Professional Documents
Culture Documents
P - P E L P C D: EAK Ower Stimation FOR OW Ower Ircuit Esign
P - P E L P C D: EAK Ower Stimation FOR OW Ower Ircuit Esign
testable
circuit
by
Focus
Issues arising from excessive test power Algorithmic solutions proposed to alleviate the lowpower test problems
PEAK POWER
Peak
power corresponds to the highest value of instantaneous power measured during testing.
Peak
power generally determines the thermal and electrical limits of the circuit and the system packaging requirements.
PEAK POWER
Peak Power can be expressed in terms of the highest energy consumed during a small instant of time during the test session:
PEAK POWER
The
average power during scan testing can be 3 times the power consumed during normal functional operation peak power can be 30 times what it is in normal functional operation
The
INTRODUCTION
Power dissipation in test mode is much higher than during functional mode
Test vectors ignore functional constraints DFT circuitry is intensively used Parallel testing is often used for efficiency Low-power functional features (e.g., gated clock) often disabled during test
has impact on the battery lifetime during power up or periodic self-test of battery operated devices
determines the thermal and electrical limits of components and the system packaging requirements
8
Dynamic power consumption coming from the application of the i-th vector is proportional to the circuit weighted switching activity (wsai)
Cg is the physical capacitance associated to gate g NT is the number of times the output of gate g toggles due to the application of the i-th vector.
Both average and peak power consumption during testing can be efficiently minimized with the MTfilling heuristics
ALPS ALGORITHM
ALPS: Analyzer of Low-Power Systems
Symbolic computation techniques, which allow computing the complete set of reachable states for sequential circuits of small and medium size
Analyzes the circuit and, by exploiting symbolic calculation techniques, computes all the reachable states reached
Peak-Power computation
3-tuple (S, V1, V2), where S reached, which maximizes the power consumption
PEAK-POWER ESTIMATION
A Genetic Algorithm has been adopted to find a 3tuple (S, V1, V2) that maximizes the peak-power consumption Encoding
Cross-over operator
Mutation operator
PEAK-POWER ESTIMATION
search for S in the previously computed states
S is substituted with the reachable one having minimum Hamming distance from S
Fitness function
RESULTS
PP(%) [Re PP(%) PP(%) Implemen [Symbolic [Approxim tation] ALPS] ated ALPS] 0 -17.28 -80.27 -59.06 -17.79 0 0 400 63.99 5.32 0 0 400 21.50 0
FF 8 14 15 19 21
H 0 3 1 6 6
CONCLUSION
Peak-power
is tightly coupled with the reachability of the initial state algorithm able to produce better results than those provided by previously proposed methods
The
QUERIES?
20