Embedded Systems Architecture Davide Bertozzi University of - - PowerPoint PPT Presentation

Embedded Systems Architecture Course Presentation

Davide Bertozzi University of Ferrara

Course Languages

2

English Italian

All the course material will be in english, but the lectures will be in italian.

Teaching Staff

3

QUIZ: which picture is outdated?

Course Website

 http://mpsoc.unife.it/~arch-dig/

 All the course material and lab exercises will

be uploaded on this website

4

Course Schedule and Material

In theory, our schedule would have been:

 TUESDAYS

 FROM 14 TO 16:30  CAN BE LECTURE OR LAB. EXERCISE

 WEDNESDAYS

 FROM 16:30 to 19

In practice:

 I will upload my recorded lectures on Google Classroom

(course code glxrzcw) every Wednesday and Thursday morning at latest (i.e., recordings available from the day after).

 Videoconference for Questions and Answers: each

TUESDAY at 3pm during a ConfCall via Hangout Meet.

5

Laboratory

 Here we need to be creative….  …I may provide you the instructions to carry out lab.

assignments via the course website.

 You would then be supposed to carry out such assignments

• n your laptop or home PC.

 You should notify as soon as possible whether YOU

DO NOT HAVE A LAPTOP OR PC AT HOME. Just send me an email if you fall into this case. If you have a laptop or home PC, do not send me any email.

6

Exam

 The course ends with an ORAL exam, consisting of three

questions that span the entire program.

 Capability to establish links between topics is preferred over

memorizing notions.

 During the first pass of the course material, ask youself: «what is

the main point this slide makes? What is the key take-away?». Do not care about the details.

 During the second pass of the course material, try to capture the

details.

 During the third pass of the material, try to test your level of

understanding of both the key take-aways and the technical

• details. You will then be in a position to estimate your final mark!

 In order to get the mark registered, you need to get a «pass»

in the lab. assignments.

7

Recommended books

 No single book covers all the course content. This is an

advanced course with several lectures covering frontier aspects

• f SoC design for embedded systems.

 Taking notes during the playback of online lectures is the best

way to learn effectively.

 However, on some specific topics I can recommend reference

books.

 See next slide.

8

Recommended Books

9

System-level design with the SystemC language:



• 1. T.Groetker, S.Liao, G.Martin, S.Swan; System Design with SystemC, Kluwer

Academic Publishers, 2002 Networks-on-chip for SoC connectivity:



• 2. J.Flich, D.Bertozzi; Designing Network-on-Chip Architectures in the Nanoscale Era,

CRC Press, 2011. Design methodologies:



• 3. Digital Integrated Circuits - A Design Perspective (second edition), J.M.Rabaey,

A.Chandrakasan, B.Nikolic, Prentice Hall

• Chapter on «Design methodologies».

Available in Italian:

• 3. Circuiti Integrati Digitali – l’Ottica del Progettista, J.M.Rabaey, A.Chandrakasan,

B.Nikolic, Pearson, 2019.

• Chapter on «Tecniche di Implementazione dei Circuiti Digitali».

Computer architecture:



• 4. David A. Patterson, John L. Hennessy; Computer Organization and Design: The

Hardware/Software Interface, Morgan Kaufmann, 2004 Available in Italian:

• 4. David A. Patterson, John L. Hennessy; Struttura e Progetto dei Calcolatori –

L’Interfaccia Hardware/Software, Zanichelli.

Electronic Board

Focus of the Course

Application Processor DRAM memory Gyroscope and accelerometer Modem LTE/GSM NFC Controller Image processor

Focus of the Course

Application Processor Our Focus is on the «brain» of the embedded system, in this case, on its «Application Processor»: Snapdragon (Qualcomm), Exynos (Samsung), Helio (Mediatek), OMAP (Texas Instr.), Kirin (HiSilicon/Huawei), Tegra (Nvidia), Ax (Apple), ….

Focus of the Course

12

SAMSUNG EXYNOS 7420 OCTA-CORE

A high-end example of «System-on-Chip» architecture

Similarly, Game Consoles

13

AMD Jaguar Architecture PS4

Similarly, Automotive

Head-Up Display Driver Alertness Monitoring Instrument Cluster Accident Recorder Event Data Recorder Auto-Dimming Mirror Interior Lighting Active Cabin Noise suppression Voice/Data Communications Cabin Environment Control DSRC Entertainment System Battery Management Lane Correction Electronic Toll Collection Digital Turn Signals Navigation System Security System Active Exhaust Noise Suppression Hill Hold Control Regenerative Braking Antilock Braking Tire Pressure Monitoring Parking System Electronic Stability Control Active Y… Control Active Wheel Control Seat Position Control Transmission Control Lane Departure Warning Bindspot Detection Remote Keyless Entry Cylinder Deactivation Idle Stop/Start Active Vibration Control Electronic Valve Timing Electric Power Steering Automatic Braking Electronic Throttle Control Adatpive Cruise Control Adaptive Front Lighting Airbag Deployment Engine Control Parental Controls Windshielded Wiper Control Night Vision

Car intelligence is provided by a densely populated network of Electronic Control Units

Consolidation of multiple computation workloads onto the same high-end embedded computing platform is well-underway in the automotive domain

14

TODAY

• 60-100 Electronic Control Units

(ECUs)

• 6-8 operating systems
• Isolated operations
• Increasing cost and complexity

Similarly, Automotive

ECU consolidation into a few high-end embedded computing platforms. Workload consolidation on the same platforms. (benefits in cost, weight, complexity, security)

ADAS Controller Cockpit Controller Safety Controller Handful and other ECUs Emission Controller Chassis Controller Body Controller

Domain Controllers

Engine Area Controller Cockpit Area Controller Safety Area Controller Back Safety Area Controller Handful and other ECUs Front Safety Area Controller Front Body Controller

Area Controllers

TOMORROW

• 6-10 Domain/Area Mega-Controllers
• Consolidated Software System
• Coordinated Operations
• Reduce Weight, Cost and Complexity

Consolidation of multiple computation workloads onto the same high-end embedded computing platform is well-underway in the automotive domain