Sei sulla pagina 1di 45

Low Power Design of CMOS

Circuits
Vishwani D. Agrawal
James J. Danaher Professor
ECE Dept., Auburn University, Auburn, AL 36849

Nov 19, 2009 Agrawal: Low Power Design 1


CMOS Logic (Inverter)

No static leakage
path exists for either
1 or 0 input.

F. M. Wanlass and C. T. Sah, “Nanowatt Logic using Field-


Effect Metal-Oxide-Semiconductor Triodes,” IEEE International
Solid-State Circuits Conference Digest, vol. IV, February 1963,
pp. 32-33.

Nov 19, 2009 Agrawal: Low Power Design 2


Power of a CMOS Gate
Transition
VDD
Dynamic Power
R
= CLVDD2/2 + Psc
Vo
Vi
CL Static power
R
= VDD Ileakage
isc
Ground

Nov 19, 2009 Agrawal: Low Power Design 3


Power Consumption of VLSI Chips

Why is it a concern?

Nov 19, 2009 Agrawal: Low Power Design 4


ISSCC, Feb. 2001, Keynote
“Ten years from now,
microprocessors will run at 10GHz
to 30GHz and be capable of
processing 1 trillion operations per
second – about the same number of
calculations that the world's fastest
supercomputer can perform now.
Patrick P. Gelsinger
Senior Vice President
General Manager
“Unfortunately, if nothing changes
Digital Enterprise Group these chips will produce as much
INTEL CORP. heat, for their proportional size, as a
nuclear reactor. . . .”

Nov 19, 2009 Agrawal: Low Power Design 5


VLSI Chip Power Density
10000
Sun’s
Surface

Rocket
Power Density (W/cm2)

1000
Nozzle

Nuclear
100
Reactor
8086 Hot Plate
10 4004 P6
8008 8085 386 Pentium®
286 486
8080 Source: Intel
1
1970 1980 1990 2000 2010
Year
Nov 19, 2009 Agrawal: Low Power Design 6
Low-Power Design
• Design practices that reduce power
consumption by at least one order of
magnitude; in practice 50% reduction is
often acceptable.
• Low-power design methods:
– Algorithms and architectures
– High-level and software techniques
– Gate and circuit-level methods
– Test power

Nov 19, 2009 Agrawal: Low Power Design 7


Components of Power
• Dynamic
– Signal transitions
• Logic activity
• Glitches
– Short-circuit Ptotal = Pdyn + Pstat
• Static
= Ptran + P + P Then
– Leakage sc stat

= Ptran + P + Pstat Now


sc

Nov 19, 2009 Agrawal: Low Power Design 8


Dynamic Power
• Each transition of a gate consumes CV 2/2.
• Methods of power saving:
– Minimize load capacitances
• Transistor sizing
– Reduce transitions
• Logic design
• Glitch reduction

Nov 19, 2009 Agrawal: Low Power Design 9


Glitch Power Reduction
• Design a digital circuit for minimum transient energy
consumption by eliminating hazards

Total transitions = 6
Essential transitions = 2
Glitch transitions = 4

Nov 19, 2009 Agrawal: Low Power Design 10


Multi-Input Gate
A

DPD: Differential path delay Delay


C
D < DPD
B

A DPD
B

D D Hazard or glitch
C
time

Nov 19, 2009 Agrawal: Low Power Design 11


Balanced Path Delays
A DPD

Delay
Delay buffer C
D < DPD
B

A
B

D No glitch
C
time

Nov 19, 2009 Agrawal: Low Power Design 12


Glitch Filtering by Inertia
A

Delay
C
D > DPD
B

A DPD
B

D > DPD

C Filtered glitch
time

Nov 19, 2009 Agrawal: Low Power Design 13


Designing a Glitch-Free Circuit
• Maintain specified critical path delay.
• Glitch suppressed at all gates by
– Path delay balancing
– Glitch filtering by increasing inertial delay of gates or by inserting delay buffers
when necessary.
• A linear program optimally combines all objectives.

Path delay = d1
Delay
Path delay = d2 D

Minimum transient energy condition: |d1 – d2| < D

Nov 19, 2009 Agrawal: Low Power Design 14


Linear Program (LP)
• Variables: gate and buffer delays, arrival time
variables.
• Objective: minimize number of delay buffers.
• Subject to: overall circuit delay constraint for
all input-output paths.
• Subject to: minimum transient energy
condition for all multi-input gates.

Nov 19, 2009 Agrawal: Low Power Design 15


An Example: Full Adder
1

1 1

1
1 1
1
1
1

Critical path delay = 6

Nov 19, 2009 Agrawal: Low Power Design 16


LP Step 1: Define Varaibles

 Gate delay variables: d4 . . . d12


 Buffer delay variables: d15 . . . d29
 Arrival time variables (earliest): t4 . . . T29
(longest): T4 . . . . T29
Nov 19, 2009 Agrawal: Low Power Design 17
LP Step 2: Specify Constraints

For Gate 7:
T 7 ≥ T 5 + d7 t7 ≤ t5 + d7 d7 > T7 - t7
T 7 ≥ T 6 + d7 t7 ≤ t6 + d7
Nov 19, 2009 Agrawal: Low Power Design 18
LP Step 2 (Cont.)

Buffer 19: T16 + d19 = T19


t16 + d19 = t19
Nov 19, 2009 Agrawal: Low Power Design 19
LP Step 2: Critical Path Constraints

T11 ≤ maxdelay
T12 ≤ maxdelay
maxdelay is specified

Nov 19, 2009 Agrawal: Low Power Design 20


LP Step 3: Define Objective Function
• Need to minimize the number of buffers.
• Because that leads to a nonlinear objective
function, we use an approximate criterion:
minimize ∑ (all buffer delays)
i.e., minimize d15 + d16 + ∙ ∙ ∙ + d29
• This gives near optimum results.

Nov 19, 2009 Agrawal: Low Power Design 21


LP Solution: maxdelay = 6
1 2

1 1

1
1 1
2
1
2
2
Critical path delay = 6

Nov 19, 2009 Agrawal: Low Power Design 22


LP Solution: maxdelay = 7
3

1 1

1
1 1
2
2
1
2
Critical path delay = 7

Nov 19, 2009 Agrawal: Low Power Design 23


LP Solution: maxdelay ≥ 11

1 1

1
3 1
2
3
4

Critical path delay = 11

Nov 19, 2009 Agrawal: Low Power Design 24


ALU4: Original and Glitch-Free

Nov 19, 2009 Agrawal: Low Power Design 25


C7552 Circuit: Spice Simulation

Power Saving: Average 58%, Peak 68%


Nov 19, 2009 Agrawal: Low Power Design 26
Components of Power
• Dynamic
– Signal transitions
• Logic activity
• Glitches
– Short-circuit
• Static
– Leakage

Nov 19, 2009 Agrawal: Low Power Design 27


Leakage Reduction Problem
65nm CMOS technology:
Low threshold transistors, gate delay 5ps, leakage current 10nA.
High threshold transistors, gate delay 12ps, leakage 1nA.
Minimize leakage current without increasing critical path delay. What is the
percentage reduction in leakage power?
What will be leakage power reduction if 30% critical path delay increase is allowed?

Nov 19, 2009 Agrawal: Low Power Design 28


Solution 1: No Delay Increase
Reduction in leakage power = 1 – (4×1+7×10)/(11×10) = 32.73%
Critical path delay = 25ps

5ps 12ps

5ps 12ps

5ps 5ps
5ps
5ps
12ps
5ps
12ps

Nov 19, 2009 Agrawal: Low Power Design 29


Solution 2: 30% Delay Increase
Several solutions are possible. Notice that any 3-gate path can have 2 high
threshold gates. Four and five gate paths can have only one high threshold gate.
One solution is shown in the figure below where six high threshold gates are shown
with shading and the critical path is shown by a dashed red line arrow.

Reduction in leakage power = 1 – (6×1+5×10)/(11×10) = 49.09%


Critical path delay = 29ps

5ps 12ps

5ps 12ps

5ps 12ps
12ps
5ps
12ps
5ps
12ps

Nov 19, 2009 Agrawal: Low Power Design 30


Integer Linear Programming (ILP) to
Minimize Leakage Power
• Assign every gate i an integer [0,1] variable Xi.
• Define ILP constraints and objective function:
• On critical path delay
• Minimize total leakage
• Let ILP find values of Xi’s:
If Xi = 1, assign low threshold to gate i
If Xi = 0, assign high threshold to gate i

Nov 19, 2009 Agrawal: Low Power Design 31


Power-Delay Tradeoff
1

0.9

0.8
C432
Normalized Leakage Power

0.7
C880
0.6
C1908

0.5

0.4

0.3

0.2

0.1

1 1.1 1.2 1.3 1.4 1.5


Normalized Critical Path Delay
Nov 19, 2009 Agrawal: Low Power Design 32
Leakage & Dynamic Power Optimization 70nm
CMOS c7552 Benchmark Circuit @ 90oC
900
800
700
Microwatts

600
500
Leakage power
400 Dynamic power
dy ce age
po n a m d s

300 Total power


er ic
ex ak
e
Le

200 Y. Lu and V. D. Agrawal, “CMOS


w

Leakage and Glitch


100 Minimization for Power-
Performance Tradeoff,” Journal
0 of Low Power Electronics
Original circuit Optimized (JOLPE), vol. 2, no. 3, pp. 378-
387, December 2006.
design
Nov 19, 2009 Agrawal: Low Power Design 33
Power Constrained Test Scheduling
R1 R2

M1 M2

R3 R4

A datapath

Nov 19, 2009 Agrawal: Low Power Design 34


Minimum Test Time
LFSR1 LFSR2

T2: test for M2

Test power
M1 M2

T1: test for M1

MISR1 MISR2
Test time

Nov 19, 2009 Agrawal: Low Power Design 35


Minimum Test Power
R1 LFSR2

Test power
M1 M2
T1: test for M1
T2: test for M2
MISR1 MISR2
Test time

Nov 19, 2009 Agrawal: Low Power Design 36


Testing of MCM and SOC
• Test resources: Typically registers and
multiplexers that can be reconfigured as test
pattern generators (e.g., LFSR) or as output
response analyzers (e.g., MISR).
• Test resources (R1, . . .) and tests (T1, . . .) are
identified for the system to be tested.
• Each test is characterized for test time, power
dissipation and resources it requires.

Nov 19, 2009 Agrawal: Low Power Design 37


Resource Allocation Graph
(A Bipartite Graph)
T1 T2 T3 T4 T5 T6

R1 R2 R3 R4 R5 R6 R7 R8 R9

Nov 19, 2009 Agrawal: Low Power Design 38


Test Compatibility Graph (TCG)
T1
(2, 100)

T6 T2
(1, 100) (1,10)

T5 T3
(2, 10) (1, 10)

Power
Test time T4 Tests that form a
(1, 5)
Pmax = 4 clique can be
performed concurrently.
Nov 19, 2009 Agrawal: Low Power Design 39
Find All Cliques in TCG
CLIQUE NO. i TEST NODES TEST LENGTH, Li POWER, Pi
1 T1, T3, T5 100 5
2 T1, T3, T4 100 4
3 T1, T6 100 3
4 T1, T5 100 4
5 T1, T4 100 3
6 T1. T3 100 3
7 T2, T6 100 2
8 T2, T5 10 3
9 T3, T5 10 3
10 T3, T4 10 2
11 T1 100 2
12 T2 10 1
13 T3 10 1
14 T4 5 1
15 T5 10 2
16 T6 100 1
Nov 19, 2009 Agrawal: Low Power Design 40
Integer Linear Program (ILP)
• For each clique (test session) i, define:
– Integer variable, xi = 1, test session selected, or xi
= 0, test session not selected.
– Constants, Li = test length, Pi = power.
• Constraints to cover all tests:
– T1 is covered if x1+x2+x3+x4+x5+x6+x11 ≥ 1
– Similar constraint for each test, Tk
• Constraints for power: Pi × xi ≤ Pmax

Nov 19, 2009 Agrawal: Low Power Design 41


ILP Objective and Solution
• Objective function:
– Minimize Σ Li × xi
all cliques
• Solution:
– x3 = x8 = x10 = 1, all other xi’s are 0
• Test session 3 includes T1 and T6
• Test session 8 includes T2 and T5
• Test session 10 includes T3 and T4
– Test length = L3 + L8 + L10 = 120
– Peak power = max {P3, P8, P10} = 3 (Pmax = 4)
Nov 19, 2009 Agrawal: Low Power Design 42
Summary
• Underlying theme in our research – use of
mathematical optimization methods for power
reduction at gate level:
• Dynamic power
• Leakage power
• Power minimization under process variation
• Test power
• Other research
• Min-max power estimation
• Architecture level power management

Nov 19, 2009 Agrawal: Low Power Design 43


Our Research Students
• T. Raja, MS 2002, PhD 2004 (NVIDIA)
• S. Uppalapati, MS 2004 (Intel)
• F. Hu, PhD 2006 (Intel)
• Y. Lu, PhD 2007 (Intel)
• J. D. Alexander, MS 2008
• K. Sheth, MS 2008
• M. Allani, PhD
• J. Yao, PhD
• K. Kim, PhD
• M. Kulkarni, MS

Nov 19, 2009 Agrawal: Low Power Design 44


Dissertations and Papers
• Dissertations:
http://www.eng.auburn.edu/~vagrawal/THESIS/thesis.html
• Papers:
http://www.eng.auburn.edu/~vagrawal/TALKS/talks.html

Nov 19, 2009 Agrawal: Low Power Design 45

Potrebbero piacerti anche