Archives

April 2001

New Weeklies
I forgot to do the weekly stats last weekend, so this week we have a two fer for ya.  You can check out the weekly stats from the previous week, and now you can check out this weeks weeklies also.  Words of congratulations are in order for both kiltannen and also wiz_of_ez.  They both popped into the top 1000 users overall in the S@H project.  Quite an accomplishment!

Correction...
I was notified by email that I had something a bit wrong In the news from the 17th.  I stated that TLC had more active members than any other team out there, but that may not have been completely correct.  Kered Ollirrac passed on word that Team MacAddict had over 1500 active members!  Woo!  I'm not exactly sure what their definition of active users is though...I may have to check to make sure.

New Addon
I also received an email from Jeff Omick.  He sent on a link to Msetimon which is a simple program that shows the status of S@H clients being run on multiple computers over a network or multiplie instances on the same computer.  Im not sure this compares with SetiWatch...but if you are interested you may want to check it out.

Catching Up
Sorry for the lack of news updates, I needed to take a slight break from things for a bit.  Now that I have the stats a bit automated it made that easier, but in the meantime the news here was neglected a tad.  So far, after passing up Sun Microsystems for the #1 spot, the TLC effort has not waned a bit.  The daily team production numbers are as high as ever while we are pulling away from the pack.  TLC now boasts over 4000 members, but not all of those members are "active" members.  A look inside the numbers show that  as of yesterday's update there are 1151 members who have posted at least one work unit within the last week, and 835 member returned work units yesterday.  That is still a pretty hefty showing.  To put those numbers in perspective, the 835 members that returned work units yesterday were more than the total members of 180 of the top 200 teams in the S@H project!  Aside from maybe Art Bell, TLC probably has the highest total of active members of any team.  period.  With the current influx of different distributed computing projects available, I am glad to see that there are so many people who still think that the SETI@Home project is worthy enuff for their spare CPU cycles. 

Even though the news on the S@H front is a bit slow, there still are some newsworthy things to report on....so lets get going.

v3.0 OS X Client Available
Recently the S@H crew released a new client for the Mac OS X operating system.  You can download the client here.  How does the client run in comparison to other clients/systems?  Well from OoklaTheMok: OS X GUI client is a genuine POS.  To quote him from the thread: 

Benchmark unit: 18 hours 30 minutes.
50% done.
This same computer did the OS 9 GUI client at 15:28!!!

...

The thing is, back with 2.4 and below, the mac gui was significantly faster than the win gui--back when backside cache size mattered. Now it's slower. But the OS X gui speed is simply abominable...the person who programmed it obviously is a newbie to OS X programming (eh, but who isn't?).

SetiQueue: Out of Beta:
There is a new version for the always wonderful SetiQueue caching program.  It is now out of beta and now is up to version 3.03.1.0.  You can check out all of the SetiQueue goodness and download the program if you wish here. 

SETIMonitor: New Version Too!
Not to be outdone, there is also a new version of SETIMonitor available.  Get the latest info here.

One Last Thing
One of the problems that lead to the increased science in the version 3.03 client was bandwidth problems on the Berkeley servers.  Of course many people suggested that compressing the work units/results could overcome some of this problems.  Here is what Eric Korpela has to say about it:

>>You will probably see compressed download in a future version, but we're not
>>in a huge rush to implement it.  There are other higher priorities for our
>>limited resources.  That could change in the future, though. 
>
>The simplest and cheapest way to implement "compression" on work units
>stored in text form would be to undo the text conversion in the server.

That is precisely the type of "compression" we would be likely to use.

Eric

That is all for now.  I still need to tweak some things on the team pages, and hopefully I can get enuff motivation to get them done.  Also I am going to work on a write up on exactly how I generate the stats for the team pages.  There have been a bunch of people who have asked me for more information on this, and I am going to try to deliver.

Unsung Hero
There is one person who is not a member of TLC who has been a huge part of the teams success over the past year plus.  That person is Jason Johnston of Nimbus Networks.  Through an arrangement with C. Eric Smith, the TLC website here has been hosted by Jason and Nimbus Networks for the past year for free.  The only stipulation for the hosting was to put a banner on the site for them, which I have had on the site under the menu bar.  Over the past 6+ months the server performance for the site has been outstanding.  It has been one constant that I could rely upon while keeping track of things for TLC (definitely more reliable than the stats and server status on the S@H site!).  The TLC site has rarely been down, every time I have had to do my updates the site has been up. 

Jason is in the process of ramping up services for web site hosting services.  If you or anyone else needs some type of web hosting, please drop Jason a line. Right now he is in the process of getting the Nimbus Networks site up and going for more information on the services.  In the mean time you can contact Jason at jj@nimbusnetworks.com.  There are also a couple of phone numbers available on the Nimbus site now to contact.  I am sure they can provide you with the web hosting that you need at competitive prices.  Please check them out!

Officially #1
Team Lamb Chop is now the overall #1 team in the SETI@Home Project.  Think about that.  The most successful team in the largest and most widely know distributed project in the world.  That's nothing to sneeze at.  TLC only ranks third in the number of users (behind Art Bell and MacAddict), but we kicked their butt.  There are major computer companies like Sun, Compaq, SGI, Intel...and software monopolies like Microsoft, but they all are looking up at us.  Hard work and dedication has made the readers of Ars Technica rise to the top and conquer all that stood in its way.  Stats are part of it, so is the publicity generated from the benchmark pages.  But what has held this team together over the long haul was the sense of community born through the Ars Forums.  It was an all around effort by all around people doing an all around kick ass job.  Congrats to all of you!

The team has had it's ups and downs in the past couple of years, and our share of controversies, but we weathered it all and strived for the top.  Being the #1 overall team was not a primary goal when I first started the pages here.  I only wanted a source of news for the team, and to provide some general stats for the team.  From such humble beginnings a juggernaut was born.  When TLC started to close in on #1, it was like Rocky fighting Apollo Creed.  When things looked good, there was Apollo to back us into the corner and give us a good whipping.  But we persevered and are now "champions".  How long will the reign last?  I don't believe that there are any other contenders left.  Unless there is some sort of increased effort from Sun, SGI, or Compaq, the #1 spot is safe.  Even if some type of dark horse would come out of the blue, I am sure that TLC could circle the ships and hold off the charge.

How did things unfold last week?  The #1 spot was actually given to us in a "gift".  TLC with its own production would have passed Sun sometime the past Friday.  Earlier in the week guru decided to leave the Sun team and move back to the rnoggin team.  When that happened, that put TLC in front based on team work units.  TLC did not show up on top of the charts because the S@H servers cache team changes and updates them on the stats on Saturdays.  Even though guru left Sun, the S@H pages would not update the actual team total for Sun until Sunday.  TLC did push ahead of Sun on work units even on the S@H pages on Friday...but the overall team pages didn't show TLC on top because they had not been updated since Thursday.

Sunday that all changed.  The S@H servers flushed the cached team change for Sun (on Saturday), and finally on Sunday, they updated the overall team pages to show Team Lamb Chop on top.  If I could buy you all a drink I would.  But I can't....so you will just have to settle for a congratulatory pat on the back :).  Way to Go TLC....lets keep it going!

More on the way from S@H
One of the things that many S@H participants have called for in the past year or so were results of the crunching that we have been doing for 2+ years on the project.  In the past week there were a couple of announcements that show that these changes are forthcoming.  First was a post on alt.sci.seti from Eric Korpela in response to some suggestions:

Bob Keeter  <rkeeter@earthlink.net> wrote:

Got a crazy idea to toss out for discussion.  (Been a while since I've last rolled a grenade under the tent flap, but I got the itch and just HAVE to scratch!  ;-))  )

What IF the S@H "personal statistics" were enhanced with a simple "acceptance rate" number.  IF, as has been said several times in this and other forums, crosschecking is being performed to prevent "bad", unrepeatable results (independent of cause!) from sneaking into the "science" of S@H, why not publish the "reject rate". 

[snip]

Yep, I'm listening (a few percent of the time :) ).  We're thinking about something like this, but right now its priority is quite a bit behind getting some new science results up on the site.  We've finally got our new "post-processing" database machine up and running, so we'll be able to have waterfall plots/RFI checking results/persistence checking results available on the web.  Access to the "canonical result" for a workunit you processed would be available, and you would be able to see its signals and the signals from adjacent/prior/later workunits and which ones have been flagged as RFI.  You'd also be able to click to other workunits for the same region of sky and same frequency coverage.  If you've kept your results, comparison against the canonical would give a motivated user a way to check it out.

User stats updates to show reliability, etc. would come after these items. I'm hoping to implement an "overclocker warning" sooner, but it may have to wait until we have undergrads available to do the job.

Eric

On top of that there was a new entry in the Technical News page on the S@H server.  This post gave a bit on an overview on what they were doing to the data, and also announced that they have a new server and they are getting it configured for the post processing of the data.  If you are interested in how they have things set up and some of the hardware that they are using on the project, go check it out!

Closing In
It is about that time.  Team Lamb Chop has been closing in slowly but surely on Sun Microsystems, and it's about that time.  At the time of the April 1 update, TLC was only about 27,000 work units behind Sun.  At the current team rates, TLC should pass up Sun sometime in the next 4-6 days.  The team has been through a bunch in the past couple of months, but this time things will be quite sweet since the work units are legit, and it has been a total team effort. 

In addition to the upcoming passing of Sun, TLC is about to mark another milestone in the next day or so.  TLC will soon pass the 4000 user mark.  Not a bad total eh?  That total is something like 3x the total members when I first joined the team quite a while ago!

Busted?
About a week or so ago Michael (from the SkazDaLimit team) posted on the DC forum about a questionable user/team that showed up on the top 200 teams overall.  This team was named Ma Bell.  There were only two members on the team, and the one member (cant remember his SETI name now) was posting around 4400 work units a day.  Why would this person be questioned?  Well from the team page the work units being crunched were averaging round 12.5 hours per work unit.  That would mean that if legit, this person would be running SETI on over 2000 machines (CPUs) all day all the time.  It seemed that the work units being sent in were not being done over all 24 hours.  I checked a couple of days ago and they had sent in 1179 work units in a period of almost 5 hours.  If they were crunching all day long that would have meant a total of 5600 WUs a day....so even that 2000 machines would have been an underestimate.  The team name "Ma Bell" would indicate it would be a TeleCo team, and from the team description it turns out the description was a slogan for Ameritech.  The logo on the page was also an Ameritech logo, but that logo is linked from the Ameritech.com front page (so anyone could have linked from there).

After hearing of this, I passed the info on to Eric Korpela to let him know.  Yes running on 2K machines is possible, but not probable.  If there was something strange going on then Eric could find out.  If nothing was wrong, then hey more power to them!  A couple of days later I received a reply from Eric saying that the person's email address was from a "very large (phone) company", and at a cursory look, the results looked legit and they matched results that others sent in.  He was going to look a bit deeper to make sure.

I haven't heard anything back from Eric, but a couple of days ago I noticed something on the stats.  The team members are gone from the team....but oddly the Team is still in the top 200 teams.  In addition, the member that had been pushing out the work units *used* to be in the top 1000 members, but now this is all that is showing for that member:

That member's total has not changed in the past couple of days, and it looks like he got shut down, and removed from the team.  I doubt that this is the same hack that has been floating around, but something different.  The only thing I can think of may be duplicate (legitimate) results being sent to one account...but at the moment we may never know what is up (and I am sure that Eric may not want to let that info out either)

Dead Drive
Ug.  I have a Win2K dual box sitting here not crunching cuz my drive seems to have died.  I tried to boot it up, but when going into Win2K it just goes Clunk, Clunk, Clunk.  Oh well, the HD was only a 5GB drive, and I know it was on its last legs :/.  I think I will have to go and get a new drive to install.  Recently I *finally* got a totally automated stats update going, but it runs on this box (Win98), and I dont want to wait for an hour till the update finishes :).  I am going to move the automated updates to the dual 2K machine after I get the new drive...

Black & White
I didn't follow the hype.  I didn't rush out to buy it when it was released.  But I have it and I'm hooked.  This game is done by the same person that did the game Populous, but I really didn't care for that game much.  It seemed a bit repetitive, and became boring.  I wasn't going to buy it either, but then I was reading the Ars Gaming Forum and some of the things people were saying about the game got my interest, and I ended up buying it.  The graphics are good, the story is pretty good, and the gameplay is awesome.  In the past couple of days I have been pulling allnighters playing because it is so addictive for me.  There are so many things going on, but they are all interesting.  Buy it, and you (probably) won't be dissappointed.  Unfortunately with my dual box going down, and my girlfriend and I playing B&W all hours of the day, I posted a total of 0 work units yesterday :P  Oh well...

-Front Page