Intro WS 2018 Ulrich Schmid s@ecs.tuwien.ac.at - - PowerPoint PPT Presentation

▶

Mar 14, 2023 688 likes •818 views

Intro WS 2018 Ulrich Schmid s@ecs.tuwien.ac.at http://ti.tuwien.ac.at/ecs/teaching/courses/brds RTOS hands-on experience Connect distributed algorithms theory and practice U. Schmid 182.704 BRDS 2 U. Schmid 182.704 BRDS 3

SLIDE 1

Intro WS 2018 Ulrich Schmid s@ecs.tuwien.ac.at http://ti.tuwien.ac.at/ecs/teaching/courses/brds

SLIDE 2

 RTOS hands-on experience  Connect distributed algorithms theory

and practice

U. Schmid

182.704 BRDS 2

SLIDE 3

U. Schmid

182.704 BRDS 3

SLIDE 4

U. Schmid

182.704 BRDS 4

http://www.omappedia.org/wiki/PandaBoard

SLIDE 5

U. Schmid

182.704 BRDS 5

SLIDE 6

 Install your prefered RTOS on the Pandaboard

QNX Neutrino
RT-Linux (Ubuntu)
Your favorite other RTOS. Caveat:

 Watch out for existing Pandaboard BSP!  Need to provide additional resources (doc links, architecture and feature

verview slides)

 Install cross-development environment  Measure local (timer-)event service time latency  Expected additional results:

Setup.txt: Detailed install instructions and configuration choices
Lab protocol (or slides): Description of latency measurement

setup and results

U. Schmid

182.704 BRDS 6

SLIDE 7

 Want to run

asynchronous
synchronous

distributed algorithms on networked embedded system

 Low-level

communication: Either via Ethernet or 802.11

U. Schmid

182.704 BRDS 7

SLIDE 8

U. Schmid

182.704 BRDS 8

RTOS Device Drivers Timers 802.3 802.11 UART IO-Ports Clocksync Async unicast message send/rcv (unreliable/reliable) Lock-step rounds Sync message bcast/rcv Asynchronous distributed algorithm Synchronous distributed algorithm

SLIDE 9

 Devise system configuration and node

addressing scheme

 Implement asynchronous unicast message

passing interface (both unreliable and reliable)

 Implement clock synchronization algorithm  Implement lock-step round abstraction and

synchronous broadcast message passing IFC

 Expected additional results:

Lab protocol (or slides): Overview of architecture and

design of your solution, including the interface to the RTOS

U. Schmid

182.704 BRDS 9

SLIDE 10

 Implement simple asynchronous distributed

algorithm. Note that the CS algorithm would also do

 Measure end-to-end delays (and termination times)  Implement simple fault-tolerant synchronous

distributed algorithm

 Measure clock synchronization accuracy and

evaluate termination time and reliability

 Expected additional results:

Lab protocol (or slides): Overview of specification and

implementation of your algorithms and evaluation results.

Presentation (15-25 min) of your solution and results in one
f the students‘ presentations
U. Schmid

182.704 BRDS 10

SLIDE 11

 Groups of 2 students are prefered  Every group gets 2 Pandaboards that can be carried home (until

Task 3 integration tests start)

 You can (also) work in the ECS lab (2nd floor), which you can

access using your TI chip card (will be enabled)

 If you select QNX Neutrino, you need to sign a non-disclosure

agreement

 You can choose the date of your presentation (every Thursday

14:15-15:45), but please send us an email one week before so that we can enter your name + RTOS into the schedule

 Your attendance in the others‘ student presentations is welcome but

not mandatory

 You must upload code, setup.txt and lab protocol to subdirectory

„BRDS“ in your ECS lab homedirectory (will be created). Make sure that the whole subtree has group „brds“ and is read- and writeable for the group!

U. Schmid

182.704 BRDS 11