43
POLITECNICO DI MILANO High Performance Processors and Systems PdM – UIC joint master 2007 PdM – UIC joint master 2007 Instructor: Prof. Donatella Sciuto Instructor: Prof. Donatella Sciuto HPPS @ PdM – March 2007 HPPS @ PdM – March 2007

HPPS 2007 Projects Presentation

Embed Size (px)

DESCRIPTION

HPPS 2007 Projects Presentation

Citation preview

Page 1: HPPS 2007 Projects Presentation

POLITECNICO DI MILANO

High Performance Processors and

Systems PdM – UIC joint master 2007PdM – UIC joint master 2007

Instructor: Prof. Donatella SciutoInstructor: Prof. Donatella Sciuto

HPPS @ PdM – March 2007HPPS @ PdM – March 2007

Page 2: HPPS 2007 Projects Presentation

2

OutlineOutline

DReAMSMatteo MurgidaAlessandro Panella

CITiESSimone CorbettaAlessandro MeroniAlessio Montone

Operating SystemIvan Beretta

PolarisMassimo MorandiMarco Novati

HLRMarco Maggioni

Page 3: HPPS 2007 Projects Presentation

3

What’s nextWhat’s next

DReAMSMatteo MurgidaAlessandro Panella

CITiESSimone CorbettaAlessandro MeroniAlessio Montone

Operating SystemIvan Beretta

PolarisMassimo MorandiMarco Novati

HLRMarco Maggioni

Page 4: HPPS 2007 Projects Presentation

POLITECNICO DI MILANO

DDynamicynamic Re Reconfigurabilityconfigurability AAppliedpplied toto M Multi-FPGAulti-FPGA

SSystemsystems

DReAMS

Page 5: HPPS 2007 Projects Presentation

DReAMSDReAMS

Dynamic ReconfigurabilityApplied to Multi-

FPGA SystemsBranch of DRESD projectInherits architectures and tools

Automatic workflow from VHDL system description to FPGA implementation

VHDL parsing and system simulationSystem creation over a specific architectureBitstream creation and download onto FPGAs

DReAMS

Page 6: HPPS 2007 Projects Presentation

Multi-FPGA Theoretical and Multi-FPGA Theoretical and Simulation Model Simulation Model 1/21/2

Project’s goals:Produce a multi-FPGA theoretical model

Architecture-independentMust capture all relevant features

Model Validation using several benchmarks

Definition/Identification of the set of benchmarksDO

VHDL description analysisPartitioningWriting a SystemC/VHDL modelSimulation

WHILE(No more improvement)

DReAMS

Page 7: HPPS 2007 Projects Presentation

Multi-FPGA Theoretical and Multi-FPGA Theoretical and Simulation Model Simulation Model 2/22/2

Project schedulingDetect relevant parameters of Multi-FPGA systemsAnalyze objective (cost) functions and architecture constraints

DimensionConnections bandwidthPower consumption…

Create a valid theoretical modelBenchmarks identification/definition Iterating process (analysis + partitioning + simulation)System implementation on Spartan-3 Multi-FPGA architecture

DReAMS

Page 8: HPPS 2007 Projects Presentation

Architecture DefinitionArchitecture Definition 1/31/3

Three Layers:Overall Multi-FPGA System

Net Topology Definition: mesh, ring, …

Single FPGADivision between fix and reconfigurable partsIP-Core selectionInternal Communication Infrastructure

Communication InfrastructurePhysical connections among FPGAsCommunication protocol

Development Environment: Digilent Spartan-3 boards

Final goal: distribuited dynamic reconfigurability

DReAMS

Page 9: HPPS 2007 Projects Presentation

Architecture Definition Architecture Definition 2/32/3DReA

MS

Page 10: HPPS 2007 Projects Presentation

Architecture DefinitionArchitecture Definition 3/33/3

Project ScheduleStudy how to use Digilent Spartan-3 boardsStudy its external interfaces and find a way to connect two or more boards togetherDesign the architecture of a single FPGA including the correct communication infrastructureDevelop the communication protocolConnect two boards togetherDevelop a simple distribuited application to test the validity of the proposed approach

DReAMS

Page 11: HPPS 2007 Projects Presentation

11

What’s nextWhat’s next

DReAMSMatteo MurgidaAlessandro Panella

CITiESSimone CorbettaAlessandro MeroniAlessio Montone

Operating SystemIvan Beretta

PolarisMassimo MorandiMarco Novati

HLRMarco Maggioni

Page 12: HPPS 2007 Projects Presentation

POLITECNICO DI MILANO

REREconfigurableconfigurable CCommunicationommunication

IInfrastructurenfrastructure F Foror EEmbedded-systemsmbedded-systems

Page 13: HPPS 2007 Projects Presentation

13

Project's objectivesProject's objectives

Communication infrastructure explorationTechnologies and paradigmsState of the art

Advantages and pitfallsComparison

Communication infrastructure for reconfigurable systems

CI requirements tailored for reconfigurable systems

Page 14: HPPS 2007 Projects Presentation

14

Schedule – Project OrganizationSchedule – Project Organization

Literature analysis

Reconfigurable devices and systems

Contextualization

Communication needs

Communication infrastructure state of the art

Paradigms

– analysis

– (potential) improvements

Communication infrastructure for

reconfigurable systems

Implementation

Subject to the De Micheli VHDL description

Page 15: HPPS 2007 Projects Presentation

15

What’s nextWhat’s next

DReAMSMatteo MurgidaAlessandro Panella

CITiESSimone CorbettaAlessandro MeroniAlessio Montone

Operating SystemIvan Beretta

PolarisMassimo MorandiMarco Novati

HLRMarco Maggioni

Page 16: HPPS 2007 Projects Presentation

POLITECNICO DI MILANO

RReconfigurationeconfiguration O Orientedriented MeMetricstrics

Page 17: HPPS 2007 Projects Presentation

Motivations and GoalsMotivations and Goals

RationaleRequirements-driven Reconfigurable SoC Communication Infrastructure design

e.g. QoS w.r.t. Load Balancing

ObjectivesDefinition and Validation of a set of Metrics tailored to identification and definition of the more effective Communication Infrastructure for Multi Processing Elements SoC architectureValidation framework definition

Simulator implementation

Page 18: HPPS 2007 Projects Presentation

Schedule - Project Schedule - Project OrganizationOrganization

Study and analysis of well-known metricsTCP/IP ProtocolsSystems migration between different Tier

Evaluation of different configurations of communication infrastructures

Topology (bus, point-to-point, cross-bar, NoC, …)Communication (connection-less, package-switching, circuit-switching, …)

Definition of metrics considering:Reconfigurable SystemDynamic changing of communication infrastructure elementsQuality of Service

Definition of a light framework Metrics Validation

Page 19: HPPS 2007 Projects Presentation

19

What’s nextWhat’s next

DReAMSMatteo MurgidaAlessandro Panella

CITiESSimone CorbettaAlessandro MeroniAlessio Montone

Operating SystemIvan Beretta

PolarisMassimo MorandiMarco Novati

HLRMarco Maggioni

Page 20: HPPS 2007 Projects Presentation

POLITECNICO DI MILANO

PProcessingrocessing E Elementslements REREconfigurationconfiguration I Inn

RReconfigurableeconfigurable A Architecturesrchitectures

Page 21: HPPS 2007 Projects Presentation

21

Project EnvironmentProject Environment

Multi Processing Elements SoC ArchitectureSupport Dynamic Partial ReconfigurabilityDeployable on FPGAs

Page 22: HPPS 2007 Projects Presentation

22

GoalsGoals

Implement and test a single Processing ElementBased on Harvard ArchitectureSoftcore Processor: MicroBlazeIt can be dynamically reconfigured on the device

Main ProblemsOn chip memory (BRAM) inizialization: current softwares (provided by FPGA’s vendors) support only total configuration bitstreams

Page 23: HPPS 2007 Projects Presentation

23

Schedule - Project Schedule - Project OrganizationOrganization

Bitstream’s structure analysisCheck differences between total configuration bitstreams and partial bitstreamsFind position of embedded memory information within the bitstream

Write bitstream memory initializator

Perform tests on physical devices

Page 24: HPPS 2007 Projects Presentation

24

What’s nextWhat’s next

DReAMSMatteo MurgidaAlessandro Panella

CITiESSimone CorbettaAlessandro MeroniAlessio Montone

Operating SystemIvan Beretta

PolarisMassimo MorandiMarco Novati

HLRMarco Maggioni

Page 25: HPPS 2007 Projects Presentation

POLITECNICO DI MILANO

Development of an OS Development of an OS architecture-independent architecture-independent

layer for dynamic layer for dynamic reconfigurationreconfiguration

Page 26: HPPS 2007 Projects Presentation

26

Scenario and GoalsScenario and GoalsCurrent scenario

Operating system support for dynamic reconfigurable architectures:

Architecture specific (e.g. Caronte)

Processor specific (e.g. PowerPC)

Tied to a particular distribution (e.g. MontaVista Linux)

Project objective

Definition of a new intermediate layer for an operating system which is:

Able to support dynamic reconfiguration

Architecture independent

High-level Linux distro independent

Implementation and validation using different FPGAs

Page 27: HPPS 2007 Projects Presentation

27

Schedule – Project OrganizationSchedule – Project Organization

Feasibility study

Study of the existing operating systems developed on the dynamic reconfigurable architectures defined in the DRESD Project

Definition of the new layer

Application

Integration of the new layer in an existing framework

Integration of the new layer in a different distribution executed on a different architecture

Implementation using Xilinx FPGAs: vp7, vp20 and vp30

Page 28: HPPS 2007 Projects Presentation

28

What’s nextWhat’s next

DReAMSMatteo MurgidaAlessandro Panella

CITiESSimone CorbettaAlessandro MeroniAlessio Montone

Operating SystemIvan Beretta

PolarisMassimo MorandiMarco Novati

HLRMarco Maggioni

Page 29: HPPS 2007 Projects Presentation

POLITECNICO DI MILANO

Effects of 2D Reconfiguration Effects of 2D Reconfiguration in a Reconfigurable Systemin a Reconfigurable System

Page 30: HPPS 2007 Projects Presentation

30

Effects of 2D ReconfigurationEffects of 2D Reconfiguration

New Generation of FPGAsVirtex-4 and Virtex-5Allow bi-dimensional reconfiguration

Improvements:Possibility for area and performance optimizations

Increased complexity:In fragmentation managementIn Placement In Communication infrastructure creationIn the Bitstream generation phase

30

Page 31: HPPS 2007 Projects Presentation

31

Project GoalsProject Goals

Project goals:Analyse effects of the new approachExamine possible remedies to the new problemsEvaluate those solutions in various scenario

31

Page 32: HPPS 2007 Projects Presentation

32

Schedule – Project Schedule – Project OrganizationOrganization

First Phase:General analysis of 2D reconfiguration

Second Phase:Detailed description of the new problems

Third Phase:Analysis of possible solutions to those problems

Fourth Phase:Evaluation of examined alternatives

32

Page 33: HPPS 2007 Projects Presentation

33

What’s nextWhat’s next

DReAMSMatteo MurgidaAlessandro Panella

CITiESSimone CorbettaAlessandro MeroniAlessio Montone

Operating SystemIvan Beretta

PolarisMassimo MorandiMarco Novati

HLRMarco Maggioni

Page 34: HPPS 2007 Projects Presentation

POLITECNICO DI MILANO

Relocation for 2D Relocation for 2D Reconfigurable SystemsReconfigurable Systems

Page 35: HPPS 2007 Projects Presentation

35

2D Relocation2D Relocation

Self dynamical run-time 2D reconfigurationVirtex-4 and Virtex-5Relocation

HW/SW solutions: advantages and disadvantagesBiRF

Project goals:Study of the new FPGA familiesAnalysis of the new bitstream structure

New version of BiRF (BiRF2)

35

Page 36: HPPS 2007 Projects Presentation

36

Schedule – Project Schedule – Project OrganizationOrganization

First Phase:Examine Xilinx documentation on Virtex-4 and 5

Second Phase:Generate Virtex-4 bitstreams to examine their structure

Third Phase:Implement the new version of BiRF

Fourth Phase:Validation of the results

36

Page 37: HPPS 2007 Projects Presentation

37

What’s nextWhat’s next

DReAMSMatteo MurgidaAlessandro Panella

CITiESSimone CorbettaAlessandro MeroniAlessio Montone

Operating SystemIvan Beretta

PolarisMassimo MorandiMarco Novati

HLRMarco Maggioni

Page 38: HPPS 2007 Projects Presentation

POLITECNICO DI MILANO

HHighigh L Levelevel RReconfigurationeconfiguration

Page 39: HPPS 2007 Projects Presentation

GoalsGoals

General Join isomorphic reconfigurable partitioning theory with reconfigurable scheduling performed by Salomone and area occupancy metricEvaluate quality of the given schedule result and optimize architecture exploiting Provide a common interface to represent TDG and scheduling output

SpecificAutomatize benchmarks productionRe-implementing Salomone to adopt the new defined interface Provide a graphical representation for the schedules

Page 40: HPPS 2007 Projects Presentation

Salomone++ workflowSalomone++ workflow

From specification to optimized scheduling…

TreeStructure

Graph

Analysis

Isomorphic

Partitioning

Specification

Area Occupation

Metrics

Salomone

SchedulingAllocationPolicies

OptimizedSchedule

Page 41: HPPS 2007 Projects Presentation

Schedule optimizationSchedule optimization

Evaluates a scheduling for a target architecture…

Based on simply considerationEach SCoNo portion depends of

its biggest node

We must modify schedule if exceeds area limit

If possible, we can save area and time anticipating loading of small different nodes of same SCoNo

Page 42: HPPS 2007 Projects Presentation

Project organizationProject organization

First phase: Development of the workflow

Isomorph partitioningSalomoneArea occupation metrics + optimization

Benchmarks

Second phase:Definition of the scheduler interfacesRe-implementation of SalomoneGraphical representation

Page 43: HPPS 2007 Projects Presentation

43

QuestionsQuestions