Archives

July 2000

Server Downtime Update
Hehehe downtime update...I made a funny.  Down Up...Get it?  Jumbo shrimp? Uh nevermind!  Just ran across this on the SETI@home page. 

NOTICE: We are in the middle of an extended outage today (Monday, 7/31/00). We currently plan to be back on line around 3:30am tomorrow morning (PDT). Until then, all CGI programs (user stats, team create/join, bug report forms) are shut off. See technical news reports.

Of course it may be a hour or so after they are up till you will be able to access them....but again...ya never know!

Check It!
Bonus Stats Today!  I thought with the server outage that everything on the SETI site would be down...but NO!  The stats pages were up so I was able to pull today's stats from the site.  Looks like everyone prepared and sent in their results before the servers went down.  Also today I updated the Weekly Stats.  Meerkat and Dyyryath keep moving on up...both of them moved up 20+ spots this week, and are looking for more.  Admins spend CPUs again was tops in work units for the week but there was a close race for second and third with svdsinner and Knight.  svdsinner edged out Knight for second though.  svdsinner was leading the way on the milestones this week.  He posted his 20,000th WU this week.  Congrats!  Other notable milestones include: VulTure's Den hitting the 10K mark, and Geordie_Cos380 and MrCatbert hitting the 5K mark.  wesley barnes and AnacalaSmaug lead the long list of new members in the past week.  I would like to give all new TLC members a warm welcome!

On the top 50 front, Beryllium keeps on moving up.  He moved up two spots in the past couple of days and now is sitting at #9 overall.  aenima slipped past Nsteussy for the #14 spot, while wesley barnes passed up Turbo for #18.  Knight keeps cranking them out and passing people by the handfull.  He moved up 3 spots in the past couple of days into the #21 spot.  Greens In Oz dumped a bunch of work units today...398 in fact.  With that bomb dropped they moved past Joker into the #25 spot.  Daryle A. Tilroe moved up one also into the the #29 spot.  XYZZY moved up a net one in the past week to place himself at #34.  Geordie and MrCatbert keeps switching places but Geordie has the best of it at #38.  JerkyChew slid by at least 4 in the past week and put himself in the #41 position,  Finally wiz_of_ez moved past _Quinn and into the #48 spot.

#4 When?
With the activity in the past week within the team and the new members, the date to pass Art Bell has crept sooner.  I don't have a firm date, but good "guess" would be sometime in the 2-4th week of August.  Maybe we will pass them before Rat's birthday! :-)

The Outage Has Begun!                                     
Hope you have things cached and are crunching away right now.  The servers are down and who knows when they will be up.  Most likely there will not be a stats update today.  If they get the site back up before maybe 3am I may be able to do an update.  If not, then don't expect anything till tomorrow.  When the servers are back up, expect to see the usual slow down in sending in those queued up results!

Athlon and Duron Still Laggin'
I reported a couple of days ago that Ancala did the beta 2.7 bench WU with a 700Mhz Duron in 5:07.  I went and clocked my PIII 650E up to 700MHz (2-2-2 timing also) and ran the benchmark.  Also add into the mix a time sent in today from Chuck Larson with a 700Mhz Athlon (classic) and you get the following chart:

* Chuck said his wifey snuck in and was surfing for about 15 min during this time, so this may be a little longer than a clean run would give

This is quite interesting...I thought that the Duron and Athlon would perform a bit better than what it is showing here.  I thrown in at the end there a "predicted" value of a Celeron (not a Celeron II) would perform at 700, based on the info from Rat Bastard's Benchmark...and it shows that the Celeron would do better than an Athlon!  Of course a benchmark of an actual Celeron @ 700 would be nice :).

With the new beta being due sometime this week, I wouldn't expect these times to vary that much.  The benchmark WU has an angle_range > 0.1 where the bimodal times start to come into effect.  The slower L2 bus and memory bandwidth seems to be evident again here with the AMD products...  -zAmboni

Kick Ass Site of the Day
This site is for the sky gazers out there.  It is called J-Track 2.5.  What is it?  It is a java based site which shows realtime the positions and orbits of a bunch o' spacecraft in the sky.  The positions and orbits are superimposed above a flat map of the earth.  You can choose to show the major NASA craft, some amateur or weather craft.  The coolness doesn't stop there.  You can bring up J-Track 3D and it gives you a 3D representation of the earth and the spacecraft orbiting around the earth.   Click on the pic to move the entire scene around with your mouse on all axes.  Shift click to zoom in or control click to zoom out.  Check out the ring of spacecraft sitting in geosynchronous orbits around the earth.....They are WAY out there!.  You can set the program to update the information from 1/4 of a second to 1 min intervals.  Then you can speed things up 100x or 1000x and set everything in motion.  You basically have to see it for yourself to see just how kickin' this is.  The final major part of the site is called J-Pass.  On this page you can either enter your zip code or your latitude/longitude coordinates and will be whisked to a page showing a of the stars above in a 360 degree range.  It will be at a time where the first sign of the spacecraft will come into view.  It also shows the track the spacecraft will take across your view.  You can bring up a list of all the spacecraft and the time range that will bring them into view.  Click on the time bar and that will bring that craft and the path that craft will take across your field of vision.  It is a great tool if you want to find out when and where in the sky the Space Station (or the Shuttle if up) will be, and plan on going out to take a look.  Highly Recommended. 

New Version 3.0 Beta Next Week
There was an explanation for the "bimodal" run times reported on earlier.  It wasn't clear if this was expected or a bug....it appears to be more of a bug, this comes from Eric...    No not THAT Eric, but from Eric Heien:

We're pretty sure this is due to the pulse finding code running an excessive amount for workunits recorded while the telescope was not moving much.  The code was just fixed, and in the next beta as well as the final 3.0 release, all workunits should take roughly the same amount of time on a given computer.

Also, the bug involving the client halting about 3% into the WU and restarting has been fixed.  It was the result of a very tiny rounding error in some floating point units.  The next beta will probably be coming out sometime next week, barring any significant problems.

-Eric Heien
eheien@setiathome.berkeley.edu

One thing for sure....don't expect the new beta to be out on Monday ;)

PIII/BX Mobo Still Can't Be Beat?
When I reported in the Version 3.0 preview, that the new client will be more cache friendly, that appears to be correct...but with some more benchmark times, and further analysis  it appears that the client DOES seem to rely on memory bandwidth and cache size still.  The first question about the benchmark times that Rat Bastard reported (both a Katmai, and CuMine 550 reporting a 5:13) came fromt he SETI newsgroup, when a poster noted that IF the new client was cache independent, then the CuMine should be faster than the Katmai because of its full speed L2 Cache.  But that isn't happening.  Lawrence Kirby tried to explain this by saying that the version 3.0 client probably uses a "recursive subdivision" in the FFT routine.  What this means is that it either starts with small blocks and combines them into larger blocks or the opposite...start with a whole chunk of data and then splits it up into smaller blocks.  Here is how he broke down the "stages" as he calls them and how the data would be accessed:

From the benchmark times the full speed L2 cache and the 256 bit L2 bus compensates exactly for the extra hit on the slow memory that the CuMine must take compared to the Katmai (1/2 speed L2 cache and 64 bit L2 bus.  The Celeron's slower time can be explained by a combination of smaller L2 cache, the 64 bit L2 bus and the extra hits accessing system memory.

Where would this put the T-Bird and the Duron???
based on the above info the working set for the v 3.0 client is still > 512kb...but the T-Bird has 128kb L1 cache and 256kb of L2, and the Duron has 128kb L1 and 64kb L2 caches (L2 caches are at clock speed).  Because the caches on the T-bird and Duron are exclusive, they act like a total instead of the inclusive cache's of the Intel CPUs.  Based on Lawrence's post I can come up with these estimates (these very well may be totally absolutely wrong :-) :

I do want to remind you this is only a guess here.  With the above table how would the T-Bird and Duron perform compared to the CuMine?  Both the T-Bird and the Duron would have the advantage of a couple more stages in the very fast L1 Cache.  But the L2 Bus of the T-Bird and Duron is only 64 bit....plus it has the disadvantage of the memory bandwidth of the Athlon chipsets.  For sure the Duron would run slower than the CuMine, but would the T-Bird?  I would say no.

Well I can actually say for sure(?) that the answer is no with the T-Bird, because I have some benchmarks!  I have a couple of times sent to me from tim Wilkens who ran the benchmark work unit with the 2.70 beta with both a T-Bird and Athlon Classic.  I ran the benchmark on my machine with the beta also.  Here is the meat:

The CuMine is definitely kicking some ass.  But there are some caveats here.  Did you see how I slipped in a (?) when I said "for sure".  This may not really be that fair of a comparison.  Yes the PIII is kicking some serious ass, but take a look at that FSB setting.  I believe that this ass kicking is more due to the PIII cache's running at a significantly higher clock speed, and the memory bandwidth totally blowing away those of the AMD machines.  Because of this you cant really determine the cache dependencies of these different CPUs.  A better comparison would be for a PIII, T-Bird and Duron at similar multipliers and FSB settings.   Ancala has done the benchmark on the beta with a 700Mhz Duron and turned a 5:07.  Sometime in the next day or two I will clock back my PIII back down to 650 and do a bench with that setting to see how the PIII compares to the Duron.  Hey it isn't exact....but what the hell!  I will keep ya informed  -zAmboni

Sorry for the Delay...
The delay in the stats last night/today was due to my inability to access the team site via FTP.  Somehow the FTP access was turned off for anyone but the server admin *DOH*.  But the stats are up now...so what are you waiting for?

One Server Outage Out of the Way
The SETI guys got the router changed...but as the norm it took longer than expected.  Apparently they changed the router, fired everything up, and then the router crashed.  Go figure.  Then it took them a while to get the router fixed and back up.  This wasn't a S@H specific router though....I believe it was one in the general Berkeley networks.  Take note of this....Be prepared for the outage on Monday!  I set up SetiQ today on my machine, and hope to load it up with tons of work units....expect the server to be down for *at least* 24 hours on Monday, but be prepared for it to be down for longer!

Note About Monday
The Team Lamb Chop server was down for a bit more than an hour on Monday.  But, it was down for a reason other than what you may think.  It was all a matter of bad timing.  One of the developers at the hosting site screwed up the server database.  They when they rebooted the server after fixing things, there turned out to be corruption on one of the SCSI drives that they had to fix.  Unfortunately it had to happen about an hr or so after the site link was posted on Slashdot.   -zAmboni

Get Those Buffers Ready...
There are two upcoming server outages in the next several days.  Today (Thursday), they will be taking the servers down for approx 1/2 hour (starting at 13:30 UT)  to upgrade their router.  Next Monday is the biggie.  They are going to rebuild the science database.  This is the same thing they did on the previous long outage, and they expect this outage to be around 24 hours like the last one.  Be prepared..stow away some work units!

More New Stuff
I was a busy doin' some Excel VBA coding tonight, and added two things today.  On the Team Daily Stats page,  I have added a column to the Top WU/Day chart.  At the end of the chart I have added the member's work unit average (for the previous 7 days) and in parenthesis next to it is the sorted rank of the WU average over the entire team.  Also on the Member Charts I have also added a sorted rank for the WU average.  -zAmboni

Today's Stats...An Explanation
If you notice on today's Top WU/Day graph and chart, it shows that msearcy cranked out 350 work units for today.  Well that really isn't true.  That number turns out to be an "artifact" with my VBA coding for Excel, and there really isn't too much I can do about it really.  Strange things tend to happen in the stats when 1) someone changes their nickname or 2) when someone joins/leaves the team with the same name as someone else on the team.  The example for msearcy is #2 above.

What happened is the the following:  There was a member on TLC with the nickname msearcy who was at position #1169 with 73 work units.  Today there was a new member who joined with the name msearcy also.  This person joined the team at #463 and brought along 423 work units....well how do you handle a situation like this?  My solution with the VBA coding is, if there are two members with the same (exact) name, the member with the most work units gets the "Name".  The member with less work units gets "Name (#1)" and down the list if there are more than two with the same name.  (There are over 100 members right now who don't even have a name and are listed "blank", "(#1)", "(#2)"..."(#n)").  In effect the recently joined msearcy took over the stats from the previous msearcy and voila! 423 - 73 = 350 WU "produced"!  Now the old msearcy is now called msearcy (#1), and because there previously wasn't a msearcy (#1) on the team yesterday, the VBA code see this person as the "new" member.

Situation 1) above happened with yesterday's stats (if you noticed).  Team member Exchequer (with a single space at the end of his name) changed his name to Exchequer (without a single space at the end of his name)...The VBA code doesn't know that he only changed his name by getting rid of a space, and took this as a member leaving the team, and as a new member joining the team, and it displayed this on the members joined and members left chart.  Luckily weird things like this doesn't happen too often...so if you see something strange, this may be an explanation!

Other Daily Stats Stuff
Another 27 members joined the team today.  Tops on that list was AncalaSmaug ranking in at #104 with 2071 WUs.  There were some team movement in the top 50 today again.  UltraMaroon passed up OoklaTheMok for the #9 spot today.  Knight moved up 3 spots today passing Joker for #27.  The Weeblows blew by mjdewitt for #34, and finally JerkyChew moved up one spot to #45.  Way to go guys!

So many people have joined the team this week, I need to make another chart page that goes from 2401-2500!  You all sure know how to keep me busy!   

PCI SETI Add-On Card a Hoax
The skeptics were right.  Check it out: http://www.krasnoconv.com/truth.html.  Here's a snippet:

In the middle of June we had a discussion about how far SETI crunchers would go to help SETI@home. And, of course, to boost their stats. How much would they pay? Everyone can just walk into a shop and buy a computer. Or more than one. But a complete computer is simply too expensive to buy just for work unit crunching. A software solution? No, this would have turned out as just another patch. So we decided to invent a hardware add on. A card which could be installed in the computer: The SETI accelerator PCI board was born.

To be honest, I had doubts from the beginning...but I thought, "If they could pull this off it would be a cool product".  Doubts crept in for me reading some of the /. comments about it being a hoax.  The final clues for me came earlier yesterday.  A search for "PR-964" came up with the product number for a Star Wars book by Dave Wolverton on a German Sci-Fi Shop, and a search for "SAMOWAR" came up with some type of German tea kettle (A tempest in a teapot?).  But sometimes you play along because they cant be lying, can they?

The final laugh may be on them.  With the future of distributed projects being pay for crunching, there will be some company to develop dedicated add on cards for crunching work units, blocks or whatever.  An idea they came up for a hoax could be an idea to make someone some cashola.

TLC Welcomes More Members!
Team Lamb Chop had another run of members joining up for the team yesterday.  There were 40+ members who decided to join.  Top of the new members is wesley barnes who debuted in the #18 spot, joining with 8700+ WUs.  Welcome One and ALL!

New Feature on the S@H Pages
...and once you are on Team Lamb Chop I sure hope you don't use it!  :-).  It was pointed out on the SETI newsgroups that they added a Quit option on the team pages.  Previously if you had joined a team and wanted to leave it, you had no options other than to join another team.  Apparently there were people who wanted to leave a team and not join another team.  This option allows you to do that.  But you won't do that will ya?  Don't you know that Team Lamb Chop wuvs ya and doesn't want ya to leave?  ;-)   -zAmboni

Weekly Stats Update!
Just posted the weekly stats stuff.  Whose the big mover this week?  None other than Meerkat.  He moved up 42 some odd places this week to put him at #185.  mckenney was second moving up 41 spots to #287, and Ackbar move up 34 spots to #289.  Its same as usual with the top producer for the week....and that was...drumroll please?  Admins spend CPUs!  WOOHOO! They crunched out 2151 work units for the week.  Knight tried as he could but couldn't topple the king.  Knight was a shade under 1200 WU for the week.  In third was svdsinner, he cranked out a bit under 1000 for the week, but he did move up to #5 today!.

Also check out the milestones for the week....there were plenty of members who reached some pretty nice milestones.  If you cant remember....that is the weekly stats page!  ;-).  Even though its on the bottom you can check the progress of the top crunchers and where they place in the overall scheme of things.  There is a lot of cris-crossing going on with the team.  I dont know about you...but I think that graph looks pretty cool!  -zAmboni

Prepare Those Buffers!  Uh...guess not!
Looks as if the server outage has been cancelled...not sure when they will do it, but be prepared anyways!

Taken from the S@H Technical News Page is this notice for a server outage tomorrow...even though they say it should be a half hour, you just never know!:

July 24th, 2000
Update: The outage for tomorrow has been cancelled by the building admin.

There will be a 1/2 hour outage tomorrow, 7/25/00, beginning at 13:30 UT. This outage is due to a router upgrade in the building where the server is housed.

Today's Stats
We have had quite a bit of additions today to the team.  At the time of today's update there were a total of 54 new members to the team.   Welcome Aboard!

There has been a bit of movement in the top 50 today, most notably we have svdsinner surging past Gonzo and into the team #5 spot.  He isn't about to stop there though...he is projected to reach the #3 spot on TLC with the month!  Beryllium also moved up one today, he passed the long idle SouthPark Admin Crew for the #11 spot.  He is set to move up into the top 10 soon.  Action passed up 77rickyli for the #21 spot.  Greens in Oz almost posted a 300 spot today, with that posting, he moved up 5 spots today to #23.  Lurking in the midst though is Knight.  He has been cranking some serious numbers over the past couple of weeks, he is sitting at #29 right now, but is slated to move into the top 10 within a month.  Both Beyond and JerkyChew moved up two spots also today into #39 and #45 place respectively.

Hopefully sometime tonight I should have the weekly update posted...I'll let ya know when it is up.  -zAmboni

PCI Card Update
The story on July 22 about the PCI SETI Card add-on continues.  The site was slashdotted late last night, and they appear to be feeling the heat.  With the combined questioning on the alt.sci.seti newsgroup and the slashdot article has led them to change some info in their FAQ.  They added this bit to their FAQ:

The client must be downloaded from Berkeley's download area, it is not part of the package due to copyright restrictions. Sorry for that misunderstanding.

With this change, it seems that the add on card would comply with the SETI Licensing Agreement.  But there are a lot of people on slashdot questioning whether or not this whole thing is a Hoax.  Comments range from the site registrant being from Germany, an Adaptec logo appearing on their card pics, to some photoshop editing done to place the 6 CPUs on the "card".  Team Lamb Chop member ColinT actually ordered a card when he first heard about it, and they said it would be shipped in "6-8 weeks".  Today he was notified that his order was cancelled because they ran out.  There are other reports of people having their orders canceled.  Each of them received a standard info letter (thanks to jonnyv and lizard for the linkage..KWSN Rocks!). 

My take on all of this?  A company has an idea...they may have developed a test card and set up a site to see how much interest this would generate.  They put up an ordering site to see how many would commit to spend cashola on the cards.  Now with all the publicity they are stuck, they have orders, there are questions, they don't have any cards to ship.  From the email, they don't even have military clearance to ship the cards!  I dub the PCI Add-On Card as Vaporware!  When someone has a card in hand and it works, then maybe it will warrant further discussion.  -zAmboni

New Things on the SETI@Home Site:
I had noticed this a couple of weeks ago, but kept forgetting to include it in the news.  There are a couple of changes on the Individual Stats pages, which now include some User Milestones.  The page now includes links to recent users passing the 10,000, 5,000, 1,000, 500, 250, and 100 WU Milestones.  If you recently passed one (like Angus at 5000 WUs) go check it out!  You just may be famous!

Also just noticed today, they have started a "Cruncher Of The Week".  It looks like it is a randomly selected S@H member who has completed at least 1,000 work units.  They should give them a S@H mug or something as a prize.....or maybe they can just give him a "no prize" :-)

-zAmboni

Initial Beta Benchmark Results
That's right! Thanks to the generous contributions of MadMac (who sent me a slotket to use for testing purposes), I've been able fo complete the preliminary benchmarks on the new 2.7 Beta client. Okay, so I tested it on a couple CPUs. However, if you're here, you probably have some interest in how fast it runs. So go find out! -Rat

New SETI & RC5 Distributed Forum
Yup It's here!  Join in on the fun!  Check out the new Distributed Computing Forum over at Ars!  Back home where we should be!  This will mean the death to the team forum over here...but its for the best.  Ciao! -zAmboni

Following In The Footsteps
Yesterday marked my 1 year anniversary of crunching SETI@Home work units.  Every minute of the way it was with Team Lamb Chop.  A couple of weeks ago Rat Bastard posted his 1 year retrospective, and now it is time for mine.  Take a look on my view of the past year, and why I think that Team Lamb Chop is da Bomb!  -zAmboni

Pretty Darn Innovative....But Is It Legal?
I originally missed this discussion, but then read up on it.  The discussion was talking about co-processors to do S@H crunching on a computer.  There was a reply with this link in it:  www.krasnoconv.com.  This site is advertising an add in PCI card for computers which contains anywhere from 1-6 CPUs, and is specifically designed to crunch SETI work units. 

There are two different cards, a single CPU card and a 1-6 CPU upgradeable card.   The cards contain 2x64 kB level 1 cache and 32MB RAM per CPU.  The CPU will run at 6.5 x the PCI clock. At the default PCI clock that woud mean a 214 MHz CPU (the CPUs have been tested up to 40MHz on the PCI Bus).  As for crunching the work units, onboard ROM chips would contain the "unmodified" linux client, as well as the required routines of linux itself.  The card ROM is flashable so when there is an upgrade in the SETI client it can be flashed into the card.

The site claims it can process 1 work unit in 15 hours, and at the most, 6 work units in 16 hours (one for each CPU installed).  The cost?  $89 for the single CPU card, $129 for the upgradable card (with one CPU) and $69 for each additional CPU.

I think that this is a pretty innovative solution.  First off because it is using CPUs harvested from surplus military hardware.  Specifically from target vector portion of the terrain following radar of cruise missiles.  With the trend leaning toward using distributed projects as a solution for large scale computing, plus many of them planning to pay for crunching data, this may help people make some cash on the side.

The problem I have with this personally is that this solution may not abide to the licensing agreement for the SETI@Home project.  Especially this portion:  "Distribution of this software is prohibited".  The company's FAQ sort of diverts this issue saying that this wasn't a patch and only uses an unmodified version of the client, but unless the company has an agreement with the SETI@Home team, they would be distributing the clients against the restrictions mentioned in the licensing agreement.   As of this point I don't know if they do have permission to do this.  I do not think that they do though.  I cannot see the SETI@Home guys (who have a hard time raising fundage for their project) allowing a company to make a profit out of their work.  Only time will tell to see how all of this plays out...legal proceedings may be a problem since the company is based in the Ukraine.  I'll keep ya posted what I find out.  -zAmboni

Beta Slow Work Units...an Explanation
So far it appears that the work units times that people have been getting for the beta 2.70 client are not out of the norm.  The different run times it turns out to be an inherent difference between the earlier 2.0 and 2.4 clients and the new beta.  Run times on the version 2.x clients (for my machine) were quite consistent, at the most an hour difference between the fastest and slowest work units.  With the beta though, the run times have been quite variable.  Some people on the newsgroups call it a "bimodal" distribution of run times.  In one group the times have been around the 3-4 hour mark (on a PIII @ 975MHz), and the other has been running at 6-7 hours for a work unit.  There *IS* an explanation, and it doesn't mean the beta is broken....it is working the way it should.

With the version 2.x clients the short run times were usually work units that did not contain gaussians, and ran  faster.  Those work units with one or many different gaussians would run slower.  With the beta on the other hand, the work units without gaussians have been the ones running the longest.  I will let Eric K. explain for ya...

Under the old client these would have been "fast" work units, because gaussian finding isn't done on them.  It's a different situation on the new client, and we haven't yet decided what to do about it.

Let me explain the situation.  After the FFT's are done, the data is broken up into chunks the width of a beam for pulse finding, if that chunk is longer than 15 points, and shorter than 40961 points, pulse finding is done.  The problem is that length of those chunks is inversely proportional to the slew rate.  Slow slew=longer chunks.  At siderial slew rate, there are less than 16 points in a chunk for long FFTs (longer than 8k give or take).  At zero slew rate, you get more than 15 points at a 64K transform.  So for zero slew rate, we do a lot more pulse searches.  We also do longer ones.  The longest pulse array we search in a normal slew rate work unit is about 25000 points.  The longest we do in a zero slew work unit is 32768 points.  Longer arrays take longer to search.  Both of these add up to a longer run time.  Of course, the longer you watch the same point, the easier it is to detect a pulse, so the pulse search in these zero slew work units is more than twice as sensitive as the same search in a normal work unit. So the question is, is the additional time worth it?  

I know what you are saying....wha?  huh?....I don't fully understand either.  But I do catch the drift of it :).  If the extra science is going to mean a time penalty...go for it.  BUT...this situation does bring up some possible problems.   The angle range is listed in the work_unit.sah file.  It would not be that hard to set up a program, batch file, etc. that could check the angle range, and if it was a low number, delete the work unit and download another.  This way they only would work on the faster work units.  Kick up the stats numbers.  I believe that the S@H guys realize this also, and they will end up doing one of two things.  1) scale back the science part a tad to get a bit faster run times on the low angle range work units....or 2) add a fudge factor to the processing that makes the faster work units run a bit slower.  I think that #1 would be more likely.  The only way to find out from the next beta, or in the final version 3.0 client.   -zAmboni

Team News!
Most of the team feels a bit lower than normal today.  That is understandable, because they had to take one step back, because rosborn joined the team yesterday.  he/she/they joined in at #19 with almost 7800 work units.  Welcome aboard!

That funny smell that you may have noticed was just Herb-Man rising up through the ranks.  He tried out a potent blend today and ended up with a whopping 219 work units.  That put him in second place for WU for the past day.  That was a nice change from his 6-7 work unit average he had been doing.  With that burst he moved up a whopping 11 places and into the top 100 at #93.  Way to go!

There were several in the top 50 who didn't get knocked down in rosborn's arrival.  Joker and Kevin N. Carpenter remained in place at the #25 and #45 positions.  JerkyChew resisted the downward flow and went against the norm moving up one spot into the #47 position.  With a work unit average of 95 per day, he will be moving up some more pretty soon!   -zAmboni

How to keep the wife from finding out why you REALLY got her a PIII
It's happened to all of us. You feel the need to rise higher in the ranks, and you can't get away with buying another computer. So you secretly upgrade the internals of one, and don't tell anyone that the 200Mhz Cyrix has been replaced by a Dual 550 Xeon. Of course, this means that you have to hide Seti too. And JonT has picked up a pen (figuratively speaking) and whipped up a handy dandy manual on installing Setilog and Setispy, to keep track of the client, and hiding the client as well. So if you're in the need to be in the know- here's your fix.  -Rat

Version 3.0 Preview
Just posted a preview of the version 3.0 client, based on the v2.66 and v2.70 betas that have been released in the past couple of weeks.  Check out what to expect on the new clients. -zAmboni

ChchchchchChanges
There are a bunch of changes in the stats portion of the page.  The first one is big one.  I am retiring the Stats News page.  The stats new isn't completely gone, I am going to be integrating the stats news with the front page news here.  Its more of a design move (less clicks for everyone!), plus I think that the Team Stats News should be up front for *everyone* to see. 

Additions - On the Stats Home page, there is now a listing and explanation of the different stats pages.  Also on the stats home page are graphs showing our progress in the annihilation of Art Bell, and a chart showing the new members to Team Lamb Chop for the day.   Now included in the Member Charts area are charts showing sorted projected rankings and work units for both team members 1-100, and 101-200.  On the Team Daily stats page, I have now included a chart showing the top 30 member WU/day.  This goes along with a graphical representation for the top 10 member WU/day.

I hope you enjoy the changes.  It is my duty to bring you the best :)....now you can't say that I don't cater to suggestions!  -zAmboni

TLC Forum Update
I received word today, that the new Distrubuted Forum on the Ars Open Fourm, should be coming soon.  How soon?  Not sure.  Apparently it was to be added...but with the Open Forum implosion last night, the addition is being held off for a couple of days to see how stable the forum is with the changes they applied.  Patience, grasshoppa!  -zAmboni

Updated Links
I updated the links page, well mostly because it was in the old uglier page format.  Let me know if there are other links that you think should be added, I may have missed a few.  I also changed the navbar a tad, just some spacing things, but nothing that you may notice :)  -zAmboni

Beta v2.70 Results
After letting the new beta client run overnight, It has processed through a total of 4 work units and is currently on a fifth.  I don't have the beta client set up to log results, so I have taken the data from my account info on the SETI site.  Here are the times and results from the work units sent in today:

The current work unit my client is running is estimated to be done with a 4hr total time.  This is on a PIII 650E running at 925MHz.   Those are some fast times even coming from a 900+ MHz machine.  It looks like they definitely made the client more cache friendly.  And with the new beta I will make a prediction on Rat Bastard's proposed experiment.  Here it goes:  The beta client will run faster on a 600+ MHz PIII than on a 1MB Xeon.  I have a feeling that the cache dependency is gone and now the SETI client performance will largely depend on raw CPU speed.

Except for the 6:46 time, the others are significantly faster than with the 2.0 CLI version that I was running previously.   The consensus on the SETI newsgroups are the same, that the beta client is faster than the previously released clients.  There hasn't been any real bug reports on the beta yet, other than some possible minor graphics glitches.  But there was one unconfirmed report, that the client may not be sending in the same results that a version 2.x client does.  I may investigate this sometime tonight.

I am still having my reservations about the beta client right now solely due to the fact that it is running faster than the 2.0 and 2.4 versions out now.  The main reason is that they expected the version 3.0 client to run about 25% slower than the 2.x client.  Since the beta is running faster, is this because the client is that more efficient?  Or is it because it is broken science wise?  This brings up one last thought.  Do the SETI@Home folks really want a client that runs faster than the 2.x client?  Currently the servers are backlogged in processing the stats.  Having a faster client will only make things worse for them.

I would not be surprised at all if the release version 3.0 client runs slower than the current beta.   -zAmboni

Chat With Team Lamb Chop!
Rayban has set up an IRC server that you can join and chat with other Team Lamb Chop persona.  If you haven't already, download the IRC Client, connect to the server rayban.dhs.org:6667, and join channel #TLC.  Easy as that.   Also if you haven't heard, there is a Team Lamb Chop ICQ Active List which can be joined.  The number for the ICQ Active List is 81362123   -zAmboni

A Strange Request
With the release of the 2.70 beta client, and the resulting stability of it, it's time to start some testing! However, before everyone gets too excited and starts sending in times from one hundred different machines, none of which are related, I'd like to point out that the 2.70 client is still beta. And that I'm not going to be maintaining a huge database of beta times. Instead, I would like to do the following: test the client on ONE machine, possibly two, under controlled circumstances. I think the interesting thing with the new client is the new FFT routine, and it's lower usage of the L2 cache. So what I'd like to do is test it with caches ranging from 128K to 1M, all on an Intel platform, to try and keep it even. To make things even more even, I'd like to test the 128K, 256k, and 512K on the same board, same install. At the same speed and FSB. Of course, the 1M cache requires a Xeon, so that's bound to be different. 

Getting back to my point. I have a 550MHz Celeron, Coppermine, and Katmai. I can run all of them in one machine, and turn out times for an otherwise identical system- which is the best test. Only one thing prevents me from doing this- I don't have a slotket adapter for the Celeron. So- if anyone in the Minneapolis/St Paul area wants to come by and hang out for a while- drop me an email. Or if you want to come by two days in a row (before and after), that will work good too. In fact, that would work better, just in case it takes 12 hours to run the work unit. After all,we can only speak Geek for so long.

Of course, the one thing this leaves out of the equation is the 1M L2 cache. So if anybody would like to run a quick test of a single instance on a 550 Xeon, and mail me the results, I'd be very appreciative. That's all that I have for now- be sure to stop by the forum to get some more info on the beta, as well as everything that's fit to rant about. Or not. -Rat

Forum Changes
One other quick note- I've made a couple changes to the Forum. The first is the renaming of the Misc category to The Rant. Why did I pick that name? Because I can. The second thing is the Elimination of the UBB Improvement section of the forum. The reason this is shut down is simple. 1)There were too many forums to for the level of posts. 2)There were no posts in that forum since the 6th, other than my notice of it's closing. 3)With the current forum plan being a move back to Ars Technica, we no longer need to consider what to do with our forum here. 
So, with the changes in the forum, it's necessary to update all the threads. I'm not sure how long it will take with the number of posts we have (probably not too long), but it will be down until the update is complete. Since this is being done late at night, you'll probably sleep through it. -Rat

Forum update complete!
The forum update is done, we apologize for the immense inconvience that's it's downtime may have proven to have been. For those who lost hope during the one minute that it took to update, feel assured that all is now well. One nice thing about our forum versus Ars Technica's- small is fast. -Rat

Beta v2.70 a Screamer?
Here is an update....I installed the new beta and I am currently running it minimized, and tracking the progress with SETIWatch.

With 10.745% done, the completed CPU time is 30:04, and estimated time left to be 4:09:45.  A total of ~4:40 for a work unit using the beta GUI.  (this is on a PIII650E @925 (142FSB)) .  For comparison the last work unit I did today with the 2.00 CLI finished the work unit in 4:57.   -zAmboni

It appears that the faster FFT routine is working its magic, and the rumors of a more cache friendly version also rings true.  This is contrary to the original estimate of the version 3.0 client to run ~25% longer than the 2.x version.  But one also has to remember what was mentioned in the beta email.....this is still a beta, and there will probably be a couple other revisions to come.  I am really itching to see how the command line client will run!   I shall update this as the work unit processes and nears completion.   -zAmboni

New Beta is Out!
...and the early reports are positive.  There hasn't been too much detail on the systems they are running on except for one post:

Right now it's a little too early to say for sure (I'm just at 10% done), but at this pace it should finish a WU in 5:00. Version 2.04 used to take 7:30, and version 2.66 about 24:00. So far, the new beta is great!  Athlon 700@900, Windows 2000.

That time seems a tad on the fast side for an Athlon....could Win2K be a SETI screamer for version 3.0?

I haven't loaded the new beta yet so I cant tell ya first hand how it will perform.  I may do that tonight and let it run uninterrupted overnight.  Here is part of the email sent to announce the release of the new beta:

FOR THOSE OF YOU WHO ALREADY DOWNLOADED 2.66 FOR WINDOWS:

We realized right away that this version was much slower than previous versions, and quickly made a few builds to figure out what was causing the painful slowdown. We found the culprit, and 2.70 is built without it. You should notice a vast decrease in CPU time per work unit, but may also notice some drawing errors.

SOMETHING ELSE TO CHECK FOR:

We have had complaints from Mac users that funny values are corrupting the state.sah file, yeilding oddly negative or out-of-bounds results.  On both MAC and WINDOWS systems, please try to recreate this problem - our guess is that it is being caused by frequent client stop/restarts that introduce random errors into the state file. Please exercise this by continually quitting the client (on windows, you right-click on the system tray icon and select exit) and restarting.

Thanks for all your testing. We're still fixing some bugs, so there may be another beta or two before release, but we're really close!       -zAmboni

For the Serious SETI Cruncher
Now you can go anywhere and keep track of your personal stats!

BAnywhere.com  has developed a free application called BaAlien that allows users to access SETI processing statistics or a SETI@home users personal statistics from any wireless enabled Palm computer.

It is available at http://www.banywhere.com

-zAmboni

New Beta Monday?
With all the server excitement several days ago the release of the new beta version 3.0 client (now dubbed 2.70) has been pushed back to Monday. 

I'm sorry to report that this weel's excitement is going to prevent us from getting the new Windows beta version (2.70) to all our testers this week.  It's in limited release right now.  It's likely that we'll have it out on Monday, though.

How will you find out about the new beta?  I am not sure on this though.  Last time they made an announcement on the alt.sci.seti newsgroup to sign up.   Eric was hinting that those who signed up for the original beta will be getting an email stating the new beta is going to be out.  But he is not absolutely sure since it is Matt Lebofsky's job to do that.

How is the beta supposed to perform?  Eric chimed in that with the limited beta release of 2.70, it performed about 3 times faster than the previous beta (2.66).

There has also been some questions on whether or not they will be going back and re-processing old work units with the new client (with pulse and triplet searching).  The answer for now is no.  But that does not rule out them changing their minds sometime in the future.   -zAmboni

Last 10 Results Wrong?
Not necessarily.  There are some people who have noticed that after the server outage that some of their returned results did not show up in the last 10 results on their user account page.  This was fallout from the server outage and them reinstalling the backup database.  They had to go back to a previous backup.  If a work unit was sent that was generated after the backup and before the fix, the database didn't know what to do with the results.  The results were not added to the database and stuck in a different directory.  The user was credited with the work unit being done, but did not show up in the last 10 results.  -zAmboni

Why Informix?
WIth the recent database problems (last week wasn't the first time), there were some newsgroup members who were questioning their choice of Informix for their database over, say, Oracle.  The answer is simple.  Money.  When a project starts out with a limited amount of fundage, sometimes you have to go with the cheapest solution.

>Goodness, you guys have all the headaches of a major development
>company.  -:)  Just curious, what was the deciding factor in using
>Informix as a database?  Some set of features?  Sponsorship?  Or was it
>simply what was readily available at the time?

Mainly sponsorship.  Given that, it was what we could afford (i.e. free).
Hard to say no to a deal like that. :)

Eric

-zAmboni

-Front Page