Future Directions on Models of Architecture Maxime Pelcat INSA - - PowerPoint PPT Presentation

▶

Oct 02, 2022 177 likes •410 views

Dataflow workshop - 2017 Future Directions on Models of Architecture Maxime Pelcat INSA Rennes, Institut Pascal System Design: Y-Chart Application Algorithm Architecture Redesign Redesign Design System Prototype Maxime Pelcat HDR -

SLIDE 1

Future Directions on Models of Architecture

Dataflow workshop - 2017 Maxime Pelcat INSA Rennes, Institut Pascal

SLIDE 2

System Prototype

System Design: Y-Chart

Maxime Pelcat – HDR - 2017 2

Architecture Design Algorithm Application Redesign Redesign

SLIDE 3

System Objectives

Maxime Pelcat – HDR - 2017 3

T°C Energy Reliability Memory Unit Cost

$

Security Maintenance Cost

$

Performance Peak Power

SLIDE 4

Model of Architecture (MoA) conform to

Model-Based Design

KPI Architecture Model KPI Evaluation Algorithm Algorithm Model Redesign

Maxime Pelcat – HDR - 2017

Model of Computation(MoC) conforms to Redesign

SLIDE 5

Models of Architecture

Maxime Pelcat – HDR - 2017 5

Model of Architecture (MoA) conform to KPI Architecture Model KPI Evaluation Algorithm Algorithm Model Redesign Redesign

SLIDE 6

MoC is not sufficient

Energy Energy Evaluation Algorithm Algorithm Model

Maxime Pelcat – HDR - 2017

Model of Computation(MoC) conforms to

SLIDE 7

Problem: Predict System Quality

How to predict a DSP system quality ?

–Efficiently (simple procedure) –Early (from abstract models) –Accurately (with a good fidelity) –With reproducibility (same models = same prediction)

7 Maxime Pelcat – HDR - 2017

SLIDE 8

Model of Architecture

Definition

–Model of a system Non-Functional Property –Application-independent –Abstract –Reproducible

8 Maxime Pelcat – HDR - 2017

Pelcat, M; Mercat, A; Desnos, K; Maggiani, L; Liu, Y; Heulot, J; Nezan, J-F; Hamidouche, W; Ménard, D; Bhattacharyya, S (2017) "Reproducible Evaluation of System Efficiency with a Model of Architecture: From Theory to Practice", IEEE TCAD. Pelcat, M (2018) “Models of Architecture for DSP Systems", Handbook of Signal Processing Systems, Third Edition, S. S. Bhattacharyya, E. F. Deprettere, R. Leupers , J. Takala, Springer.

SLIDE 9

Model of Architecture

Maxime Pelcat – HDR - 2017 9

Model Reproducible Application- independent Abstract AADL

  

MCA SHIM

  

UML MARTE

 / 

AAA

  

CHARMED

  

S-LAM

  

MAPS

  

LSLA

  

SLIDE 10

NFP = MoA( ) activity( )

MoA depends on MoC

Model of Architecture

One and always the same quality evaluation Model H conforms to MoA Model G conforms to MoC Activity

MoC( )

Maxime Pelcat – HDR - 2017

application

Performance Power Energy Memory T°C Reliability Security Cost

SLIDE 11

Model of Architecture

KPI MoA MoC Act

Maxime Pelcat – HDR - 2017

SLIDE 12

LSLA: First MoA

LSLA = Linear System-Level Architecture

Model

Motivated by the additive nature of energy

consumption

Maxime Pelcat – HDR - 2017 12

SLIDE 13

LSLA Model of Architecture

Task1 signal signal Task2 Task3 Task4 Task5 1 1 1 1 1 1 1

PE1 PE2

10x+1 2x+0 3x+0

16+12+22=50

Maxime Pelcat – HDR - 2017

token quantum Compositional

SLIDE 14

LSLA Model of Architecture

Task1 signal signal Task2 Task3 Task4 Task5 1 1 1 1 1 1 1

PE1 PE2

10x+1 2x+0 3x+0

16+12+22=50

Maxime Pelcat – HDR - 2017

SDF: Model of Computation Activity LSLA: Model of Architecture

SLIDE 15

LSLA MoA for Energy Prediction

86% of fidelity on octo-core ARM 

15 Maxime Pelcat – HDR - 2017

SLIDE 16

LSLA MoA for Energy Prediction

The model is learnt from energy

measurements

PE PE

PE PE PE PE

PE PE

Maxime Pelcat – HDR - 2017

SLIDE 17

LSLA MoA for Energy Prediction

The model is learnt from energy

measurements

PE PE

α 1.5W 1.5W PE PE 1.5W 1.5W PE PE

γ 0.3W 0.3W PE PE 0.3W 0.3w

Maxime Pelcat – HDR - 2017

SLIDE 18

MoAs: Limits of LSLA

Energy

 Linear model OK

Latency
Latency does not have an additive nature



Maxime Pelcat – HDR - 2017

Task1 Task2 1 1 1 Task1 Task2 1 1 1 1

Latency = sum Latency = max

!

SLIDE 19

Activity & MoA for Latency

Task1 signal signal Task2 Task3 Task4 Task5 1 1 1 1 1 1 1

SDF a) b)

Maxime Pelcat – HDR - 2017

SLIDE 20

Activity & MoA for Latency

PE1 PE2

10x+1 2x+0 3x+0

Σ  12+12+11=35 Σ 8+6+11=25 max(35,25)=35 a) b)

Maxime Pelcat – HDR - 2017

MaxPlus

SLIDE 21

System Prototype

Accuracy?? No! Fidelity!!

Maxime Pelcat – HDR - 2017 21

Architecture Design Algorithm Application Redesign Redesign

SLIDE 22

Directions for Research on MoA

Try existing models on new KPIs
Create new models for new KPIs

–When existing ones do not match

Co-explore MoAs - multi-objective optim.
Learn more complex, non-linear models

Maxime Pelcat – HDR - 2017 22

KPI MoA MoC Act