View
217
Download
0
Category
Tags:
Preview:
Citation preview
SETI@homeand Citizen Cyber-Science
David P. Anderson
Space Sciences LaboratoryU.C. Berkeley
Citizen Cyber-Science
The use of Internet and computer technology to involve the public in scientific research
Volunteer computing Volunteer thinking
Stardust@home, GalaxyZoo, FoldIt!
SETI@home Classic
Volunteer PC SETI@home serverInternet
SETI@homeclient
download,install
SETI@home Classic
Volunteer PC SETI@home serverInternet
SETI@homeclient
serverrequest a job
SETI@home Classic
Volunteer PC SETI@home serverInternet
SETI@homeclient
serversend data
database
get job
SETI@home Classic
Volunteer PC SETI@home serverInternet
SETI@homeclient
(process data)
SETI@home Classic
Volunteer PC SETI@home serverInternet
server
return results,request new job
database
store resultsget jobSETI@home
client
Problems with S@h Classic
• Hard to deploy new applications or versions
• Wasteful replication
– “Always have work” policy
• No result validation
• Monolithic, inflexible server
Problems with S@h Classic
• Client only downloads 1 job at a time
• Client uses only 1 CPU
• Unit of credit is “job”
• Hard to participate in >1 project (e.g. Folding@home and SETI@home)
BOINC
• General-purpose “middleware” for volunteer computing
• Funded by NSF, 2002 - present
• Open-source (LGPL)
SETI@home/BOINC
Volunteer PCInternet
BOINC client download,install
SETI@home/BOINC
Volunteer PCInternet
BOINC client
“attach” to project
SETI@home server
SETI@home/BOINC
Volunteer PCInternet
BOINC client
request jobs type of computer RAM, CPU speed connection interval
SETI@home server
scheduler
BOINCDB
get jobs
SETI@home/BOINC
Volunteer PCInternet
BOINC client
job descriptions list of application files list of input files list of output files
SETI@home server
scheduler
SETI@home/BOINC
Volunteer PCInternet
BOINC client
download program,input files
SETI@home server
data servers
SETI@home/BOINC
Volunteer PCInternet
BOINC client
SETI@home server
programprogram
(run applications)
SETI@home/BOINC
Volunteer PCInternet
BOINC client
upload outputfiles
SETI@home server
data servers
SETI@home/BOINC
Volunteer PCInternet
BOINC client
report completed jobs
SETI@home server
scheduler
BOINCDB
ScienceDB
assimilate
validate
Problems solved
• Easy to deploy new apps
• Efficient replication
– 2-fold, adaptive
• Errors, cheating are detected immediately
• Server is scalable, fail-soft
• Credit is based on FLOPS, not jobs
• Client can queue lots of jobs, use >1 CPU
Painful transition to BOINC (2005)
• Awkward registration process
• “techie” GUI
• Classic credit didn’t carry over
• User base: 500K -> 200K
The volunteer computing ecosystem
projects
CPDN
LHC@home
WCGattachments
volunteers
Climateprediction.net
Einstein@home
• Gravitational waves
Other projects
• LHC@home
• Rosetta@home
• IBM World Community Grid
• Etc.
– epidemiology, plant ecology, genetic linkage, phylogenetics, graph theory, number theory, cognitive modeling, fluid dynamics, galactic structure, quantum chemistry, nanotechnology, quantum computing, cosmology
History of volunteer computing
Applications
Middleware
1995 2005distributed.net, GIMPS
SETI@home, Folding@home
Commercial: Entropia, United Devices, ...
BOINC
Climateprediction.netPredictor@homeIBM World Community GridEinstein@homeRosetta@home ...
20052000 now
Academic: Bayanihan, Javelin, ...
Applications
Performance
• Current
– 500K people, 1M computers
– 6.5 PetaFLOPS (3 from GPUs, 1.4 from PS3s)
• Potential
– 1 billion PCs today, 2 billion in 2015
– GPU: approaching 1 TFLOPS
– 1 ExaFLOPS: 4M GPUs * 0.25 availability
– Freescale i.MX51: 2 GFLOPS, 200mw
Organizational models
Umbrella projects
• Institutional
– Lattice, VTU@home
• Corporate
– IBM World Community Grid
• Community
– AlmereGrid
• Research community
– MindModeling.org
Project
publicityweb developmentsysadmin
Stardust@home
• The Stardust mission
• Where’s the dust?
• Stardust@home
– 23K volunteers
– 43M viewings
– 64 tracks found
jobs
middleware
people orcomputers
identityaccounting
queuingassignmentvalidation
What’s different?
• People vary
• Jobs may not be well-defined
aptitudetraining
Bossa
• Open-source PHP-based middleware for volunteer thinking
– http://bossa.berkeley.edu
• Policies
– replication, volunteer assessment, job assignment
– Bossa doesn’t provide policies, but makes it easy to implement a wide range of policies
Projects in development
• Hominids@home
– Collect photos of Middle Awash (Ethiopia)
– Look for hominid and other fossils
• AfricaMap
Conclusion
• Citizen Cyber-Science
– volunteer computing
– volunteer thinking
• Benefits to science
• Benefits to society
Recommended