Analysing gCSP Models Using Runtime and Model Analysis Algorithms - - PowerPoint PPT Presentation

▶
analysing gcsp models using runtime and model analysis
SMART_READER_LITE
LIVE PREVIEW

Analysing gCSP Models Using Runtime and Model Analysis Algorithms - - PowerPoint PPT Presentation

Nov 03, 2022 250 likes •553 views

Analysing gCSP Models Using Runtime and Model Analysis Algorithms Communicating Process Architectures 2009 Maarten Bezemer, Marcel Groothuis and Jan Broenink Control Engineering, University of Twente, The Netherlands Contents Introduction

slide-1
SLIDE 1

Analysing gCSP Models Using Runtime and Model Analysis Algorithms

Communicating Process Architectures 2009 Maarten Bezemer, Marcel Groothuis and Jan Broenink

Control Engineering, University of Twente, The Netherlands

slide-2
SLIDE 2

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 2

Contents

  • Introduction
  • Runtime Analysis Algorithm
  • Model Analysis Algorithm
  • Conclusions
slide-3
SLIDE 3

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 3

  • CSP usage at Control Engineering
  • Modelling tool → gCSP

Introduction

Par4 = REP_P || REP_C REP_C = Seq_C; REP_C Seq_C = Consumer1; Consumer2 Consumer1 = C1_Rd; C1_C Consumer2 = C2_Rd; C2_C REP_P = Seq_P; REP_P Seq_P = Producer1; Producer2 Producer1 = P1_C; P1_Wr Producer2 = P2_C; P2_Wr

slide-4
SLIDE 4

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 4

  • CSP usage at Control Engineering
  • Modelling tool → gCSP
  • Code generation for (robotic) controllers
  • Using Communicating Threads (CT) library
  • Debugging possibilities while running the code
  • Animating the model (processes and channels)
  • Stepping through model, while showing channel values

Introduction

slide-5
SLIDE 5

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 5

Introduction Problem

  • Designer Point of View
  • Detailed modelling
  • Lots of small processes
  • Executing Point of View
  • Fast code
  • A few bigger processes
  • Both Points of View conflict!
slide-6
SLIDE 6

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 6

Introduction Solution

  • Translate Designer PoV to Executing PoV
  • Requires
  • Analysis of the gCSP model
  • Model transformation
  • Solution: two analysis algorithms
  • Runtime analysis for static ordering of processes
  • Model analysis for process scheduling
slide-7
SLIDE 7

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 7

Contents

  • Introduction
  • Runtime Analysis Algorithm
  • Introduction
  • Algorithms
  • Results
  • Model Analysis Algorithm
  • Conclusions
slide-8
SLIDE 8

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 8

Runtime Analysis Algorithm

  • Why static ordering of processes
  • No complex scheduler required
  • Possibility for grouping of processes
  • Goal of Runtime Analysis Algorithm
  • Find a static running order for the processes
slide-9
SLIDE 9

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 9

Runtime Analysis Algorithm Processes

  • Process states
  • New

Process is created

  • Ready

Process is ready to be started

  • Running

Process is started and still running

  • Blocked

Process is blocked

  • Finished

Process is ended

  • Algorithm mainly uses Finished state to determine the

static running order

slide-10
SLIDE 10

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 10

Runtime Analysis Algorithm Notations

  • Set of chains
  • Clear view of groups of processes
  • Cross-Reference types
  • To other chain
  • To start of same chain
  • Comparable with a CSP Trace
  • Traces
  • Finished processes of running model
  • For demonstration purposes

D → C → F → B → (B,D*) B → E → A → C → (D) [start] → A → B → C → D → (B) A → B → C → D → B → E → A → C → E → A → C → D → C → F →…

slide-11
SLIDE 11

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 11

[start]

Runtime Analysis Algorithm Algorithm

  • Process Ordering Rules
  • Chains with no cross-refs (the active chain is not finished yet)
  • Add processes to chain
  • Rules
  • If the state of a process changes to Finished add it to the end
  • f the active chain.
  • If a process is Finished and is already present in the active

chain, it will become a cross-reference of this chain pointing to a chain starting with this process.

A → B → C → B → D → E → F → B [start] → A → B → C B [start] → A → B → C → (B) B → D → E → F → (B*) [start] → A → B → C → (B)

slide-12
SLIDE 12

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 12

Runtime Analysis Algorithm Algorithm

  • Process Ordering Rules
  • Chains with cross-refs (the active chain got finished already)
  • Perform validation on chains
  • Rules
  • If the active process does not match the Finished process the

chain must be split.

… → D → E → F → B → D → G → H B → D → E → F → (B*) [start] → A → B → C → (B) G E → F → (B) B → D → (E,G) [start] → A → B → C → (B) G → H E → F → (B) B → D → (E,G) [start] → A → B → C → (B) Rules for ‘chains with no cross-references’ applied

slide-13
SLIDE 13

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 13

Runtime Analysis Algorithm Results

  • Set of chains as expected
  • All processes could be placed in one big process
  • Writer-Reader combinations can be removed
  • Channels become internal variables

P1_C → C1_Rd → C1_C → P1_Wr → P2_C → P2_Wr → C2_Rd → C2_C →(P1_C*) [start]→(P1_C)

slide-14
SLIDE 14

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 14

Runtime Analysis Algorithm Results

  • Scalability of the algorithm
  • Complex traces
  • Hard to verify results
  • Static ordering available
  • Working Cartesian plotter model
slide-15
SLIDE 15

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 15

Runtime Analysis Algorithm Results

  • Working Cartesian plotter model

1 2 3 4 →

Sc_Rd8 → DoubletoBooleanConversion → Sc_Wr17 → Sa_Wr8 → Sa_Rd4 → MC_Wr4 → Sa_Rd7 → MC_Wr7 → Sa_Rd6 → MC_Wr6 → Sa_Rd5 → MC_Wr5 → Sa_Rd_ESX2_2 → Sa_Rd_ESX2_1 → Sa_Rd_ESX1_2 → Sa_Rd_ESX1_1 → MC_Rd12 → MC_Rd13 → Safety_X → Sa_Rd_ESY1 → Sa_Rd_ESY2 → Safety_Y → Safety_Z → MC_Rd1 → MS_Wr1 → MC_Rd2 → MS_Wr2 → Sa_Wr9 → Sc_Rd9 → MC_Rd3 → LongtoDoubleConversion → Controller → MS_Wr3 → Sc_Rd10 → Sa_Wr10 → (Sc_Rd11) Sc_Rd11 → DoubletoShortConversion → Sc_Wr14 → Sc_Wr15 → Sc_Wr16 → Sa_Wr11 → (Sc_Rd8, HPGLParser) MC_Rd12 → MC_Rd13 → Sa_Rd_ESX2_2 → Sa_Rd_ESX2_1 → Sa_Rd_ESX1_2 → Sa_Rd_ESX1_1 → MC_Rd1 → MS_Wr1 → Safety_X → Sa_Rd_ESY1 → Sa_Rd_ESY2 → Safety_Y → Safety_Z → MC_Rd2 → MS_Wr2 → MC_Rd3 → LongtoDoubleConversion → Controller → MS_Wr3 → Sa_Wr9 → Sc_Rd9 → Sc_Rd10 → Sa_Wr10 → (HPGLParser) HPGLParser → (MC_Rd12, Sc_Rd11) [start] → MC_Rd12 → MC_Rd13 → HPGLParser → MS_Wr1 → MC_Rd1 → MC_Rd2 → MS_Wr2 → MC_Rd3 → LongtoDoubleConversion → Controller → MS_Wr3 → MC_Wr5 → Sa_Rd5 → MC_Wr6 → Sa_Rd6 → MC_Wr7 → Sa_Rd7 → MC_Wr4 → Sa_Rd4 → (HPGLParser)

slide-16
SLIDE 16

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 16

Contents

  • Introduction
  • Runtime Analysis Algorithm
  • Model Analysis Algorithm
  • Introduction
  • Algorithm
  • Results
  • Conclusions
slide-17
SLIDE 17

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 17

Model Analysis Algorithm

  • More towards model analysis/ model transformation
  • What processes are related?
  • How to schedule large models onto a target system?
  • Goal
  • Schedule processes on cores/ networked nodes
slide-18
SLIDE 18

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 18

Model Analysis Algorithm Architecture

  • Algorithm Architecture
  • Build modular
  • gCSP model & User Interface feed the algorithm with data
  • Process weights (or execution times)
  • Available cores or networked nodes
  • Communication (setup) time
slide-19
SLIDE 19

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 19

Model Analysis Algorithm Algorithms

  • Model Tree Creator
  • Recreates the model tree
  • Only for displaying purposes for the user interface
  • Dependency Graph Creator
  • Finds dependencies between processes
  • Sequential relations
  • Channels
  • Critical Path Creator
  • Finds the critical path using the

dependencies

slide-20
SLIDE 20

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 20

Model Analysis Algorithm Heaps

  • Heaps
  • Groups of ‘related’ processes
  • Influenced by
  • Process weight
  • Communication (setup) time
  • Reduce complexity of core scheduler
  • Index blocks
  • Subdivision of heaps
  • When multiple outgoing

dependencies are available

slide-21
SLIDE 21

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 21

Model Analysis Algorithm Cores

  • Cores
  • Groups of processes to be scheduled on the

same core/ networked node

  • Find optimum for end time
  • Amount of cores
  • Relative speed of cores
slide-22
SLIDE 22

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 22

Model Analysis Algorithm Results

  • The processes are optimally scheduled
  • For the given process weights
  • For the given communication times
  • For the given target systems (mostly)
  • Scalable for models of real-life setups
  • Cartesian plotter model
  • Production cell model
  • 597 Processes
  • 210 Heaps
  • ~50 Cores for optimal ending time
slide-23
SLIDE 23

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 23

Model Analysis Algorithm Results

slide-24
SLIDE 24

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 24

Contents

  • Introduction
  • Runtime Analysis Algorithm
  • Model Analysis Algorithm
  • Conclusions
slide-25
SLIDE 25

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 25

Conclusions

  • Both analysis algorithms work as expected
  • Functional
  • Scale well
  • Both analysis algorithms complement each other
  • Runtime analysis algorithm
  • Groups processes into bigger processes
  • Reduces the amount of context switches
  • Model analysis algorithm
  • Schedules processes onto multiple cores
  • Reduces the amount of network communication
  • Both reduce execution time
  • Without losing concurrency aspects
slide-26
SLIDE 26

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 26

Recommendations

  • Refinements are necessary
  • Better representation of the results
  • Include more CSP constructs
  • Support for allocating specific processes on a core
  • Better support for networked nodes
  • Next steps
  • Include model transformations after the analysis phase
  • Implement the algorithms in the gCSP2 tool
slide-27
SLIDE 27

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 27

slide-28
SLIDE 28

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 28

Runtime Analysis Algorithm Algorithm

  • The active chain should be split at process B when process p

is unexpected, but a chain starting with process B is present.

  • Compare the processes after B with the chain starting with

process B → equal!

  • Remove the remaining process (E) in the active chain starting at

process B.

  • Add the cross-references (G) to the chain starting with B if they

are not present at this chain.

  • Create a new chain starting with I and make it the active chain.

A → B → C → B → D → E → F → B → D → G → H → B → D → G → H → I I G → H → (B, I) E → F → (B) B → D → (E,G) A → B → C → (B) G → H → B → D → (G*) E → F → (B) B → D → (E,G) A → B → C → (B)

slide-29
SLIDE 29

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 29

Cartesian plotter

  • 56 processes
  • 10 heaps
  • ~4 cores
  • 1 core is almost optimal
slide-30
SLIDE 30

02-11-2009 Analysing gCSP Models Using Runtime and Model Analysis Algorithms 30

Ongoing Work

  • gCSP2
  • Eclipse based
  • Much more stable

compared to gCSP