SETI@Home Transitions To BOINC

SeaDour writes "The team at SETI@Home have finally released their highly-anticipated new client software based on the BOINC (Berkeley Open Infrastructure for Network Computing) software platform. This new platform promises transparent version upgrades, more efficient work unit distribution, and the ability to seamlessly integrate other distributed computing projects that are also using the BOINC standard. For now, SETI@Home is allowing both the Classic and BOINC clients to run, but eventually they will shut down the Classic data server and force everyone to upgrade. You can read more about the transition here."

Re:SETI running out of Work-Units?

[Thng]

Not sure if this is before or after the news you heard, but for version 3.03, they added additional processing capability in the client so workunits would take longer to process
News posting [berkeley.edu]
Text:

Added additional science coverage. We now do a thorough search out to a chirp rate of +- 20 Hz/second. The cost of the additional coverage is that clients will take longer to process a workunit

However, as 3.03 is rather old, I wouldn't be surprised if the new and faster computers and old clients that weren't upgraded negated some of the effect.

thng

Re:a great joke

[Kiryat Malachi]

Best you'd conceivably get is a detection from one dish; without actually putting a transmitter way out in space, the confirmations that would be required (things like confirming parallax of the signal, motion of the signal consistent with it being X light years away, etc.) would require access to every radiotelescope in the world.

Best you might get is "oh, neat, a candidate signal" until one or the other of the rejection mechanisms coughs and says "Bullshit."

comparative advantage in distributed computing

[k2enemy]

this is the first i've seen BOINC. it looks like a good platform to implement concepts of comparative advantage in distributed computing projects [jahana.com] . the idea is to apply some of the concepts that drive international trade to distributed computing.

BOINC

[Sunspire]

Boinc is more than just an updated Seti@Home, it's a generic delivery platform for distributed projects. That means you, yes you, can develope a BOINC app [berkeley.edu]. Just gather some people to run it for you and compute away without needing any approval from the guys at Berkeley. Basically the participants enter a project URL into the BOINC application, the program then downloads your code and the crunching begins. BOINC handles all the network, workunit, results, distribution, security, versioning etc. issues for you.

Participants can even choose to split their resources among several projects, say, Seti@Home and Folding@Home. Another thing that will also be used in the new Seti@Home is that you can have clients participating in the same project working on completely different computation sets. For example, clients that have proven themselves to have a fast workunit turnaround time and a long history of participating and that have a gigabyte or more of RAM can be given special tasks that would normally be impossible because of the high number of griefers on the net.

BOINC SETI@home - Ready for Prime Time?

[wintermute1974]

In the hopes that someone working on the SETI@home client reads Slashdot, I am posting my bug report about the new BOINC SETI@home client here. I tried to post a bug report on their Windows Client Forum [berkeley.edu], but the authentication routine failed to recognize my account ID when I copied and pasted it from the e-mail message I had just received. (If you don't yet have a BOINC account, it looks something like 213ed9ba2da1696f77b0d0fa3165a3ab, but no, this is NOT my real user account.)

Anyway, enough preamble. Here's the problem:
In the Work tab, when I right-click on the currently-running work unit, the context-sensitive menu displays one option, Show Graphics.

When I select Show Graphics, a window pops up, the entire contents of which is black. At this point, my Windows 2000 SP4 computer freezes. CTRL-ALT-DEL doesn't bring up the Windows Security window. CTRL-SHIFT-ESC doesn't bring up the Task Manager. I can't move the mouse. The keyboard is completely unresponsive.

Being a sucker for punishment, I sent a non-maskable interrupt to my CPU, and rebooted the machine. Then I tried the exact same steps, and got the same results. Yup, this bug is repeatable.

So is the new client ready for prime time? Um, not really. Add the insult of the website not recognizing the account ID that it gave me to begin with and I'd say this program should stay in beta a while longer.

A final note: If you happen to be one of the programmers for the client, and know why this problem is happening, reply here. I'd appreciate a reply.

Re:That's unfair.

[Anonymous Coward]

Here [go.com] you go.

Re:OT: Re:File format is not XML: why not?

[holloway]

1. Binary data can be referenced in external files. Consider the OpenOffice xml format, where it's a single zip file with multiple xml files and the binaries that are referenced.
2. It's inefficient compared to binary, but then it's more readable to programmers (and even some non-programmers feel comfortable opening a file, searching for a term, and replacing it - as my non-programmer boss did once).

If you're after a binary format try EBML at sourceforge. It's a binary equivalent of XML syntax.

Generally I think you're being too harsh on text formats. Plaintext configuration files (non-xml, I'm thinking of unix config files) have shown their worth, and binary configuration files mean you need custom editors. The processing time of XML vs binary is meaningless and is not a bottleneck for most applications -- especially this one where it'd only be XML for the data transfer. So far as bandwidth goes it could be gzipped.

XML isn't a particular language, it's a metalanguage, and depending on the format it can provide information on how to interpret a document as much as a binary source file could. Whether the logic should be with the file is again a question of implementation, so there's no reason to complain about binary or XML unless you'd like to get more specific about which format is lacking (and then it would probably be a mistake in that format, not XML).

XML is so much easier for programmers than binary. Many people who think they know XML think it's just the XML spec rather than the surrounding standards such as XSLT, XSLFO, XQuery, Schema, RelaxNG - maybe even Tamino. People can't really say they know XML well unless they understand those specs and the implementations.

Re:Different Projects?

[kpearson]

BOINC doesn't run multiple projects at the same time. It runs one project at a time, but it divides its time between projects according to percentages that you choose.

There are no active, public projects besides SETI@home yet. Predictor@home [scripps.edu] is running a public alpha test of its client that anyone can participate in. climateprediction.net [climateprediction.net] began a private alpha test of its client today, and plans to begin a public beta test next month. Folding@home [stanford.edu] is developing a client, but has not announced any alpha or beta testing for it yet. BOINC Beta Test [berkeley.edu] is still beta testing the BOINC client and may create an Astropulse project based on the client. Einstein@Home [physics2005.org] may be developing a client based on BOINC for its project which begins in 2005.

Re:SETI running out of Work-Units?

[Graymalkin]

I suspect the Allen Telescope Array [seti.org] will be providing quite a bit more data for SETI@Home to chew on. Not only will be it scanning a wider range of frequencies but an order of magnitude more stars. The extra data along with the enhanced processing taken from the 3.03 S@H client will likely keep the project plenty busy for a while longer. Optical SETI is also gaining some mindshare and research dollars. I don't think it will be too long before an optical scan tool is added to the new BOINC client.

Full source *is* available on BOINC site!

[Anonymous Coward]

It's just not easy to find:

http://boinc.berkeley.edu/seti_source/

Unless I'm missing something, this has all the source for the client, including the signal analysis code and the communications protocol.

You're right, to my knowledge this is the first time the source has been available. I previously would never touch SETI because of their security-through-obscurity mentality. Apparently they finally got with the program.

Re:Me predicts...

[Anonymous Coward]

Umm... This new version is GPL'ed

http://boinc.berkeley.edu/seti_source/ [berkeley.edu]

Re:Uh oh. New virus vector

[Anonymous Coward]

Absolutely nothing. http://boinc.berkeley.edu/security.php

This looks far more secure than having the user try to verify themselves.

SETI@home is GPL

[SETIGuy]

This is, of course, one of the reasons why I won't use SETI@Home until it is GPL or similar [Would it be GPL with BOINC?]

Um, the SETI@home version that runs under BOINC is GPL, and has been so for some time. The BOINC client is BOINC Public License, which, because of a legal settlement, restricts commercial use until late this year. After the agreement expires, BOINC will transition to Mozilla license or GPL. I don't think we've decided which.

You are also free to download both BOINC and SETI@home and compile them on your home machine under the "anonymous platform" mechanism. That way you don't need to download binaries.

BTW, we sign our BOINC/SETI@home binary code on a non-networked machine kept under lock and key.