Special Topic in CE Embedded Multicore Systems: Architecture, Application and Compiler

1. Special Topic in CEEmbedded Multicore Systems: Architecture, Application and Compiler Jongeun Lee Fall 2013

2. Technology Trend • CPU trends • #transistors • clock speed • power • perf/clock (via http://extremetech.com)

3. Power Wall In CMOS IC technology (source: intel.com) ×30 5V → 1V ×1000

4. Transistor (source: wikipedia.com)

5. Semiconductor Scaling (source: IBM Academy of Technology)

6. (source: Borkar 1999)

7. Energy Efficient Computing • Challenges • New approaches

8. Multicore • relies on simpler cores, multi-threaded execution • challenges: programming, scalability, … • may not necessarily reduce power – unless freqis reduced DVFS (Dynamic Voltage & Frequency Scaling) (source: Intel Inc. via Embedded.com)

9. System Designer’s Perspective • ASIC delivers best performance & power efficiency • hardware accelerators extensively used in embedded systems • use multiple accelerators for higher energy efficiency (eg, Tegra2) (source: Retarget.com)

10. Embedded Systems • Embedded computing system: any device that includes a programmable computer but is not itself a general-purpose computer • Application-specific optimization • don’t need all the general-purpose bells and whistles Modems MPEG decoders Network cards Network switches/routers On-board navigation Pagers Photocopiers Point-of-sale systems Portable video games Printers Satellite phones Scanners Smart ovens/dishwashers Speech recognizers Stereo systems Teleconferencing systems Televisions Temperature controllers Theft tracking systems TV set-top boxes VCR’s, DVD players Video game consoles Video phones Washers and dryers Anti-lock brakes Auto-focus cameras Automatic teller machines Automatic toll systems Automatic transmission Avionic systems Battery chargers Camcorders Cell phones Cell-phone base stations Cordless phones Cruise control Curbside check-in systems Digital cameras Disk drives Electronic card readers Electronic instruments Electronic toys/games Factory control Fax machines Fingerprint identifiers Home security systems Life-support systems Medical testing systems

11. Embedded Systems Characteristics • Sophisticated functionality • Cell phone, navigation (sophisticated UI) • Real-time operation • Hard vs. soft • Multi-rate • Low manufacturing cost • Non-functional requirements • Low power • Designed to tight deadlines by small teams • 6 months market window is common

12. Embedded Systems as Technology Frontier • Late 1940’s: MIT Whirlwind computer was designed for real-time operations • Originally designed to control an aircraft simulator • First microprocessor was Intel 4004 in early 1970’s • Designed for embedded application (calculator) • HP-35 calculator used several chips to implement a microprocessor in 1972 • Automobiles used microprocessor-based engine controllers starting in 1970’s • Control fuel/air mixture, engine timing, etc. • Multiple modes of operation: warm-up, cruise, hill climbing, etc. • Provides lower emissions, better fuel efficiency HP-35 (US $395)

13. Embedded Systems • Much earlier multicore adoption • CPU + DSP was a norm since at least 20 years ago • Many application-specific accelerator processors (video, media, audio, graphics, etc.) • STI Cell BE • Much earlier power-conscious design • To increase battery life

14. About this course:

15. Course Objective • Quick intro to main topics in the areas of architecture and compiler • Focus on a few research topics • selected by student+professor

16. Course Format • Try to be informal • to encourage participation and discussion • Course format • 2 sessions per week, 10 weeks • Visit course homepage for up-to-date info • http://ecl.unist.ac.kr/~jlee/2013/ece610/ • Grading • Class participation • Midterm report (oral) • Final project (written)

17. Course Topics • Research topics • Mobile application processor • Cache coherency and transactional memory • Mobile (=low-power) GPU • Programming FPGA Using OpenCL • Programming Xeon Phi Using OpenCL • Optimizing application XXX for platform YYY • (you may add to this list)

18. Schedule • Review of computer architecture topics • D. Patterson & J. Hennessy, Computer Organization • Parallel applications • Multicore programming (not covered here) • Modern multi- or many-core architectures • Programming framework • OpenCL, StreamIt (in addition to OpenMP, MPI) 2~3 weeks? 2 weeks? 2 weeks? 2 weeks?

19. To-dos • Required readings • give presentations (how often?) • slides are not required • papers • Esmaeilzadeh et al. (2013) “Power Challenges May End the Multicore Era” • parallel applications • parallel architectures (?) • Project • Midterm report

20. (slide from M. Pedram, USC, 2005)

21. (slide from M. Pedram, USC, 2005)