Hardware Virtualization for
Pre-Silicon Software Development
in Automotive Electronics
Frank Schirrmeister, Filip Thoen
fschirr@synopsys.com
Market Trends & Challenges
Growing electronics content & complexity
y
Multi-node, networked ECU architecture typical
8Complex intra-ECU behavior & inter-ECU interaction
8
Meeting quality & reliability requires analysis & diagnosis of complex
hardware/software problems
y
Dramatic increase in software content & complexity
8
40% of vehicle cost is attributed to electronics & software cost
850-70% of ECU development cost is due to software
Need to reduce design cycle times & cost
y
While satisfying global regulations on passenger safety, emissions &
fuel economy
y
While guaranteeing high quality & reliability
Software & hardware are focused on reliability & quality
y
Both for safety-critical functionality and to control warranty costs
Competitive OEM market, complex dynamic of cost & reliability
y
Compete heavily in terms of safety, fuel efficiency & new features
Growing Complexity in Car Software
Source: BMW
System Complexity
Source: Toyota
#1 Automotive Challenges
Cost Reduction & Supply Chain Collaboration
Cost Reduction cited as #1
challenge facing the Automotive
Design and Engineering Community
(32% of users)!
Source: Dupont, 2008
53% of users think an increase in
collaboration across the value and
supply chain is needed to strengthen
the automotive industry!
HW/SW Challenge #2: Late Start
Software has to wait for
hardware => Semi cannot sell
silicon and bugs are found late
Cockpiut
CAN
Chip(s)
Peripheral
Models
Virtual Platform Example
Fast Transaction-level Models - Technology Details
Device
Board(s)
Virtual I/O
Mem
CPU
Model
Bus Model
Simulation Infrastructure
Transaction-level
Interfaces
High-speed
(SystemC™)
C++ Models
Graphical Hardware
Models
Emulation of
system I/O
Sensor(s)
Actuator(s)
Device
Under
Control
User Interface
Emulation
Fast
Instruction-set
Simulators
ECU in Innovator
Virtual Platform Demo
ECU in Innovator
Software Debugger
Module Authoring Module Authoring
Electronics in the Car
A Distributed Embedded System
Source: Alberto Sangiovanni-Vincentelli, “Design of Embedded Systems: Methodologies, Tools and Applications”
Today: More than 80
microprocessors
and millions of lines
of code !
But: Accuracy
requirements are
widely different
depending on which
part of the car
electronics is
Transaction-level Modeling Abstraction Levels
What Synopsys offers in TLM space
40-60 MIPS
1-10 MIPS
1-100 KIPS
Functionally
Cycle
Cycle
80+ MIPS
Prog. View
TLM (PV)
PV with Timing
TLM (PVT)
Cycle
Accurate
Pre-silicon Software
Development & Integration
Architectural Exploration
& Real-Time SW Development
(Profiling)
System Verification
& Timing Validation
App View
TLM (AV)
Application
Development
Automotive
Area
Information
Systems
Body
Functions
Body
Electronics
Speed
Accuracy
The Automotive Supply Chain
Companies
Hardware Developers
Programmers Processor &
Sub-system & IP Developer Users
Software IP / Software Suppliers
Independent Software Developers: Operating Systems: WindRiver, Linux, ecos, itron
Development Tools: WindRiver, Lauterbach, Greenhills, Etas, DSpace
System Integrators BMW Toyota Honda GM Chrysler Ford Subsystem Suppliers Bosch Continental Magneti Marelli Denso Delphi Visteon Semiconductor ST Freescale Infineon … Hardware IP
Processors: ARM, MIPS, PPC, X86, Tensilica, ARC, MeP, CEVA Interconnect: Arteris, Sonics, Silistix, Peripherals: Synopsys, Denali
Driv
ers
System
SW
SW
integration
Test
ing
Firm
ware
Driv
ers
•
Device drivers,•
OS•
Software integration•
Software development•
Validate architecture(s)& elect semi supplier
•
ECU softwareSoftware Responsibility
Value Proposition:
Design Cycle Reduction
Companies
Semiconductor System Integrators
Hardware IP Subsystem Suppliers
Software IP / Software Suppliers
Arch.
Design
Hardware
Design
Silicon
Proto
Firm
ware
Driv
ers
Driv
ers
System
SW
Part
Design
Proto
System
Design
Bench
/ Proto
SW
integration
Test
ing
Virtual
Platform
• Long, sequential development flow
• HW/SW meet late = risk!
• Reduced design cycle
The Automotive Supply Chain
Concerns, Responsibilities, Value of Virtual Platforms
Actual Platform usage (wireless example) Early (Pre-HW) Software Development
•
Device drivers,•
OS•
Software integration•
Software development•
Validate architecture(s)& elect semi supplier
•
ECU softwareSoftware Responsibility
Companies
Semiconductor System Integrators
Hardware IP Subsystem Suppliers
Software IP / Software Suppliers
•
Device drivers,•
OS•
Software integration•
Software development•
Validate architecture(s)& elect semi supplier
•
ECU softwareSoftware Responsibility
Value Proposition:
Design Cycle Reduction
Companies
Semiconductor System Integrators
Hardware IP Subsystem Suppliers
Software IP / Software Suppliers
Virtual Platforms as Collaboration Tool
Actual Platform usage (wireless example)
System
Integrators will
start to request
models here
Early (Pre-HW) Software Development
Value Proposition:
Higher Quality & Reliability, Lower Cost
Companies
Software IP / Software Suppliers
Technical Concerns Software Responsibility Hardware IP
•
Architecture analysis•
Feature definition, get designed in bysemiconductor house
•
Early supplier feedback•
Architecture trade-offsVirtual
Platform
Value
•
Software development productivity•
Quality & reliability•
Validate architecture(s) & elect semi supplier•
ECU software•
Reduce cycle time•
Improve quality
More time to test
Systematic test methodology•
Improve softwareproductivity
•
Reduce cost
Less bench time
Reduce need for prototypeSubsystem Suppliers System Integrators
•
Software development productivity•
Quality & reliability•
Software integration•
Some: Proprietary OS plus all development•
Early access•
Reduce cycle time•
Improve quality
More time to test
Systematic testmethodology.
•
Improve software productivity•
Reduce cost
Less bench time
Reduce need forprototype Semiconductor