Вы находитесь на странице: 1из 22

# Expected Performance Centering for

Analog Designs
Expected Performance Centering for
Analog Designs
Bao Liu and Andrew Kahng
UC San Diego
Outline
Outline
Existing Robust Analog Design Techniques
Expected Performance Centering
Statistical Computation Techniques
Experiments
Conclusion
VLSI Variability
VLSI Variability
Increased variability in nanometer VLSI designs
Process:
OPC Lgate
CMP thickness
Doping Vth
Environment:
Supply voltage transistor performance
Temperature carrier mobility and Vth
These (PVT) variations result in circuit performance
variation
p
1
p
2
PVT Parameter
Distributions
d
1
d
2
Gate/net Delay
Distribution
Analog Design Parametric Yield
Optimization
Analog Design Parametric Yield
Optimization
Analog design
Complex metrics
Strong sensitivity to process variations
1. Robust programming
2. Stochastic programming
3. (explicit parametric yield optimization)
4. Design centering
5. Performance centering
6. Taguchis quality loss function
Robust Programming
Robust Programming
Bound design and process parameter variations in
polyhedrons or ellipsoids
If the objective and the constraints are all
posynomials, this is a geometric programming
Maximize f x
Subject to f x i m
g x i p
x i n
x u
x u i
i
i
sup ( )
sup ( ) , ,...
( ) , ,...
, ,...

=
= =
> =
0
1 1
1 1
0 1
Stochastic Programming
Stochastic Programming
Explicit parametric yield maximization
Y = parametric yield
y = design metrics
U = design specifications
R = acceptability region
Maximize Y
Subject to Y y U x dx D x dx
D x x
R R
= < = =
=

Pr( ) Pr( ) ( )
( ) Pr( )

Design Centering
Design Centering
Explicit parametric yield maximization
Recursive partition of ellipsoids
Recursive partition of polytopes
Maximum parameter distance (to boundaries of an
acceptability region in the design parameter space)
Maximum parameter distance
maximum parametric yield
e.g. 15% Lgate variation
Max distance Max yield
Performance Centering
Performance Centering
Maximum distance to the boundaries of the
performance acceptability region in the
performance space
Performance targeting yield maximization
Performance centering
Design centering
U
y=f(x)
Taguchis Quality Loss Function
Taguchis Quality Loss Function
Increases as the circuit performance deviates from
the target performance
Reducing performance variability as a means for
both parametric yield maximization and
performance targeting
What is the domain of the quality loss function?
Taguchi chooses minimum quality loss
Target performance
U
y=f(x)
Outline
Outline
Existing Robust Analog Design Techniques
Expected Performance Centering
Statistical Computation Techniques
Experiments
Conclusion
Expected Performance Margin
Expected Performance Margin
Expected performance margin in the presence of
design parameter variation P(x) is given by
x = design parameters
= variations
y = design metrics
U = performance constraints
R = acceptability region
= performance margin
( ) ( ) Pr( ) ( ( )) Pr( ) x x x d U y x x d
R
i i i
R
= + + = + +

## Expected Performance Margin

Expected Performance Margin
For extremely small variabilities, e.g., P(x+)=()
expected performance margin = performance margin
For extremely small performance margin (x) =
Expected performance margin = parametric yield
Expected performance margin Taguchis quality
loss function
( ) ( ) ( ) ( ) x x d x
R
= + =

( ) Pr( ) x x d Y
R
= + =

## Expected Performance Centering

Expected Performance Centering
Given
Design parameters x
Design parameter variabilities Pr(x+)
Performance constraints y < U
Find design parameters x* such that the expected
performance margin is maximized
Maximize x x x d
Subject to x U y x
y x U x R i
x
R
i i i
i i

( ) ( ) Pr( )
( ) ( ( ))
( ) , ,
= + +
+ = +
+ +

0
Expected Performance Centering
Expected Performance Centering
Input: x Pr(x+), y < U
Output: x* s.t. f(x*) is maximized
1. Construct a performance macromodel y = f(x)
2. Find acceptability region boundaries y = f(x) = U
3. Compute performance margin (x)
4. Computer expected performance margin (x)
5. Find maximum expected performance margin (x*)
Expected Performance Margin
Computation
Expected Performance Margin
Computation
Sampling
posynomial geometric programming
Gaussian integral
if y = f(x) is polynomial, Pr(x) is Gaussian
Surface integral

sampling i i i i
i
N
N
U y x a = + =
=

1
1
( ( ))
1
2
1
2
2
2
2
2
0

e d erf
z
z

=

( )
( )

( ) ( ( )) Pr( ) ( ( )) ( )
( ) Pr( )
x U y x x d U y x D x d
D x x
i i i
R
i i i
R
= + + = + +
=

Outline
Outline
Existing Robust Analog Design Techniques
Expected Performance Centering
Statistical Computation Techniques
Experiments
Conclusion
Experimental Setup
Experimental Setup
Three-inverter ring oscillator
Synopsys HSpiceRF simulator
70nm Berkeley Predictive Technology Model
Performance metrics:
Oscillation frequency f
0
> 1GHz
Phase noise N < 40dB
Power consumption P < 1mW
Delay parameters
Transistor channel width and length
Experimental Results
Experimental Results
Acceptability region and performance contours
Design Center
Expected Perf. Center
Perf. Center
Experiments
Experiments
I. Design centering
II. Performance centering
III. Expected performance centering
= performance margin
= expected performance margin
3.00
3.04
2.91
L
79.76
80.33
75.37

67.12 128.0 II
III
I
128.0
128.0
W
67.53
65.42

Outline
Outline
Existing Robust Analog Design Techniques
Expected Performance Centering
Statistical Computation Techniques
Experiments
Conclusion
Summary
Summary
We propose maximum expected performance
margin as a new analog design objective
performance targeting in the presence of small
process variations
parametric yield maximization (design centering)
for small performance margin
Taguchis quality loss function
Expected performance margin computation and
maximization methods
Thank you !
Thank you !