1. Volunteer Computing: SETI and Beyond David P. Anderson University of California, Berkeley 7 June 2007

2. Outline ● Who am I? ● SETI@home ● Volunteer computing ● BOINC ● The future

3. Flashback: 1994 ● Computers – 100 Mhz Pentium II is hot chip – computers are faster than humans ● Internet – Expanding from institutions to consumers – AOL grows to 10M subscribers ● 25 th anniversary of Apollo 11 moon landing ● David Gedye has an idea

4. Radio SETI Blackbody radiation (natural) Narrow-band signal (synthetic)

5. The Extraterrestrial Life Conjecture: is there life outside Earth? ● Early idea (Gauss, 1820):

6. How to investigate the ELC? ● Estimate: Drake’s Equation N = R×fp×ne×fl×fi×fc×L Milky Way: 400 billion stars 100 billion other galaxies R* is the average rate of star formation in our galaxy fp is the fraction of those stars that have planets ne is the average number of planets that can potentially support life per star that has planets fl is the fraction of the above that actually go on to develop life at some point fi is the fraction of the above that actually go on to develop intelligent life fc is the fraction of civilizations that develop a technology that releases detectable signs of their existence into space L is the length of time such civilizations release detectable signals into space.

7. Extrasolar planet discovery (1996 -) Doppler shifting of absorption lines: 236 planets discovered so far

8. How to detect life? ● Physical evidence ● Find synthetic EM waves – radio, optical ● Look at natural EM waves – planet imager proposal 100,000 light years

9. Pros and cons ● Should we listen? +: get encyclopedia; learn about alien science, culture, sociology. Join Galactic Internet. – : Destructive technology; computer viruses ● Should we transmit? +: Share legacy; need to send to hear – : Get noticed, invaded

10. Radio SETI: some details ● What frequency range? – dust transparency – Hydrogen line: 1.42 Ghz (21 cm) ● Where to look? – Targeted search – Sky survey ● Sensitivity and noise – RFI rejection ● How far can we hear? – probably < 100 light years

11. Radio SETI: history ● 1959: proposed by Phil Morrison and Giuseppe Cocconi ● 1960: Frank Drake: Project Ozma, Green Bank 25-meter dish ● 1963: NSF funds Ohio State project: The Big Ear – heard "Wow" signal in 1977 (not ET) ● 1971: NASA funds Cyclops, led by Frank Drake ● 1979-: SERENDIP (UC Berkeley) ● 1983-: Paul Horwitz, Harvard (telescopes at Harvard, Argentina) ● 1992: NASA funds sky survey (MOPS) ● 1993: Congress cancels MOPS, cuts off funds for SETI ● 1995: SETI Institute forms, creates Phoenix project – (Parkes/Australia, Green Bank, Arecibo) ● 2006: Allen Telescope Array (UC Berkeley, SETI Institute) – 350 dishes planned; 40 completed – multibeaming: use interferometry to look more than one place ● 2008-2015: Square Kilometer Array (multinational consortium) – 50X more sensitive than Arecibo

12. Arecibo and SERENDIP ● Arecibo Observatory – Funded by NSF – study atmosphere/planets/space – Sees 35 degree band – Sees 1 pixel, 0.1 degree ● SERENDIP – piggyback observation – analyze 100 Mhz band, 1 Hz resolution – hardware: FFT chips, Xilinx FPGA – record “spikes” above noise – look for repeats – reobserve

13. SETI@home: science ● Like SERENDIP, but use home PCs for analysis ● Data recording – 2.5 Mhz band, 2-bit samples – One DLT tape every 16 hours (now using hard disks) ● Splitting data into “workunits” – 10 Khz band, 105 seconds == 350 KB ● Better sensitivity because: – frequency resolution down to 0.1 Hz – detect Gaussians – detect “chirped” signals (changing frequency) – detect pulsed signals (all rates, phases, duty cycles)

14. SETI@home: history ● 1995: work out scientific plan ● 1996-7: unsuccessful fundraising ● 1998: get money from Paramount, Planetary Society; write software ● 1999 (May): launch, get 400,000 users ● 1999-2003: battle various problems – server scaling, network bandwidth – air conditioning, disk failures, DB integrity – DB performance problems – security issues (hackers, cheaters) – bad results from overclocked clients ● 2004: analyze accumulated signals (5 billion) – reobserve 250 best repeat signals – don’t find anything ● 2005: switch to BOINC ● 2007: using multibeam receiver at Arecibo; preparing to release new application (Astropulse, detects very short pulses)

15. Radio SETI: the big picture

16. Cycle-scavenging ● Within an organization – 1985: Condor – 1992: Prowess (Pratt and Whitney) ● With home PCs – 1996: GIMPS – 1997: distributed.net – 1999: SETI@home – 2000: Folding@home

17. Volunteer computing Internet Projects Volunteers ● Helps science ● Involves public in science

18. Volunteer computing ≠ Grid computing Resource owners Managed systems? Clients behind firewall? anonymous, unaccountable; need to check results no – need plug & play software yes – pull model yes – software stack requirements OK no – push model identified, accountable ISP bill? ye s nono... nor is it “peer-to-peer computing”

19. Volunteer computing software Application Infrastructure Manage work Transfer files Security Screensaver Accounting Do scientific computation

20. Berkeley Open Infrastructure for Network Computing (BOINC) ● Middleware for volunteer computing – also works great for desktop grid computing ● Funded by NSF, 2002-present ● Open-source (LGPL license) Volunteers Projects 80% 20%

21. Creating a BOINC project ● Set up a server (Linux/MySQL/Apache/BOINC) ● BOINC-enable application ● Compile app for Win, Mac, Linux etc. ● Develop programs to generate and handle work ● Test ● Create web pages about your research ● Publicize ● Keep volunteers informed

22. BOINC (Volunteer’s view) ● 1-click install, zero configuration ● All platforms ● Invisible, autonomic

23. Communication: “Pull” model client scheduler I’m a Windows/x86 computer with 512 MB RAM 20GB free disk 2.5 GFLOPS CPU Here are three jobs. Job 1 has application files A,B,C, input files C,D,E and output file F...

24. What applications work well with BOINC? ● Resource-intensive – 1 CPU year: do it yourself – 10,000 CPU years: use BOINC ● Lots of independent tasks ● Moderate data/compute ratio Examples: - Physical simulations (molecule, Earth, universe) - CPU-intensive data analysis - Search of large spaces (math)

25. Example: ClimatePrediction.net ● Application: UK Met Office Unified Model ● State-of-the-art global climate model – 1 million lines of FORTRAN ● High-dimensional search space – model parameters – boundary conditions – perturbed initial conditions

26. ClimatePrediction.net ● Using supercomputers: – 1 day per run – 10-20 total runs ● Using BOINC: – 6 months per run – 50,000 active hosts – 171,343 runs completed – Nature papers – 60-fold savings

27. Other BOINC-based projects ● Einstein@home – LIGO; gravitational wave astronomy ● Rosetta@home – U. Washington; protein study ● SETI@home – U.C. Berkeley; SETI ● LHC@home – CERN; accelerator simulation ● Africa@home – STI, U. of Geneva; malaria epidemiology ● IBM World Community Grid – several biomedical applications ●...and about 30 others

28. Computing power ● BOINC-based projects: ● Folding@home: 650 TeraFLOPS ● 200 from PCs ● 50 from GPUs ● 400 from 10,000 PS3s (Cell processor)

29. Future directions for BOINC ● Data-intensive computing – BitTorrent integration ● Low-latency computing ● Coexistence with other background activities – disk indexing, defragmentation, web prefetch, etc. ● Credit mechanism – credit == utility to projects ● Efficient validation – reduce use of replication ● Use of game consoles, GPUs, set-top boxes

30. What else can volunteers do? ● Testing ● Translation ● Program optimization ● Message-board moderation ● Online customer support ● Skilled tasks – Stardust@home

31. Conclusion ● SETI@home – we haven’t found ET, but we... – increased public awareness of SETI – popularized volunteer computing ● Volunteer computing – a new paradigm – more/cheaper computing power – public involvement in science – enabling technology: BOINC ● Contact me (davea@ssl.berkeley.edu) about: – Using BOINC for desktop grid computing – Organizational use of BOINC – Other ways the public can help science