Modeling and Analysis of Data Flow Graphs using the Digraph - PowerPoint PPT Presentation

Modeling and Analysis of Data Flow Graphs using the Digraph Real-Time Task Model Morteza Mohaqeqi, Jakaria Abdullah, and Wang Yi Uppsala University Ada-Europe 2016

Overview Introduction Data Flow Graphs: [ 1 , 3 ] [ 2 ] Signal processing a b Stream processing [ 2 ] [ 1 , 1 , 4 ] Data dependency Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 1 / 19

Overview Introduction Data Flow Graphs: [ 1 , 3 ] [ 2 ] Signal processing a b Stream processing [ 2 ] [ 1 , 1 , 4 ] Data dependency Design Objectives Throughput maximization Design Constraints Buffer overflow/underflow avoidance Schedulability Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 1 / 19

Overview An Overview The Problem: Our Approach: A Set of Data A Set of Data Flow Graphs Flow Graphs Transform Schedule Real-Time Tasks Processor Schedule Processor Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 2 / 19

Overview An Overview The Problem: Our Approach: A Set of Data A Set of Data Flow Graphs Flow Graphs Transform Schedule Real-Time Tasks Periodic Processor Schedule Processor Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 2 / 19

Overview An Overview The Problem: Our Approach: A Set of Data A Set of Data Flow Graphs Flow Graphs Transform Schedule Real-Time Tasks Digraph RT Processor Schedule Processor Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 2 / 19

Overview Previous Work a b periodic periodic DRT periodic periodic task task Our work Our work Strictly Period Schedule Strictly period schedule Digraph Real-Time (DRT) Digraph Real-Time (DRT) [Bouakaz 2014] [Bouakaz 2014] task model [Stigge et al 2011] task model [Stigge et al 2013] Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 3 / 19

Outline Introduction 1 Method 2 Evaluation 3 Conclusion 4 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 4 / 19

Introduction (Static) Data Flow Graphs Synchronous Cyclo-Static Data Flow Data Flow [ 2 ] [ 1 ] [ 1 , 3 ] [ 2 ] a b a b [ 2 ] [ 1 , 1 , 4 ] Fixed token production Variable token production (consumption) rate (consumption) rate Fixed execution time Variable execution time Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 5 / 19

Introduction Semantics Empty buffer [ 2 ] [ 1 ] a b size = 2 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 6 / 19

Introduction Semantics Empty buffer ’ a ’ can be fired [ 2 ] [ 1 ] a a b size = 2 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 6 / 19

Introduction Semantics Full buffer [ 2 ] [ 1 ] a b size = 2 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 6 / 19

Introduction Semantics Full buffer ’ a ’ cannot be fired ’ b ’ can be fired [ 2 ] [ 1 ] a b b size = 2 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 6 / 19

Introduction Semantics ’ a ’ cannot be fired ’ b ’ can be fired [ 2 ] [ 1 ] a b b size = 2 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 6 / 19

Introduction Semantics Empty buffer ’ a ’ can be fired [ 2 ] [ 1 ] a a b size = 2 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 6 / 19

Introduction Constraints Design Constraints Underflow/overflow avoidance Schedulability Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 7 / 19

Introduction Underflow and Overflow Underflow avoidance [ 2 ] [ 1 ] a b T a t T b t 0 5 10 Cannot be fired Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 8 / 19

Introduction Underflow and Overflow Underflow avoidance [ 2 ] [ 1 ] a b T a t offset T b t 0 5 10 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 8 / 19

Introduction Underflow and Overflow Produce token as soon as possible Consume token as late as possible [ 2 ] [ 1 ] a b T a t offset T b t 0 5 10 tokens t 0 5 10 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 8 / 19

Introduction Underflow and Overflow Produce token as soon as possible Consume token as late as possible [ 2 ] [ 1 ] a b T a t offset T b t 0 5 10 tokens buffer size=5 no overflow t 0 5 10 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 8 / 19

Introduction Constraints Design Constraints Underflow/overflow avoidance Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 9 / 19

Introduction Constraints Design Constraints Underflow/overflow avoidance Schedulability Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 9 / 19

Introduction Constraints Design Constraints Underflow/overflow avoidance Schedulability [ 2 ] [ 1 ] a b C a = 2 T a t offset C b = 2 T b t 0 5 10 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 9 / 19

Introduction Constraints Design Constraints Underflow/overflow avoidance Schedulability [ 2 ] [ 1 ] a b C a = 2 T a t offset C b = 2 T b t 0 5 10 Unschedulable! Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 9 / 19

Introduction Constraints Design Constraints Underflow/overflow avoidance Schedulability [ 2 ] [ 1 ] a b C a = 2 T a t offset C b = 2 T b t 0 5 10 Larger Periods Lower Throughput Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 9 / 19

Introduction The Problem Design Parameters Periods Offsets Constraints No underflow No overflow Schedulability Objective Throughput maximization Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 10 / 19

Introduction Cyclo-Static Data Flow Graphs Repeating pattern [ 10 , 10 , 0 , 0 ] [ 5 , 5 , 5 , 5 ] a b 10 10 0 0 10 10 0 0 10 10 T a t 5 5 5 5 5 5 5 5 T b t 0 5 10 15 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 11 / 19

Introduction Cyclo-Static Data Flow Graphs Repeating pattern [ 10 , 10 , 0 , 0 ] [ 5 , 5 , 5 , 5 ] a b 10 10 0 0 10 10 0 0 10 10 T a t 5 5 5 5 5 5 5 5 T b t 0 5 10 15 tokens size=25 25 20 15 10 5 t 0 5 10 15 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 11 / 19

Introduction Cyclo-Static Data Flow Graphs Repeating pattern [ 10 , 10 , 0 , 0 ] [ 5 , 5 , 5 , 5 ] a b 10 10 0 0 10 10 0 0 10 deadline T a t 5 5 5 5 5 5 5 5 T b t 0 5 10 15 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 11 / 19

Introduction Cyclo-Static Data Flow Graphs Repeating pattern [ 10 , 10 , 0 , 0 ] [ 5 , 5 , 5 , 5 ] a b 10 10 0 0 10 10 0 0 10 deadline T a t 5 5 5 5 5 5 5 5 T b t 0 5 10 15 tokens 25 size=20 20 15 10 5 t 0 5 10 15 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 11 / 19

Introduction We need a non-periodic task model Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 12 / 19

Introduction Scheduling Data Flow Graphs Synchronous Data Flow Fixed behavior Periodically repeating [ 2 ] [ 1 ] a b T a t 0 4 8 12 T b t 0 4 8 12 offset Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 13 / 19

Introduction Scheduling Data Flow Graphs Synchronous Data Flow Cyclo-Static Data Flow Changing behavior Fixed behavior Periodically repeating Repeating pattern [ 1 , 3 ] [ 2 ] [ 2 ] [ 1 ] a b a b [ 2 ] [ 1 , 1 , 4 ] T a t 0 4 8 12 T b t 0 4 8 12 offset Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 13 / 19

Introduction Scheduling Data Flow Graphs Synchronous Data Flow Cyclo-Static Data Flow Changing behavior Fixed behavior Periodically repeating Repeating pattern [ 1 , 3 ] [ 2 ] [ 2 ] [ 1 ] a b a b [ 2 ] [ 1 , 1 , 4 ] T a t 0 4 8 12 The Digraph Real- T b t Time task model. 0 4 8 12 offset Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 13 / 19

Introduction The Digraph Real-Time (DRT) Task Model A graph-based representation Different job types 8 v 1 v 2 � 2 , 5 � � 1 , 3 � 10 8 5 v 3 � 1 , 6 � 10 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 14 / 19

Introduction The Digraph Real-Time (DRT) Task Model A graph-based representation Different job types 8 v 1 v 2 � 2 , 5 � � 1 , 3 � 10 8 5 v 3 � 1 , 6 � 10 v 1 v 2 v 3 v 1 v 1 v 2 v 3 v 1 t 0 5 10 15 20 Mohaqeqi et al. (Uppsala University) Analysis of Data Flow Graphs 14 / 19

Modeling and Analysis of Data Flow Graphs using the Digraph - PowerPoint PPT Presentation

Modeling and Analysis of Data Flow Graphs using the Digraph Real-Time Task Model Morteza Mohaqeqi, Jakaria Abdullah, and Wang Yi Uppsala University Ada-Europe 2016 Overview Introduction Data Flow Graphs: [ 1 , 3 ] [ 2 ] Signal processing a

HW/SW Codesign w/ FPGAs Data Flow Modeling III ECE 522 DFG Performance Modeling and

HW/SW Codesign w/ FPGAs Data Flow Modeling II ECE 522 Synchronous Data Flow Graphs Synchronous

1 Liveness by Example (cont) Control Flow Graphs (CFGs) Live range of a Definition a is live

1 What Is Control-Flow Analysis? Loop Concepts Control-flow analysis discovers the flow of

Introduction to Data-flow analysis Last Time Implementing a Mark and Sweep GC Today

Flow Analysis Data-flow analysis, Control-flow analysis, Abstract interpretation, AAM Helpful

1 Ways to improve data-flow analysis efficiency Example (liveness) 1st 2nd 3rd

Data-flow analysis Introduction to data-flow analysis Michel Schinz based on material by

!System(on(Chip!Design! Data!Flow!Modeling! (Based!on!slides!at!ECE!522!at!UNM)! Hao$Zheng$

Test%Driven+Data+Modeling+With+Graphs + Twi7er:+@ianSrobinson+ #neo4j+ + Outline+

HW/SW Codesign w/ FPGAs Data Flow Modeling I ECE 522 Data Flow Models As we discussed, hardware

Inductive Data Flow Graphs Azadeh Farzan 1 Zachary Kincaid 1 Andreas Podelski 2 1 University of

An Interactive Web Based Platform for Modeling and Analysis of Large Scale Argus Network Flow

Analysis and simulation of a two phase flow F. Sma , A. Bourgeat, M. Jurak model with phase

Diagnostic Information for Control-Flow Analysis of Workflow Graphs (aka Free-Choice Workflow Nets)

CMSC 430 Introduction to Compilers Spring 2016 Data Flow Analysis Data Flow Analysis A

Verification and Qualitative analysis of hybrid systems: dimension 2 Joint work with Gerardo

Distributed Synthesis for LTL Fragments Krishnendu Chatterjee, Thomas A. Henzinger, Jan Otop ,

Review of 136, Preview of 186 Peter Cappello Department of Computer Science University of

Design and Analysis of Algorithms

Composition study of ultra-high-energy cosmic rays based on the Telescope Array surface detector

Vertex reconstruction Vertex reconstruction in large liquid scintillator detectors in large

FPChecker Detecting Floating-Point Exceptions in GPUs Ignacio Laguna, Harshitha Menon, Tristan

Measurements of ASIC V2. M.Borri STFC Measurements in h-mode. Overview. Test01 GOAL: