASPDAC 06
Lei HE, UCLA
45
Algorithm
• Simulated annealing with objective function
F ( pi , ci )    pi    ci
iIO
j
• pi: Penalty function for noise violation
• ci: cost of decoupling capacitor
• α, β: weights
ASPDAC 06
Lei HE, UCLA
46
Example
• 4 types of decoupling capacitors
• 3 I/O ports
– Each connected to 10 I/O cells
• 90 possible location for decoupling capacitors
• Total 93 ports
• Worst case noise bound: 0.35V
Power planes
Chip I/O Cells
Type
1
2
3
4
50
100
50
100
ESR(Ω)
0.06
0.06
0.03
0.03
ESL(pH)
100
100
40
40
1
2
2
4
ESC(nF)
Price
ASPDAC 06
Lei HE, UCLA
47
Experiment results: noise based
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
3
4
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
0
0
0
0
0
0
0
3
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
3
1
0
Type
1
2
3
4
50
100
50
100
ESR(Ω)
0.06
0.06
0.03
0.03
ESL(pH)
100
100
40
40
1
2
2
4
2
3
ESC(nF)
Price
Chip
port
• Cost=20
ASPDAC 06
1
before optimization
2.52V
after optimization
0.344V 0.343V 0.344V
Lei HE, UCLA
2.49V
2.48V
48
Comparison: Impedance Based
• Cost=72
– 3X larger than noise based
• Impedance bound is not met but noise bound has
already been met.
– Overdesign
port
1
2
3
bound
Maximum
Impedance
5.31Ω
5.59Ω
7.12Ω
0.7Ω
worst-case
noise
0.256
V
0.302
V
0.284
V
0.35V
0
0
0
0
1
0
0
0
2
0
0
0
0
0
0
0
0
0
4
0
0
0
0
0
0
0
0
2
3
0
0
0
0
0
0
0
0
4
3
0
0
4
0
0
0
0
0
1
1
0
0
0
0
0
0
0
2
2
1
0
0
0
0
0
0
0
0
0
0
0
0
0
4
4
0
1
0
4
1
0
0
0
0
0
0
0
2
2
4
0
0
0
0
0
0
4
3
2
1
0
Chip
ASPDAC 06
Lei HE, UCLA
49
0
0
0
0
0
1
4
1
4
0
Runtime Comparison
1
Noise via incremental impedance + decap
2
Noise via admittance matrix inversion
[Zhao et al, EPEP 2004] + decap
3
Impedance + decap [Zheng et al, CICC 2003]
approach
ports
1
2
3
93
93
20
iterations
5881
5403
1920
runtime(s)
389.5
4156.1
2916
0.0662
0.7692
1.519
avg. runtime(s)
• 10x speedup compared to method based on admittance
matrix inversion
ASPDAC 06
Lei HE, UCLA
50
Conclusions
• High-speed IO signaling requires package-aware design
and analysis (co-design)
• Package-aware chip IO planning improves convergence
and turnaround time
• On-chip devices are increasingly exposed to package
effects
• Power integrity is getting harder
• Efficient and accurate macro models are needed to
enable chip-package co-design
ASPDAC 06
Lei HE, UCLA
51
Benefit of Chip-Package Co-Design
(Design from client of Rio Design Automation)
~24% reduction
− Package Size 27mm x 27mm
− Substrate Layers: 3-2-3
− Original Die Size: 7.2 x 7.4mm
− New Die Size: 6.3 x 6.5 mm
− #Voltage Domains: 7
− Two different voltages: 3.3V, 1.8V
− Total IOs: 341
− Frequency: 200Mhz
− Original Bump pitch: x:225, y:225
− New Bump Pitch: X:201, 225, 275
Y: 216, 225
− TSMC 0.18u process
ASPDAC 06
Lei HE, UCLA
52
Reading Assignment
•
J. Xiong, Y. Wong, E. Sarto, L. He, "Constraint Driven I/O Planning and
Placement for Chip-package Co-design", IEEE/ACM Asia South Pacific
Design Automation Conference, Jan. 2006.
•
J. Chen and L. He, “Noise driven in-package decoupling capacitor
optimization for power integrity," in International Symposium on Physical
Design, 2006.
ASPDAC 06
Lei HE, UCLA
53
References
• Extraction and sparsification
– K. Nabors and J. White, "Fastcap: A multipole accelerated 3-d capacitance
extraction program," IEEE Trans. on Computer-Aided Design of Integrated Circuits
and Systems, pp. 1447-1459, Nov. 1991.
– J. Cong, L. He, A. B. Kahng, D. Noice, N. Shirali and S. H.-C. Yen, "Analysis and
Justification of a Simple, Practical 2 1/2-D Capacitance Extraction Methodology",
–
–
–
–
ACM/IEEE Design Automation Conference, June 1997, pp.627-632.
M. Kamon, M. Tsuk, and J. White, "Fasthenry: a multipole-accelerated 3D
inductance extraction program," IEEE Trans. on MIT, 1994.
L. He, N. Chang, S. Lin, and O. S. Nakagawa, "An Efficient Inductance Modeling
for On-chip Interconnects", IEEE Custom Integrated Circuits Conference, pp. 457460, May 1999.
F. Grover, “Inductance Calculations: Working Formulas and Tables”, Dover
Publications, New York, 1946.
M. Beattie, H. Zheng, B. Krauter, A. Devgan, “Spatially Distributed 3D Circuit
Models”, IEEE/ACM Design Automation Conference, 2005
ASPDAC 06
Lei HE, UCLA
54
References
• Model order reduction
– A. Odabasioglu, M. Celik, and L. Pileggi, “PRIMA: Passive reduced-order
interconnect macro-modeling algorithm," IEEE Trans. on Computer-Aided Design
of Integrated Circuits and Systems, pp. 645-654, 1998.
– R. W. Freund, “SPRIM: Structure-preserving reduced-order interconnect macromodeling," in IEEE/ACM ICCAD, 2004.
– H. Yu, L. He, and S. X.D. Tan, “Block Structure Preserving Model Reduction,” IEEE
International Behavioral Modeling and Simulation Conference, September 22-23,
2005.
ASPDAC 06
Lei HE, UCLA
55
References
• P/G noise
– M. Zhao, R. V. Panda, S. S. Sapatnekar, and D. Blaauw, ``Hierarchical analysis
of power distribution networks,'' IEEE Trans. on Computer-Aided Design of
Integrated Circuits and Systems, no. 2, pp.159--168, 2002
– E. Chiprout, ``Fast flip-chip power grid analysis via locality and grid shells,'' in
ICCAD, 2004.
– J. M. Wang and T. V. Nguyen, Extended Krylov subspace method for reduced
order analysis of linear circuits with multiple sources," in Proc. Design
Automation Conf. (DAC), 2000.
– Y. Lee, Y. Cao, T. Chen, J. Wang, and C. Chen, HiPrime: Hierarchical and
passivity preserved interconnect macro-modeling engine for RLKC power
delivery," IEEE Trans. on Computer-Aided Design of Integrated Circuits and
Systems, vol. 26, no. 6, pp. 797-806, 2005.
– Y.Y Shi, H. Yu, and L. He, “Generalized Second-Order Arnoldi Method for Model
Order Reduction with Multiple Non-impulse Sources”, ACM International
Symposium of Physical Design, 2006.
ASPDAC 06
Lei HE, UCLA
56
References
• IO planning and placement
– W.-K. Mak, "I/O placement for FPGAs with multiple I/O standards," IEEE Trans. On
Computer-Aided Design of Integrated Circuits and Systems, vol. 23, pp. 315-320,
February 2004.
– J. Wang, K. Muchherla, and J. Kumar, "A clustering based area I/O planning for
flip-chip technology," in Quality Electronic Design, 5th International Symposium on,
pp.196-201, 2004.
– A. Caldwell, A. Kahng, S. Mantik, and I. Markov, "Implications of area-array I/O for
row-based placement methodology," in IC/Package Design Integration, IEEE
Symposium on, pp. 93-98, Feb. 1998.
– M. M. Ozdal and M. D. Wong, "Simultaneous escape routing and layer assignment
for dense PCBs," in Proc. Int. Conf. on Computer Aided Design, Nov. 2004.
– C. Tan, D. Bouldin, and P. Dehkordi, "Design implementation of intrinsic area array
ICs," in Advanced Research in VLSI, Seventeenth Conference on, pp. 82-93, Sept.
1997.
– J. Xiong, Y. Wong, E. Sarto, L. He, "Constraint Driven I/O Planning and Placement
for Chip-package Co-design", IEEE/ACM Asia South Pacific Design Automation
Conference, Jan. 2006.
ASPDAC 06
Lei HE, UCLA
57
References
• Decoupling capacitors
– L. Smith, R. Anderson, D. Forehand, T. Pelc, and T. Roy, "Power distribution
system design methodology and capacitor selection for modern cmos
technology," IEEE Transactions on Advanced Packaging, vol. 22, pp. 284-291,
1993.
– H. Zheng, B. Krauter, and L. Pileggi, "On-package decoupling optimization with
package macromodels," in Custom Integrated Circuits Conference, 2003.
– J. Zhao and O. P. Mandhana, "A fast evaluation of power delivery system input
impedance of printed circuit boards with decoupling capacitors," in IEEE Topical
Meeting on Electrical Performance of Electronic Packaging, 2004.
– J. Chen and L. He, “Noise driven in-package decoupling capacitor optimization
for power integrity," in International Symposium on Physical Design, 2006.
ASPDAC 06
Lei HE, UCLA
58