a framework to model self-adaptive computing...
TRANSCRIPT
A Framework to ModelSelf-Adaptive Computing Systems
Cristiana BOLCHINIMatteo CARMINATIAntonio MIELEElisa [email protected]
AHS 2013 @Politecnico di Torino - ItalyJune 25, 2013
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Motivations
2
‣ Context-awareness and self-adaptiveness are growing trends in designing new computing systems
‣ But rigorous definitions and formal models are usually neglected or overlooked
‣ This becomes a limitation when the number of elements determining the context grows: the quest for a flexible and powerful support arises
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Context
3
Lot of attention from SE and DB research areas
‣ context: information that can be used to characterize situation of an entity [1]
‣ context-aware system: uses context to provide relevant information and/or services to the user, where relevancy depends on the user’s task [1]
[1] A.K. Dey and G.D. Abowd - Towards a better understanding of context and context-awareness - Workshop on the What, Who, Where, When, and How of Context-Awareness, 2000
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Context
4
[2] M. Baldauf, S. Dustdar, and F. Rosenberg - A survey on context-aware systems - International Journal Ad Hoc and Ubiquitous Computing (IJAHUC), 2007.
acquisition
representation
usage interpretation
reasoning
context [2]
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Context
5
[2] M. Baldauf, S. Dustdar, and F. Rosenberg - A survey on context-aware systems - International Journal Ad Hoc and Ubiquitous Computing (IJAHUC), 2007.
acquisition
representation
usage interpretation
reasoning
modeling
context [2]
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Context
6
[2] M. Baldauf, S. Dustdar, and F. Rosenberg - A survey on context-aware systems - International Journal Ad Hoc and Ubiquitous Computing (IJAHUC), 2007. [3] B. Schilit, N. Adams, and R. Want - Context-aware computing applications - Workshop on Mobile Computing Systems and Applications (HotMobile), 1994.
[3] context [2]
acquisition
usage interpretation
reasoning
user
physical
computing
representationmodeling
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Context
7
[2] M. Baldauf, S. Dustdar, and F. Rosenberg - A survey on context-aware systems - International Journal Ad Hoc and Ubiquitous Computing (IJAHUC), 2007. [3] B. Schilit, N. Adams, and R. Want - Context-aware computing applications - Workshop on Mobile Computing Systems and Applications (HotMobile), 1994.
[3] context [2]
acquisition
usage interpretation
reasoning
user
physical
computing
representationmodeling
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Self-Adaptiveness[4]
8
[4] T. Becker, A. Agne, P. R. Lewis, R. Bahsoon, F. Faniyi, L. Esterle, A. Keller, A. Chandra, A. R. Jensenius, and S. C. Stilkerich - EPiCS: Engineering Proprioception in Computing Systems - International Conference on Computational Science and Engineering (CSE), 2012.
Sensor
State & ContextPrivate
Self-AwarenessEngine
LearntModel(s)
PublicSelf-Awareness
Engine
Self-ExpressionEngine
Monitor/Controller
GoalsValues
ObjectivesConstraints
Sensor
Sensor
Actuator
Actuator
Environment
ExternalActions
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Self-Adaptiveness[4]
9
Sensor
State & ContextPrivate
Self-AwarenessEngine
LearntModel(s)
PublicSelf-Awareness
Engine
Self-ExpressionEngine
Monitor/Controller
GoalsValues
ObjectivesConstraints
Sensor
Sensor
Actuator
Actuator
Environment
ExternalActions
[4] T. Becker, A. Agne, P. R. Lewis, R. Bahsoon, F. Faniyi, L. Esterle, A. Keller, A. Chandra, A. R. Jensenius, and S. C. Stilkerich - EPiCS: Engineering Proprioception in Computing Systems - International Conference on Computational Science and Engineering (CSE), 2012.
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Self-Adaptiveness[4]
10
[4] T. Becker, A. Agne, P. R. Lewis, R. Bahsoon, F. Faniyi, L. Esterle, A. Keller, A. Chandra, A. R. Jensenius, and S. C. Stilkerich - EPiCS: Engineering Proprioception in Computing Systems - International Conference on Computational Science and Engineering (CSE), 2012.
Sensor
State & ContextPrivate
Self-AwarenessEngine
LearntModel(s)
PublicSelf-Awareness
Engine
Self-ExpressionEngine
Monitor/Controller
GoalsValues
ObjectivesConstraints
Sensor
Sensor
Actuator
Actuator
Environment
ExternalActions
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Self-Adaptiveness[4]
11
[4] T. Becker, A. Agne, P. R. Lewis, R. Bahsoon, F. Faniyi, L. Esterle, A. Keller, A. Chandra, A. R. Jensenius, and S. C. Stilkerich - EPiCS: Engineering Proprioception in Computing Systems - International Conference on Computational Science and Engineering (CSE), 2012.
Sensor
State & ContextPrivate
Self-AwarenessEngine
LearntModel(s)
PublicSelf-Awareness
Engine
Self-ExpressionEngine
Monitor/Controller
GoalsValues
ObjectivesConstraints
Sensor
Sensor
Actuator
Actuator
Environment
ExternalActions
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Self-Adaptiveness[4]
12
[4] T. Becker, A. Agne, P. R. Lewis, R. Bahsoon, F. Faniyi, L. Esterle, A. Keller, A. Chandra, A. R. Jensenius, and S. C. Stilkerich - EPiCS: Engineering Proprioception in Computing Systems - International Conference on Computational Science and Engineering (CSE), 2012.
Sensor
State & ContextPrivate
Self-AwarenessEngine
LearntModel(s)
PublicSelf-Awareness
Engine
Self-ExpressionEngine
Monitor/Controller
GoalsValues
ObjectivesConstraints
Sensor
Sensor
Actuator
Actuator
Environment
ExternalActions
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Self-Adaptiveness[4]
13
[4] T. Becker, A. Agne, P. R. Lewis, R. Bahsoon, F. Faniyi, L. Esterle, A. Keller, A. Chandra, A. R. Jensenius, and S. C. Stilkerich - EPiCS: Engineering Proprioception in Computing Systems - International Conference on Computational Science and Engineering (CSE), 2012.
Sensor
State & ContextPrivate
Self-AwarenessEngine
LearntModel(s)
PublicSelf-Awareness
Engine
Self-ExpressionEngine
Monitor/Controller
GoalsValues
ObjectivesConstraints
Sensor
Sensor
Actuator
Actuator
Environment
ExternalActions
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Work Goals
14
‣ The definition of a model for self-adaptive Computing Systems to express:
‣ the elements affecting their behavior, including existing relations and constraints
‣ the conditions that trigger adaptation
‣ The validation of the completeness and flexibility of the model, by applying it to self-adaptive systems from literature
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
What is relevant?
15
Self-AdaptiveComputing Systems
(SACS)
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
What is relevant?
16
elements
ODA control loop
O: high-level quantitiesD: aspects to reason onA: knobs and strategies
Self-AdaptiveComputing Systems
(SACS)
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
What is relevant?
17
elements relations
ODA control loop
O: high-level quantitiesD: aspects to reason onA: knobs and strategies
Both direct and indirect effects of planned actions
Self-AdaptiveComputing Systems
(SACS)
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Context Dimensions
18
MainDimensions
goals
requirements
observations
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Context Dimensions
19
MainDimensions
goals
requirements
observations
Collected DataDimensions
raw data
measures
metrics
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Context Dimensions
20
MainDimensions
goals
requirements
observations
Collected DataDimensions
raw data
measures
metrics
Self-ExpressionDimensions
controlactions
methods
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
observations
measuresmetrics raw datacontrol actions methods requirements
goals
Context Meta-Model
21
drivingdimensions
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
observations
measuresmetrics raw datacontrol actions methods requirements
goals
Context Meta-Model
22
drivingdimensions
may not existin some contexts
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
observations
measuresmetrics raw datacontrol actions methods requirements
goals
Context Meta-Model
23
drivingdimensions
may not existin some contexts
segments representrelations between
dimensions
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
observations
measuresmetrics raw datacontrol actions methods requirements
goals
Context Meta-Model
24
drivingdimensions
may not existin some contexts
segments representrelations between
dimensionsOR relation AND relation
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
observations
measuresmetrics raw datacontrol actions methods requirements
goals
Context Meta-Model
25
drivingdimensions
may not existin some contexts
OR relation AND relationsegments representrelations between
dimensions
New dimensions can be added by simply connecting them to the existing ones through (possibly new) relations
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Context Model
26
control actions methods metrics measures raw data goals/requirements/observations✦ #Allocated Cores✦ Off-Chip Memory Bandwidth✦ Shared Cache Space✦ Idle Cycle Injection✦ Core Frequency✦ . . .
✦ SIMO controller & ARMA model✦ Controlled Task Scheduling✦ Round Robin✦ FIFO✦ Constant Value✦ . . .
✦ Performance✦ Resource Exploitation✦ Power✦ Temperature✦ Reliability✦ Area✦ Manufacturability✦ . . .
✦ Weighted IPC✦ Harmonic Speed-Up✦ Resources Utilization✦ Mean Time To Failure✦ . . .
✦ Instructions per Cycle (IPC)✦ Average Resource Usage✦ Execution Time✦ Transactions per Second✦ . . .
✦ Cores Temperature✦ Screen Brightness✦ Acceleration✦ Execution Time✦ Memory Usage✦ CPU Utilization✦ Cache Hits✦ . . .
Each dimension has a domain of values whose selection will define the specific context model
driving dim. secondary dim.
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Context Model
27
‣ For a given SACS, the context model is defined as
‣ a set of <dimension, value> pairs
where each is a dimensionand is its value
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
‣ a set of <dimension, value> pairs
where each is a dimensionand is its value
Context Model
28
‣ For a given SACS, the context model is defined as
‣ satisfying the following constraints
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
goals requirementscontrol actions methods metrics measures raw data✦ #Allocated Cores✦ Off-Chip Memory Bandwidth✦ Shared Cache Space
✦ SIMO controller & ARMA model
✦ Performance✦ Resources Exploitation
✦ Weighted IPC✦ Harmonic Speed-Up✦ Resources Utilization
✦ Instructions per Cycle (IPC)
✦ ∑ Allocated Cores✦ ∑ Off-Chip Memory Bandwidth✦ ∑ Shared Cache Space
✦ #Executed Instructions✦ Execution Time✦ #Allocated Cores/App✦ Off-Chip Memory Bandwidth/App✦ Shared Cache Space/App
METE[4]
29
[4] A. Sharifi, S. Srikantaiah, A. K. Mishra, M. Kandemir, and C. R. Chita - METE: meeting end-to-end QoS in multicores through system-wide resource management - International Conference on Measurement and Modeling of Computer Systems (SIGMETRICS), 2011.
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
goals requirements✦ Performance
✦ Resources Exploitation
METE[4]
30
[4] A. Sharifi, S. Srikantaiah, A. K. Mishra, M. Kandemir, and C. R. Chita - METE: meeting end-to-end QoS in multicores through system-wide resource management - International Conference on Measurement and Modeling of Computer Systems (SIGMETRICS), 2011.
goals requirementscontrol actions methods metrics measures raw data✦ #Allocated Cores✦ Off-Chip Memory Bandwidth✦ Shared Cache Space
✦ SIMO controller & ARMA model
✦ Performance✦ Resources Exploitation
✦ Weighted IPC✦ Harmonic Speed-Up✦ Resources Utilization
✦ Instructions per Cycle (IPC)
✦ ∑ Allocated Cores✦ ∑ Off-Chip Memory Bandwidth✦ ∑ Shared Cache Space
✦ #Executed Instructions✦ Execution Time✦ #Allocated Cores/App✦ Off-Chip Memory Bandwidth/App✦ Shared Cache Space/App
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
goals requirementscontrol actions methods metrics measures raw data✦ #Allocated Cores✦ Off-Chip Memory Bandwidth✦ Shared Cache Space
✦ SIMO controller & ARMA model
✦ Performance✦ Resources Exploitation
✦ Weighted IPC✦ Harmonic Speed-Up✦ Resources Utilization
✦ Instructions per Cycle (IPC)
✦ ∑ Allocated Cores✦ ∑ Off-Chip Memory Bandwidth✦ ∑ Shared Cache Space
✦ #Executed Instructions✦ Execution Time✦ #Allocated Cores/App✦ Off-Chip Memory Bandwidth/App✦ Shared Cache Space/App
METE[4]
31
[4] A. Sharifi, S. Srikantaiah, A. K. Mishra, M. Kandemir, and C. R. Chita - METE: meeting end-to-end QoS in multicores through system-wide resource management - International Conference on Measurement and Modeling of Computer Systems (SIGMETRICS), 2011.
metrics measures raw data✦ Weighted IPC✦ Harmonic Speed-Up✦ Resources Utilization
✦ Instructions per Cycle (IPC)
✦ ∑ Allocated Cores✦ ∑ Off-Chip Memory Bandwidth✦ ∑ Shared Cache Space
✦ #Executed Instructions✦ Execution Time✦ #Allocated Cores/App✦ Off-Chip Memory Bandwidth/App✦ Shared Cache Space/App
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
goals requirementscontrol actions methods metrics measures raw data✦ #Allocated Cores✦ Off-Chip Memory Bandwidth✦ Shared Cache Space
✦ SIMO controller & ARMA model
✦ Performance✦ Resources Exploitation
✦ Weighted IPC✦ Harmonic Speed-Up✦ Resources Utilization
✦ Instructions per Cycle (IPC)
✦ ∑ Allocated Cores✦ ∑ Off-Chip Memory Bandwidth✦ ∑ Shared Cache Space
✦ #Executed Instructions✦ Execution Time✦ #Allocated Cores/App✦ Off-Chip Memory Bandwidth/App✦ Shared Cache Space/App
METE[4]
32
[4] A. Sharifi, S. Srikantaiah, A. K. Mishra, M. Kandemir, and C. R. Chita - METE: meeting end-to-end QoS in multicores through system-wide resource management - International Conference on Measurement and Modeling of Computer Systems (SIGMETRICS), 2011.
control actions methods✦ #Allocated Cores✦ Off-Chip Memory Bandwidth✦ Shared Cache Space
✦ SIMO controller & ARMA model
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
METE[4]
33
[4] A. Sharifi, S. Srikantaiah, A. K. Mishra, M. Kandemir, and C. R. Chita - METE: meeting end-to-end QoS in multicores through system-wide resource management - International Conference on Measurement and Modeling of Computer Systems (SIGMETRICS), 2011.
goals requirementscontrol actions methods metrics measures raw data✦ #Allocated Cores✦ Off-Chip Memory Bandwidth✦ Shared Cache Space
✦ SIMO controller & ARMA model
✦ Performance✦ Resources Exploitation
✦ Weighted IPC✦ Harmonic Speed-Up✦ Resources Utilization
✦ Instructions per Cycle (IPC)
✦ ∑ Allocated Cores✦ ∑ Off-Chip Memory Bandwidth✦ ∑ Shared Cache Space
✦ #Executed Instructions✦ Execution Time✦ #Allocated Cores/App✦ Off-Chip Memory Bandwidth/App✦ Shared Cache Space/App
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
goals requirementscontrol actions methods metrics measures raw data✦ #Allocated Cores✦ Off-Chip Memory Bandwidth✦ Shared Cache Space
58.1
6 15
4
621.9 1.4 15
10245 300 2 680.2
0.8
✦ Weighted IPC✦ Harmonic Speed-Up✦ Resources Utilization
25 325 0.8
✦ SIMO controller & ARMA model
✦ Performance✦ Resources Exploitation
✦ Instructions per Cycle (IPC)
✦ ∑ Allocated Cores✦ ∑ Off-Chip Memory Bandwidth✦ ∑ Shared Cache Space
✦ #Executed Instructions✦ Execution Time✦ #Allocated Cores/App✦ Off-Chip Memory Bandwidth/App✦ Shared Cache Space/App
goals requirementscontrol actions methods metrics measures raw data✦ #Allocated Cores✦ Off-Chip Memory Bandwidth✦ Shared Cache Space
✦ SIMO controller & ARMA model
✦ Performance✦ Resources Exploitation
✦ Weighted IPC✦ Harmonic Speed-Up✦ Resources Utilization
✦ Instructions per Cycle (IPC)
✦ ∑ Allocated Cores✦ ∑ Off-Chip Memory Bandwidth✦ ∑ Shared Cache Space
✦ #Executed Instructions✦ Execution Time✦ #Allocated Cores/App✦ Off-Chip Memory Bandwidth/App✦ Shared Cache Space/App
Re-cap
34
observations
measuresmetrics raw datacontrol actions methods requirements
goalsContextMeta-Model
ContextModel
ContextInstance
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
goals requirementscontrol actions methods metrics measures raw data✦ #Allocated Cores✦ Off-Chip Memory Bandwidth✦ Shared Cache Space
✦ SIMO controller & ARMA model
✦ Performance✦ Resources Exploitation
✦ Weighted IPC✦ Harmonic Speed-Up✦ Resources Utilization
✦ Instructions per Cycle (IPC)
✦ ∑ Allocated Cores✦ ∑ Off-Chip Memory Bandwidth✦ ∑ Shared Cache Space
✦ #Executed Instructions✦ Execution Time✦ #Allocated Cores/App✦ Off-Chip Memory Bandwidth/App✦ Shared Cache Space/App
Why do we need it?
35
Context representation:
‣ Help the computing system designer in understanding which resources are needed to be able to pursue its goal
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
goals requirementscontrol actions methods metrics measures raw data✦ #Allocated Cores✦ Off-Chip Memory Bandwidth✦ Shared Cache Space
✦ SIMO controller & ARMA model
✦ Performance✦ Resources Exploitation
✦ Weighted IPC✦ Harmonic Speed-Up✦ Resources Utilization
✦ Instructions per Cycle (IPC)
✦ ∑ Allocated Cores✦ ∑ Off-Chip Memory Bandwidth✦ ∑ Shared Cache Space
✦ #Executed Instructions✦ Execution Time✦ #Allocated Cores/App✦ Off-Chip Memory Bandwidth/App✦ Shared Cache Space/App
Why do we need it?
36
Context representation:
‣ Help the computing system designer in understanding which resources are needed to be able to pursue its goal
sensors
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
goals requirementscontrol actions methods metrics measures raw data✦ #Allocated Cores✦ Off-Chip Memory Bandwidth✦ Shared Cache Space
✦ SIMO controller & ARMA model
✦ Performance✦ Resources Exploitation
✦ Weighted IPC✦ Harmonic Speed-Up✦ Resources Utilization
✦ Instructions per Cycle (IPC)
✦ ∑ Allocated Cores✦ ∑ Off-Chip Memory Bandwidth✦ ∑ Shared Cache Space
✦ #Executed Instructions✦ Execution Time✦ #Allocated Cores/App✦ Off-Chip Memory Bandwidth/App✦ Shared Cache Space/App
Why do we need it?
37
Context representation:
‣ Help the computing system designer in understanding which resources are needed to be able to pursue its goal
actuators
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Why do we need it?
38
Context representation:
‣ Help the computing system designer in understanding which resources are needed to be able to pursue its goal
‣ Document the system to provide a common and systematic classification of self-adaptive computing systems
‣ Describe, at run-time, the current context of the considered system
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
What can be done?
39
Context usage:
‣ Development of a framework able to generate, in a template fashion, the main control loop of the self-adaptive engine, according to the driving dimensions value
‣ Automate the testing phase, by creating optimal test sets to stimulate all possible system configurations
‣ Exploit the knowledge contained in our formalization to keep track of the context evolution, to deal with unforeseen contexts
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Questions?
40
Cristiana BOLCHINIMatteo CARMINATIAntonio MIELEElisa QUINTARELLI
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Questions?
41
Cristiana BOLCHINIMatteo CARMINATIAntonio MIELEElisa QUINTARELLI
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Into The Wild[5]
42
goalscontrol actions methods metrics measures raw data✦ Cores Frequency
✦ Display Brightness
✦ On Demand DFS✦ Gradual Reduction
✦ Resources Exploitation✦ Power
✦ Resources Utilization✦ Power Model
✦ Average Resource Usage
✦ #Used Resources✦ CPU Utilization✦ System Up-Time✦ Screen Brightness✦ Connections Utilization✦ SD Card Accesses
[5] A. Shye, B. Scholbrock, and G. G. Memik - Into the wild: studying real user activity patterns to guide power optimizations for mobile architectures - International Symposium on Microarchitecture (MICRO), 2009.
M. CARMINATI - A Framework to Model Self-Adaptive Computing Systems June 25, 2013
Into The Wild[5]
43
goalscontrol actions methods metrics measures raw data✦ Cores Frequency
✦ Display Brightness
✦ On Demand DFS✦ Gradual Reduction
✦ Resources Exploitation✦ Power
✦ Resources Utilization✦ Power Model
✦ Average Resource Usage
✦ #Used Resources✦ CPU Utilization✦ System Up-Time✦ Screen Brightness✦ Connections Utilization✦ SD Card Accesses
[5] A. Shye, B. Scholbrock, and G. G. Memik - Into the wild: studying real user activity patterns to guide power optimizations for mobile architectures - International Symposium on Microarchitecture (MICRO), 2009.
Metronome[6]
goals requirementcontrol actions methods metrics measures raw data✦ Virtual Run-Time
✦ Scheduling Policy
✦ Performance✦ Resources Exploitation
✦ Application Progress
✦ Heartrate ✦ Heartbeat✦ Execution Time
[6] F. Sironi, D. B. Bartolini, S. Campanoni, F. Cancare, H. Hoffmann, D. Sciuto, and M. D. Santambrogio - Metronome: operating system level performance management via self-adaptive computing - Design Automation Conference (DAC), 2012.