Digital On-Demand Computing Organism Dod Org SPP OC Kolloquium DFG - - PowerPoint PPT Presentation
Digital On-Demand Computing Organism Dod Org SPP OC Kolloquium DFG SPP 1183 Organic Computing Mnchen, Oktober 7/8, 2010 KIT The cooperation of Forschungszentrum Karlsruhe GmbH and Universitt Karlsruhe (TH) Talk Overview Project
KIT – The cooperation of Forschungszentrum Karlsruhe GmbH and Universität Karlsruhe (TH)
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 2 7.10.2010
Organic Hardware Organic Middleware Organic Low Power Management Organic Middleware
Overview Scenarios
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 3 7.10.2010
Classic Scenario: Only those scenarios can be handled that were considered in advance, where the cause can be detected, where the corresponding reaction had been explicitly programmed. Lack of adaptation leads to insufficient reactions (e.g. shutdown …) DodOrg Scenario:
System reaction based on indications (higher level of abstraction) e.g. CRC/bit error rate, network bottleneck, environmental change or change on application level Proper reaction possible even if Scenario was not considered in advance. Cause was not detected, Reaction was not explicitly programmed. Flexible response to changed
environmental situation
Scenario detection: recognize that something is different Adapt to changed requirements either by known path or gradual process of rearrangement (optimization, healing) Plasticity: Stabilization but not “petrification’’
Demonstrator platform
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 4 7.10.2010
Robustness Extending the stable system property towards more serious system changes . Stability The ability of the system to provide the required service while reacting upon external and internal events.
+ Attack resistance + Fault resistance + Increased tolerance
+ Oscillation avoidance + Normal operating conditions
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 6 7.10.2010
Brain Level Organ Level Cell Level
Myo- cardial Cell Nervous System Application API Application Monitoring Hardware Monitoring Middleware Monitoring, Feedback Application Testbed (all groups)
Organic Middleware (Brinkschulte) Organic Processing Cells (Becker) Distributed Low Power Management (Henkel) Biological Considerations (Brändle) Organic Monitoring System (Karl)
Heart Hormone Level Computation Dynamic Power Management Real-time considerations Temperature, Local Traffic
Proactive Intelligent Data Analysis Self-Adaptation Self-Optimization Self-Healing Stable Hormone Interaction Thermal-aware Energy distribution OPC Extension
Stabiltiy Aspects
Goals
Seamless and transparent expansion of the on chip communication services (artNoC) Dynamic OPC resource pool physical growth of DodOrg organism
Challenges
Physical Connection # I/O Pins Bandwidth Latency Hot Plug /Unplug Control Flow Overhead Dynamic Address Space
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 7 7.10.2010
µProc Cell IO Cell FPGA Cell IO Cell µProc Cell IO Cell IO Cell IO Cell IO Cell System 1 System 2 Transparent Extension
Broadcast Real-time Adaptive routing
µProc Cell IO Cell FPGA Cell IO Cell µProc Cell IO Cell IO Cell IO Cell IO Cell System 1 System 2 Transparent Extension
Broadcast Real-time Adaptive routing
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 8 7.10.2010
38 38 38 38 R R R R R R R R R R R R 38 38 38 38
Chip
2 2 2 2 R R R R R R R R R R R R 2 2 2 2
Chip
Off-Chip Interface
Off-Chip-Interface (OCI) Parallel Serial Asynchronous Transmission
artNoC Handshake Signals artNoC Flit-Data
3 Phase Operation Link Negotiation
Backchannel Detection # Virtual Channels (VCs) # Real-Time Channels
Data Transmission Link Cutback
Terminate open artNoC VCs
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 9 7.10.2010
data(7:0) vc_id(1:0) valid full(3:0) rtc_ready(2:0) bc_grant data(7:0) vc_id(1:0) valid full(3:0) rtc_ready(2:0) bc_grant data(7:0) vc_id(1:0) valid full(3:0) rtc_ready(2:0) bc_grant data(7:0) vc_id(1:0) valid full(3:0) rtc_ready(2:0) bc_grant
1 1 port_out_a port_in_a port_in_b port_out_b OCI_a OCI_b Router_a Router_b Data Transmission: Adaptive Control Flow Protocol
Goal: Increase Serial Link Efficiency Differential Control Flow Transmission
1 VC used: P(Cch ) < 0,1 > 1 VC used: P(Cch ) > 0,5 Amount of Control Flow mainly depends on VC usage
2 Protocol Modes
Differential Mode Auto Sequence Mode
10 20 30 40 50 60 70 80 90 1 VC 2 VC's 50/50 3 VC's 50/20/30 2 VC's 10/90 link efficiency % virtual channel usage Diff Mode Auto Mode
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 10 7.10.2010
Objectives
Classification of the system state with regard to environmental conditions Identification of bottlenecks Determine the outcome of an
Providing these information to Organic Middleware and Thermal Management
Application API Application Monitoring Hardware Monitoring Middleware Monitoring, Feedback Application Testbed (all groups) Organic Middleware (Brinkschulte) Organic Processing Cells (Becker) Distributed Low Power Management (Henkel)
Biological Considerations (Brändle) Organic Monitoring System (Karl)
Hormone Level Computation Dynamic Power Management Real-time considerations Temperature, Local Traffic
Proactive Intelligent Data Analysis Self-Adaptation Self-Optimization Self-Healing Stable Hormone Interaction Stable Energy Distribution OPC Interaction, Metrics
Plasticity Aspects
Application Hardware Monitoring Raw & Cooked Data Feedback Configuration Requirements Status Status Middleware Thermal Management
Flexible, rule-based Approach
Based on the data gathered by our low-level monitoring-infrastructure Evaluation rules are defined at runtime in a dedicated learning phase Rules can be updated at runtime to adapt to new environmental conditions
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 11 7.10.2010
Rule Layout
rule = (t, v, p1, p2, p3) One Rule for each event type
Rule Creation
Determine the normalized
at a predefined time Creation of a histogram Determination of the points p1, p2, p3
Evaluation
Using a simple transfer function (TF) TF converts the ratio into an evaluation score System State is classified using a weighted arithmetic mean of all evaluation scores
Rule Layout
rule = (t, v, p1, p2, p3) One Rule for each event type
Rule Creation
Determine the normalized
at a predefined time Creation of a histogram Determination of the points p1, p2, p3
Evaluation
Using a simple transfer function (TF) TF converts the ratio into an evaluation score System State is classified using a weighted arithmetic mean of all evaluation scores
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010v 12 7.10.2010
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 13 7.10.2010
Phase/ Trend Detection and Prediction
Prediction of future system states Identification of potentially harmful system states in advance
Avoiding Bad System States through Proactivity
Initiating proper system changes to avoid bad or harmful system states (e.g. high temperature or performance bottleneck) Introducing a feedback-loop for on-line evaluation of the system changes Proactive self-healing and self-optimization Past System States Current System State Predicted System States Avoid this state
Organic Low Power Management: Managing Energy-Distribution (Prof. Henkel)
Cost Function
Organic Middleware
Influencing Hormone Expression
Power / RT Manager
Organic Monitoring
Consume Fade Trade & Negotiate
Policy
Energy Budget Manager Local Energy Budget
P4 P3 P2 P1
Fill
Energy Input
Efficiency RT criteria Temperature Local traffic
Power source Peers (in neighbored OPCs) OPC
Voltage / frequency setting Power States Assigned Tasks Scheduled Tasks
data / information actions
Legend:
Future energy level Actual energy level Energy level Actual power state
Energy Distribution: goals
Low energy consumption Avoidance of local thermal hot-spots
Energy Distribution: main concept
Each OPC has a Local Energy Budget Determines the local available energy Global Power Source Assigns energy budgets to OPCs (pulse-based) Energy Budget Manager Agent controlling Local Energy Budget Receives temperature data each pulse Negotiates & Trades energy budget with neighboring OPCs Influences Power Manager policies
7.10.2010 14 SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010
Organic Low Power Management:Thermal Management (Prof. Henkel) Thermal simulation of energy distribution Agent negotiation based on simple economic principles Energy budget made up of Energy Units Trading based on supply & demand Temperature incorporated as a negotiation penalty Agents only trade with their direct neighbors One Energy Unit per pulse Energy Units propagate across the chip over time
OPC OPC OPC
Agent Agent AgentThermal-aware power negotiation
7.10.2010 15 SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010
Organic Low Power Management: Outlook (Prof. Henkel) Current approach succeeds in keeping temperature below a threshold, however the effectiveness of agent trading can still be improved Improvement of negotiation process using economic learning Organic state classification as input Trading multiple energy units at a time will decrease propagation time Considerable improvement for large chips (hundreds of cores) Better trading between different thermal domains (i.e. physically separated chips)
7.10.2010 16 SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 10 20 30 40 50 60 1 581 1161 1741 2321 2901 3481 4061 4641 5221 5801 6381 6961 7541 8121 8701 9281 9861 10441 11021 11601 12181 12761 13341 13921
Time Interval (ms) Temperature (°C) Temperature below threshold
Organic Middleware: Re-Intruduction of the Artificial Hormone System (Prof. Brinkschulte)
OPC OPC OPC OPC OPC OPC OPC OPC
Task Mapping Providing Self-X Properties: Self-Configuration Self-Healing Self-Optimization Good mapping regarding Requirements of each task Relationships of the tasks Condition of each cell and it’s neighborhood Reacting and Adapting to changes (plasticity) e.g. increased bit-rate errors Reaching stable mapping conditions Oscillation avoidance
7.10.2010 17 SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010
Organic Middleware: Implementation of the AHS for µCs (Prof. Brinkschulte)
7.10.2010 18 SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010
AHS Interface AHS Task Management AHS Error Management AHS Message Communication AHS List Management AHS Hormone Communication AHS Log Management AHS Basic OS Support AHS Basic Communication Support
Operating System Communication System Distributed Application
Pure ANSI C for deployment in environments from small µCs to large PCs 7729 Total Physical Source Lines of Code (SLOC) Simple exchange of underlying OS due to the “AHS Basic OS Support” abstraction layer Network protocol easily interchangeable due to “AHS Basic Communication Support” interface
Organic Middleware: Outlook (Prof. Brinkschulte)
Prototype implementation: Analyzing the effects on system stability with the influences of: Power Manager Monitoring Setting boundaries for the control influences Conflict avoidance through proactivity “Immune System” for the AHS: Immune mechanisms for advanced Self-Healing and Self-Protecting Aspects to increase Robustness (together with the Organic Processing Cell Capabilities and the Organic Monitoring) Robustness against mal-behaving internal/external components (comparable to illness in a biological system) Being able to react to „ill“ OPCs Counter-measures against malicious attacks
7.10.2010 19 SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010
OPC Network-Interface
Monitor
Power- Manager Config- Manager
MicroBlaze µC
AHS
ware
OPC
Config- Manager Power- Manager Monitor Network- Interface artNoC- Router DCM clk_dp M B M B data path MicroBlaze-µC with Middleware FSL FSL FSL 8 8 8 Ring Oscilator Temperature Sensor DCM clk_net Config- Manager Power- Manager Monitor Network- Interface artNoC- Router DCM clk_dp M B M B data path MicroBlaze-µC with Middleware FSL FSL FSL 8 8 8 Ring Oscilator Temperature Sensor Config- Manager Power- Manager Monitor Network- Interface artNoC- Router DCM clk_dp M B M B data path MicroBlaze-µC with Middleware FSL FSL FSL 8 8 8 Ring Oscilator Temperature Sensor Config- Manager Power- Manager Monitor Network- Interface artNoC- Router DCM clk_dp M B M B data path MicroBlaze-µC with Middleware FSL FSL FSL 8 8 8 Ring Oscilator Temperature Sensor
Prototype: Interaction of HW, Mon, MW and TM
7.10.2010 20 SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010
HW-Interfaces: Networking via HW- Interface Energy Budget from power management System Status and Estimations of Future Changes from monitor Develop Protocols: Infos Ranges Packet Formats etc
OPC
Config- Manager Power- Manager Monitor Network- Interface artNoC- Router DCM clk_dp M B M B data path MicroBlaze-µC with Middleware FSL FSL FSL 8 8 8 Ring Oscilator Temperature Sensor DCM clk_net Config- Manager Power- Manager Monitor Network- Interface artNoC- Router DCM clk_dp M B M B data path MicroBlaze-µC with Middleware FSL FSL FSL 8 8 8 Ring Oscilator Temperature Sensor Config- Manager Power- Manager Monitor Network- Interface artNoC- Router DCM clk_dp M B M B data path MicroBlaze-µC with Middleware FSL FSL FSL 8 8 8 Ring Oscilator Temperature Sensor Config- Manager Power- Manager Monitor Network- Interface artNoC- Router DCM clk_dp M B M B data path MicroBlaze-µC with Middleware FSL FSL FSL 8 8 8 Ring Oscilator Temperature Sensor OPC
Config- Manager Power- Manager Monitor Network- Interface artNoC- Router DCM clk_dp M B M B data path MicroBlaze-µC with Middleware FSL FSL FSL 8 8 8 Ring Oscilator Temperature Sensor DCM clk_net Config- Manager Power- Manager Monitor Network- Interface artNoC- Router DCM clk_dp M B M B data path MicroBlaze-µC with Middleware FSL FSL FSL 8 8 8 Ring Oscilator Temperature Sensor Config- Manager Power- Manager Monitor Network- Interface artNoC- Router DCM clk_dp M B M B data path MicroBlaze-µC with Middleware FSL FSL FSL 8 8 8 Ring Oscilator Temperature Sensor Config- Manager Power- Manager Monitor Network- Interface artNoC- Router DCM clk_dp M B M B data path MicroBlaze-µC with Middleware FSL FSL FSL 8 8 8 Ring Oscilator Temperature Sensor
Demonstrator Hardware Floorplan
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 21 7.10.2010
Microblaze OPC Empty- OPC I/O- OPC
DodOrg Demonstrator Scenarios
OPC OPC OPC OPC OPC OPC OPC OPC
Providing Self-X Properties: Self-Configuration Self-Healing Self-Optimization
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 22 7.10.2010
DodOrg Demonstrator Scenarios
OPC OPC OPC OPC OPC OPC OPC OPC
Providing Self-X Properties: Self-Configuration Self-Healing Self-Optimization
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 23 7.10.2010
Providing Self-X Properties: Self-Configuration Self-Healing Self-Optimization DodOrg Demonstrator Scenarios
OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 24 7.10.2010
Providing Self-X Properties: Self-Configuration Self-Healing Self-Optimization DodOrg Demonstrator Scenarios
OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC OPC
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 22 7.10.2010
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 25 7.10.2010
Working demonstrator
the interior of what is happening while the system is running Visualisation of the OPC: Activity Monitoring Info Temp and Energy Budget Running Tasks Cycles Bandwidth MW sends infos via serial interface to an PC where the data is visualized
SPP 1183 OC Kolloquim – München, 7.-8. Oktober 2010 26 7.10.2010
Concepts individually tested and applicability proven Monitoring: hormone-inspired associative event coding and use of associative counters Middleware: reaching stable hormone and mapping situations while still being able to react to changes Low-Power-Processing: thermal-aware local agent-based energy budget distribution Processing Cells: Growth of the OC-hardware platform through adaptive chip to chip communication interface Incorporation of the DodOrg subsystems into demonstrator platform
21.09.2009 28 SPP 1183 OC Kolloquim – Augsburg, 21.-22. September 2009
Application Testbed (all groups) Organic Middleware (Brinkschulte) Organic Processing Cells (Becker) Organic Low Power Management (Henkel)
Organic Monitoring System (Karl)
List of Publications:
21.09.2009 29 SPP 1183 OC Kolloquim – Augsburg, 21.-22. September 2009
ACACES,2009
IT - Information Technology Journal 5/08, 2008
BICC08, 2008
Environment,“ ARCS, 2008
IWSOS, 2006
Networks“, CASES, 2006
2009
middleware”, ISORC, 2008
Allocation”, in Organic Computing, 2007
2007
Architectures”, ICCAD 2009
Dynamic Power-Gated Instruction Set” , ICCAD 2009
Communication", DAC 2008
Organic Computing“, ReCoSoC, 2008
IPDPS, 2008
2007