Archives

May 2000

Tuesday! Tuesday! Tuesday! Update! Update! Update!
All The Kings Are Coming! Coming! Coming! Coming!
I have a whopping update for you guys today...Most of it has to do with some ramblings and discussions going on over on the alt.sci.seti, and sci.astro.seti newsgroups.  Now that I have your attention...lets get to it!

Hiram 'O Hiram!
Hiram has been pretty busy on the newsgroups recently with lots of good info.  In a thread named "SETI on a Cray???" there was some questions dealing with the naming of the clients, and the confusion therein with the i386, i486, i686, etc. nomenclature.   It appears that the names of the client are more out of convenience and a way of differentiating between the clients:

The so-called i386 clients are most likely optimized for Pentium class machines.  Especially if they were done with a good compiler that can do that.  They just so happen to also run on 386 machines too, hence the name.

I have been thinking about trying to fix up the names for the v3 release given this constant confusion.

Currently most of the names are blindly chosen by the autoconf process which doesn't really do the job that we need done here. A few of them are manually adjusted after autoconf makes a first stab.

He goes on to explain that he hopes to pare down the numbers of different clients...but he is unsure if "combining" different clients will be feasible in the porting process. 

I am somewhat inclined toward having all the names to be simply: x86-whatever  Anyone else have a brilliant idea of how to name these things ?

He also provides a list of average work units produced per day by the different clients.  The CLI has settled in nicely as the #2 producing client behind the Win98 GUI (sorry Mac!)

In the same thread there was also some talk about the version 3.0 client.  The news sounds good (except for the 10-20% performance hit), but don't expect to see the new client soon (I don't think it has been beta tested, nor sent to the porters yet)

The v3 client is pretty much in final form at this point.
The additional new science code is being adjusted so that the expected WU completion times will be similar to current WU completion times.  It will probably be a little bit longer, maybe 10 to 20 %.  Your milage may vary. There is a new FFT algorithm which is faster, and thus allows more science to be done.  There is now a spike detection phase in the processing.  I have not made myself familiar with the details of these changes.  My concern is easing the job of the porters so I have been adjusting tiny items that allow the client to compile out of the box with minimum interaction.

In a different thread, Hiram pointed out a link over at distributed.net talking about overclocking and their project(s).  I don't necessarily agree with some of what they state, but again the distributed.net projects have different goals and doesn't have the redundancy that the S@H project has.

Re: overclocking
I note this information in the FAQ at the distributed.net site: http://n0cgi.distributed.net/faq/cache/148.html

That is interesting that a single error by an overclocked system could ruin their entire project.  At least S@H has some redundancy to cover client errors.

To Open Source, or not to Open Source
There is a good discussion going on about the merits/non-merits of releasing the SETI code via Open Sourcing.  I really don't know anything about open sourcing and the supposed benefits/downfalls of it.  Hiram started it off with this post under "The open source code problem" and a healthy discussion has ensued:

For the open source code debate, I hereby offer the discussion about this problem from the distributed.net site: http://www.distributed.net/source/specs/opcodeauth.html

This is a very interesting problem and is worthy of research.

Join Now!  Win Prizes!
I am not sure if this is a good trend or not.  First there was The Register "contest" a couple of months ago...now several sites have started giving out prizes for participation, referrals, or work unit crunching for their teams.  The first in this vein was the Trust No One team, and now it seems that the Overclockers Network has done the same.  Personally I don't like the idea of such contests.  I tend to think the idea of Teams in the SETI@Home project is to get a bunch of people together and crunch work units as a "community".   Just how do contests and giveaways help promote a sense of community?  Are they just trying to bribe people to increase membership and work unit totals because they couldn't do so otherwise?

Linkage!
Team Lamb Chop has gotten linked up over at bbspot.com!  I think the recent influx of new members have been due to link click throughs from their site.  The word is spreading!

Finally...
The work unit totals for the individual teams have not been accurate for a couple of weeks.  Somewhere, somehow there is a backlog of results that haven't been posted to the team totals.  The derived team stats and pages on the SETI@Home site reflect these inaccurate numbers.  If you want to see "more accurate" team numbers, take a look at TKWSN "enhanced" stats page.  This page gets the team totals from the sum of the individual team members.  One thing to note is that the work unit totals for Art Bell are most likely underestimated.  The team pages on the SETI servers only show the top 9,999 members, and currently the team has over 12,000 members.

I am (slowly) working on improving the stats that I am keeping,  and may implement some of the same type of stats that TKWSN pages have....be patient, I am working on it :)  -zAmboni

Its Official!
With todays Overall Team update, it is official on the SETI servers that we have passed Intel (of course we actually passed them a couple of days ago :).  The actual team totals for the two teams are" TLC: 910414, Intel 896554.  We were helped out today in overtaking Intel by the addition of three members who immediately jumped into the top 200.  The addition of wiz_of_ez, Barf, and Thom Bouch boosted the team totals by over 5300 work units!  Art Bell?  Want some?  We are under 100,000 work units away from them right now (not sure how many because of the work unit backlogs...).  The current projected take over date from Art Bell is 6/22/00.  That date is most likely around the second week of June taking into account the work units that havent been posted because of the backlog.

To look a bit farther ahead....I have changed the Overtake Graph a bit to show the recent fall in overtake dates for the top three teams.  The dates for overtaking SGI, Compaq and Sun have steadily creeped downwards.  At current rates we would pass Compaq and Sun within a year!  And depending on how long the SETI@Home project lasts, be may also eventually pass SGI!  -zAmboni

Benchmarks, Xeons, and retarded monkeys
Okay, I lied, there are no retarded monkeys- I've decide to take them down. For a while at least. However, the benchmarking page has been updated with all sorts of interesting info by Max, and I ramble on about the current state of hardware, and what you should do to upgrade. It's been a while since Max or I have updated, so there's alot of info to digest. And if that's not incentive enough to read, I promise that, unlike other people, I will not use any bad puns.  -Rat

Kiss Our Ars!
Well, It will be a bit easier for Intel to do that....now that they are behind Team Lamb Chop!  Sometime around 5:30-6:00 PM EST, Team Lamb Chop "unofficially" passed Intel into the #5 Overall spot teamwise.  The team pages do not point this out yet since the backlog of members work units have not gotten posted to the team totals.  The current team totals show TLC being around 7600 work units behind Intel, but the more accurate numbers (totaling the individual team members) has TLC on top!

Who's Next?  Yup you guessed it....venerable Team Art Bell.  The SETI totals puts us about 115,000 work units behind them......but the actual numbers are something on the order of 60 - 70,000 work units behind.  When will we pass them?  Current estimates (from the SETI server data) puts us passing them sometime at the end of June.....but I have a feeling that we shall ass them around the 2nd week of June! w00p!

Servers Up!
Of course you probably already knew that!  I was just too lazy to update the front page here :-)

SETI Crunching on the Playstation 2?
I am not sure if I had covered this before...but here it goes anyways :-).  Many of you may have already read the Ars comparison of the PS2 with the current PC technology (If you haven't read it now! OK... after you go through stuff here!).  The new PS2 looks pretty impressive with its RISCIII CPU core, and many have asked how fast a SETI WU could be crunched on one...  SOme of the SETI crew has looked into the possibilities.  Hiram chimed in with this reply about the doing a port for the PS2:

<It does make one think, since the playstation 2 will be released in the US
<with a HD upgrade capability and internet connectability, why not do a port
<to the Emotion Engine for seti.  Just leave the PSX2 running all day and let
<that custom, high memory bandwidth, massively complex, number crunching
<monster of a CPU/Controller setup burn through units all day.  You'd need to
<burn a custom CD probably. :)

I had a playstation engineer try the port.  It was not practical.

--Hiram - WB6RSS

pHear For Your Life!
Just when you thought overclocking your rig was safe (ok maybe at the risk of a fried CPU or MoBo :-)....then comes along this report:

Overclocker Creates Rift in Space-Time Continuum

Santa Cruz, CA - A rift in the space-time continuum was created today when overclocker Jamie Aperman ran a 750 MHz Coppermine Pentium III at 1.6 GHz.  Overclocking has long been blamed for causing global warming, but this is the first occasion that the fabric of space-time has been damaged.

Hrmm.....makes you wonder why we are crunching work units in search of ET.....when extreme overclocking techniques, may actually let you visit ET!  BTW....the site this was found has a bunch of hilarious stories.  Check them out!

One Last Thing...
If you want some tangible proof of your SETI@Home participation, there is a link over at the SETI page that will allow you to download and print out a "Participant Certificate".  It may not be much, but hey, if you don't like it after you print it out, it could be used as a coaster or lining for the litter box. -zAmboni

Servers Down...
At the time of this writing (4:15pm EST) the SETI servers are down.  This was a planned outage according to the SETI Site (albeit with short notice):

May 22nd, 2000
We will have a 3 hour outage today for hardware and software upgrades. Unlike most server outages, this one will include our web server. The outage will begin at 1PM PDT.

Hopefully this outage will only last 3 hours not like the last time when it turned into an "extended" downtime for the servers.  :-).  Because of the planned server outage, The daily stats updates were done a bit early today.  you can check out the daily stats updates and weekly updates (posted today) on the stats pages. 

Shut Down?
Many of you may have noticed...the daily output for CCI SETI Crunshers has virtually dissapeared.   If you are not familiar with the CCI Saga check out the May 9 and May 10 news blurbs from the Stats Archives.  Over the past 8 days they have only averaged about 100 work units per day...the low number has not been due to the backlog of the SETI servers.  And most of the numbers posted for the team were actually from the other 28 members with almost none coming from CCI Europe.  To quote CCI Europe's post on the TKWSN message boards:  "we do not have uninterrupted access to all of the computers at all times".  Does this mean that they haven't had access to the computers for the past week or so?  Or has the SETI@Home crew shut them down?  I guess we can only wait and see how it plays out.  -zAmboni

Are You Willing to Face the Consequences?
Many of you may have already noticed Xanrel and his quick ascension through the ranks over the past month or so.  There is news today that his run looks to be over in the SETI crunching world.  Xanrel had been doing maintenance for the computers at his High School  He also took the opportunity to install SETI clients on the computers also.  The problem is that he never really got permission to run the SETI clients on the machines.  People at the school found out about the SETI clients he had been running, and Xanrel fessed up to being the culprit.  The result?  He was "relieved" of his maintenance duties, suspended for a month (right before finals).  He was also presented with a $3000 bill saying it took them 60 hours @ $50/hr to figure out what the client was.  (Note: I don't think they expect him to pay the $3000, but he wants to pay it to sort of clear things up).

IMHO, I think that the punishment is a bit on the heavy side.  On the other hand I really don't know the school's side of the story either.  There is a discussion in the thread if the punishment is justified or not.  With the recent email virus threats, and possibilities of installed trojans, password sniffers, et. al., school administrators do have reason to be worried about installed programs that they are not familiar with. 

The donation of spare CPU cycles for the SETI@Home project is a noble thing to do for (what I think is) a worthy cause.   Ditto for donating the processed work units to the Team Lamb Chop totals.  If you are currently crunching work units for Team Lamb Chop please go about it through the correct channels.  At work if you think the computers could be put to work running SETI@Home, ask for permission to run the clients. If they say no, it ain't no big thing! If that only leaves you with a computer or two at home to run it on then that is still GREAT!   Every little bit helps! 

Even though it may help the team out, helping out Team Lamb Chop by installing clients on computers without permission is not worth risking a job or facing the possibility of monetary damages.  If found out, the negative publicity does no good for Team Lamb Chop, Ars Technica, or the SETI@Home project in general.  So if you are thinking of running he SETI@Home client on machines that you do not have permission to run them on, you have to ask yourself...."Am I willing to face the consequences if I am found out?".   Only you can answer that question.   -zAmboni

Happy Birthday SETI@Home!
Today marks the 1 year anniversary of the start of the SETI@Home project.  To mark the birthday the SETI crew held themselves a little birthday party (birthday cake at the right).  To commemorate the 1 year anniversary, they put up a small photo album at the on the SETI@Home webpage.  You can see mug shots of the SETI Staff, and also get some pictures of the servers that have let us down so many times.

Move On Over Microsoft!
Even though the Stats update for today, nor the SETI page for the Overall Team standings show it, Team Lamb Chop has passed up Microsoft for the #6 team overall!  Congratulations Team Lamb Chop!  Most likely the update tonight in the overal team standings will put us up ahead of Microsoft.  An unidentified Team Lamb Chop member made a trip up to Redmond to "mark" the occasion and show Microsoft his appreciation.  A hidden camera caught his exploits on film:

With the SETI servers being a bit slow in logging in the work units to the teams, their numbers aren't quite accurate.  Below is a table showing the reported team totals and actual team totals for the relevant teams.  The reported team totals is what the SETI servers say the totals are (inaccurate) , and the actual team totals are the sum of all the members work units processed for the teams (as of 5/17/99 7:00pm):

* note: the actual totals for Art Bell are probably several hundred WU short since the web page only shows the top 9999 members and Art Bell has well over 12000 members ( the 2000 members not on the page have 2 work units each or less)

See more about the stats on the Team Stats page!  Right now Team Lamb Chop is just about 20,000 work units behind our next target.....ChipZilla.  They shall go down too....and words of warning to Intel.  TLC will 0wN j00!    -zAmboni

Team Lamb Chop....A Stock Market Analysis
SETI and stocks?  Do they go hand in hand?  Apparently they do!  On a whim, I wanted to see how Team Lamb Chops performance has effected the stock performance of the large corporations who crunch SETI work units.  Lo and Behold!  Team Lamb Chop has directly (or maybe indirectly) influenced the stock performance for several large corporations.  There is a strong correlation to the immediate threat that Team Lamb Chop has had with them in the SETI@Home standings and how their stock performed.  (any other market factors that didn't relate to what i wanted to say here were completely ignored and discarded).

The following is a graph showing the relative performances of 6 of the heavy SETI hitters in the overall team standings.  The companies in this graph are: SGI, Sun, Compaq, Intel, Microsoft, and Apple (ripped from CNet).  The color of the team names on the graph are the same as the line colors for that company.  The stock prices were normalized at 0% for all teams on May 16, 1999, and the graph shows the relative performance of each stock over the past year.  There are three major influences Team Lamb Chop has had on the performance of these stocks and they are pointed out by the numbers 1-3.

1. There was a huge loss in Apple stock following Team Lamb Chop passing Mac Addict in the team standings.  Of course Apple wants to "think different" but they had to think differently when some hardcore PC users with alternative OSs kicked their butt.

2. Right before the beginning of the year Team Lamb Chop started to turn the tide, and started gaining on Microsoft.  This was not good news for the monopoly, sensing the impending doom their stock starts a prolonged slide into oblivion (DOJ hearings?  never heard of 'em)

3. Team Lamb Chop to crank things up some more, and starts gaining on ALL of the teams above us (SGI, Sun, Compaq, Intel, Microsoft).  This news struck fear into stockholders and their stocks have been in a near free fall afterwards.  Apple shareholders didn't think different, they only thought one thing also "SELL SELL SELL!"

Don't believe me?  Compare the trends with the top graph on the "Catching Up" graph here.  You shall become a believer also!!!!!!!

Update!
It appears that there is another backlog of work units that have yet to get posted to the individual team WU totals (i.e. the team total is less than the total of the team member WU totals).  Taking that into account for both Team Lamb Chop and Microsoft, we are currently only 6,000 WU behind Microsoft instead of the 10,000 WUs that the team totals suggest!     -zAmboni

2 Million S@H Members
Sometime yesterday the SETI@Home project passed the 2,000,000 member mark.  They have passed this mark even before the first anniversary of the unveiling of the project.  Pretty impressive eh?  Well, OK, its not that impressive. Why?  Well of those 2 million members, 650,000+ of them have not returned one work unit :/.

Microsoft and  Intel are Running Scared!
There are scattered and unconfirmed rumors that M$ and ChipZilla are running scared with the impending stomping that Team Lamb Chop will deliver to them.  Apparently Microsoft will include in their next service pack for NT, and Windows98, and automatic install of the SETI@Home client.  There are also reports that they are now currently shipping the S@H client integrated into Win98/IE5.0.  This attempt may be short-lived with the possibility of the M$ breakup.  Further rumors say that the SETI@Home project will also comply with the DOJ ruling, splitting up the Microsoft team in half, the two teams without interaction, and WU sharing. 

Intel happens to be running scared also.  With their recent "Caminogate" fiasco, they don't want to be embarrassed even further by a hardware enthusiasts site passing them in the S@H rankings...  There is a top secret plan in which Intel will offer a "trade-in upgrade for any current BX/VIA 133/SDRAM systems.  They will replace free of charge those memory/motherbard combos with an i820/RDRAM combo.  Luckily for us, TLC members can see through their scam....Intel wants to slow other team's processing because the i820/RDRAM combo gets smoked by even a BX/SDRAM combo in SETI.  JUST SAY NO!  (of course all of the above isn't true, but only done in jest :-)

The actual countdown begins for Microsoft.  With current estimates Team Lamb Chop will pass Microsoft with a week!  Intel would be the the next to go (sometime around 5/29).  We are also set to pass Art Bell sometime around 8/1.  Unfortunately, unless something changes, CCI Europe is on track to pass Team Lamb Chop before we will pass Art Bell :-(.  -zAmboni

Update
On the newsgroup I guess Joe tried to defend himself with the following in a post (I think):

There are numerous conceivable possibilities under which the terms of
the license would not be violated:

1) I retain ownership of systems which I rent to clients.  I do, infact, do this sort of thing.  I actually have systems that I own which I have not seen in excess of five years.  Would you care to be a dearie and explain to me how it is that I am not allowed to run SETI@home on such machines?  Would you care for me to explain to you how much of a pain it is to upgrade software on such a machine, due to a "forced upgrade"?

2) I am authorized by the customer to make high-level decisions about their computing systems and how they are used.  I do, in fact, also do that sort of thing.

3) I've actually asked the customer if they'd like to run SETI@home.  I've also done that sort of thing.

These are just three trivial possibilities, none of which would cause any violation of the licensing agreement.

Dunno if it is actually a defending of what he does or not.  Sounds like a carefully worded reply, which he doesn't really say he does this or that but is more of a "hypothetical".  Who knows.

Where does this leave us?  I don't know. With Joe and CCI Europe, these appear to be extreme cases here.  I guess it all depends on your interpretation of the license agreement.  Oh what a fine line to walk.  What constitutes ownership, or permission.  When it comes down to decisions like this, many SETI@Home participants (probably even TLC members) have to interpret it their own way.    -zAmboni

Where Do They Get These People?
The SETI newsgroups yesterday had another show of arrogance from one of the participants.  In the thread "Q about version 3.0" there was an interesting exchange of posts.  It all started out with talk about the forced upgrade path that will be part of the version 3.0 rollout.  Then Joe Greco crawled out of the sewers to reveal some questionable business practice (hey guess it isnt any worse than CCI Europe's practices shown on the stats news).

While the prospect of one Windows user having to upgrade their home PC's SETI@home client doesn't seem to be all that big of an issue, there are some of us have been dumping SETI@home on every server-class system that goes out the door.  Many of these, I'll never get a chance to see again (and hence will never upgrade).

I'm talking _hundreds_ of machines.
You lose computing capacity.
And that's a damn shame.
... JG

Fun.  Could you imagine buying a workstation/server from what you think that is a reputable company, only to find out that they have installed who knows what kind of programs on your computer?  I thought that M$ with its bloated Windows preinstall was bad enough!  George Dishman questioned whether or not he has informed these people that he is installing S@H on their computers, but 'ol JG skirted the issue and complained again about future loss of work units on his account:

:Since the SaH rules insist you obtain the permission of the owner before
:using the machine, presumably you are informing the customers of
:"sol.net Network Services - Milwaukee, WI" (from your header) that it has
:been installed and including instructions for upgrading with the product
:documentation so that they can keep it up to date.

You can presume whatever you like.  Unfortunately, on systems which are essentially embedded application servers, the customer does not have the _option_ to upgrade the client, because it's part of a base OS package that cannot be field-upgraded or field-modified.
... JG

Hiram then decides to step in:

Are you shipping the SETI client in your application server product ?  Do your customers know that you have loaded their machine with this application ?  Are you an authorized source of the seti client binary ?

Something doesn't sound right with this picture ?
-Hiram

In case you dont know...Hiram is the person who organizes the porting of the S@H clients over to the unix platforms.  He also has been one of the more outspoken people warning and reminding people that according to the License Agreement for the S@H clients that you should only install the client with permission of the owner of the computer.  Obviously Joe didn't know who he is:

Yes, your comprehension of the actual legal meaning of the SETI@home License Agreement, combined with lame attempts to reverse-engineer the meaning of statements that I have made, in an attempt to imply that I said things which I did not.

Unless you happen to be a lawyer representing one of the customers whom I have allegedly jilted by running SETI@home on a server that they allegedly own, allegedly without their permission  (God, that's a lot of allegations)...  I'd suggest you stuff it.

Have a good one,
... JG

BZZZZZZZZZZZZZ!  Wrong Answer!

I was simply amazed at what Joe claimed he had done.  (placing the client on production appliance servers)  I have to assume that Joe is simply bragging a bit and wouldn't be so foolish as to place the client on a shipping product.

It wouldn't be up to me to enforce the license agreement.  If he really did place the client on a shipping product as his claim seemed to indicate, I'm sure the feedback from customers to his company would soon bump the unemployment figures next week by one.
--Hiram

One more unemployed, and possibly one less SETI@Home participant.  I sort of chimed in on the thread with some of what CCI Europe has been doing.  I am sure that Hiram will notify the people who are in position to enforce the license agreement about both of these people.  We will have to wait and see what comes of it.  -zAmboni

Interview!
SETI@Home guru Eric Korpela has agreed to do a email Q&A "interview" with me for the page.  If any of you have any dire or pressing questions about the project or even hardware used for the SETI@Home project, drop me a line.  I am in the process of preparing some questions for him to answer, and the more questions to choose from the better! :)  -zAmboni

A Peek Behind
For the past couple of days when doing the team stats, I have been looking a bit behind us to check out the scene.  Well unfortunately, there may be a couple of threats on the horizon.  I have started keeping track of the stats for CCI SETI Crunshers, and The Knights Who Say Ni!.  The numbers from those two teams for the past couple of days are pretty impressive.  CCI have been posting alot of numbers but they are only a 30 member strong team.  The majority of their work units 575388 (98.7% of the team total) is coming from a "member" CCI Europe.  CCI Europe is most likely a team or fleet of computers posting to one account, and they are the #2 overall member of SETI@Home (behind SETI@Sun).   TKWSN are also posting pretty high numbers also, but their production seems to be a tad below Team Lamb Chop's.  I will be keeping an eye on things....and you can keep an eye on their production with the team graphs.    -zAmboni

RAMBUS.  Bust? or Misunderstood?
After seeing the unimpressive benchmark numbers from a RAMBUS based i820 system over the weekend, I started reading up a little on RAMBUS and come of the controversy with therein.  I found this pretty good article at Hardware Central called "SDRAM vs. RDRAM, Facts and Fantasy".  I will give several bits of info from the article that may be of particular interest in SETI crunching:

• With all of the talk about RDRAM's "low latency" we may be talking apples and oranges here.  at low memory bandwidth stress the latency of RDRAM is around the same as SDRAM latency.  But the way RDRAM is made, It is really made for high memory bandwidth applications, and RDRAM should outperform SDRAM when the there are high memory bandwidth requirements.  (this doesn't really explain the low SETI scores though)
• Power States and you:  With RDRAM there are different power states that "balance power consumption and access latency".  These different power states are important for consideration in ALL computer products especially when power consumption is an important factor in portable computer's battery life.  There are apparently 4 different power settings for RDRAM which are accessible through system BIOS.  Two of the settings consume more power than the other two (but provide lower access latency).  Apparently many of the default configurations of RDRAM systems place some of the RDRAM pools in the lower power/higher latency settings, which may hinder performance unnecessarily.  BIOS tweaking for RDRAM systems should help increase the performance for these systems, and this may explain some of the poor performance of the RDRAM SETI benchmark.
• As what usually happens with new technology, programs need to evolve to showcase the power of the new technology (i.e. there has been quite a lag for hardware T&L in current games even though the GeForce supports it).  Many programs out today are written for the current tech, but also compatible for legacy technology.  The SETI client may have been written to take advantage for the CPU and memory architecture that is supported by today's processors and memory architecture.  Therefore it may not be surprising that the SETI client doesn't perform as well on RAMBUS as it does on SDRAM systems.
• One last note.  It is about the price of RDRAM modules.  The tech is pretty new and most fabrication plants have not ramped up appropriately for the demand of the modules.  Right now the demand far outweighs the supply.  Add into that the manufacturers want to recoup some of the ramp up prices, this gives RDRAM a high price.  When the ramp up of RDRAM gets going in several months the supply will start to meet the demand, and prices will fall.    -zAmboni

Eric Strikes Back
When there is an extended server outage over at the SETI@Home offices, there is usually quite a bit of outcry from members on the newsgroups.  Eric Korpela gave a couple of explanations on what exactly happened to cause the server downtime...and hinted that there may be a planned down time in the future to help prevent the same problems from happening:

RossM <[email protected]> wrote:
>Get your f****** act together guys I am getting pissed off that every time I
>want to connect to the server to get a WU it's down for one reason or
>another.
>
>Amateurs.
>
>RossM

I can at least offer a reason for the down time, but I'm surely not going to offer an apology.  The index in the database that decided which work units are to be sent became corrupted.  The corrupted portion was esentially selecting on two values (i.e. "where a=0 and b=1").  Selecting on work units where a=0 was returning a lot of work units many of which had b=1.  Selecting on b=1 was returning a lot of work units many of which had a=0.  But selecting on both failed to return any work units.  We tried fixing the problem basically trying the quickest fix first.  The final fix took the amount of time it took and nothing was going to change that.  You don't like it?  Tough.  Maybe it's time to grow up.
Eric

There also was this post:

Roger Halstead <[email protected]> wrote:
>On a serious note though...An index which is such a simple thing, can be
>relativley easy to fix and reconcile the records...Unless you have a large
>database.  When you have a very large database reconciling the index with
>the proper records can be a realy time consuming chore.
>
>Many times, reindexing is not a viable option as there may be thousands,
>hundreds of thousands, or more records that relate back to the original
>index.  That means reconciling the index which can be some pretty heavy
>work, or at the least time consuming.

Yep.  The indexed table had (as of friday) 43,979,239 records.  I'm hatching a scheme to build a second, smaller "active" table that will act as a cache to the larger table.  That way if the indices of the large table get corrupted we can still operate from the cache table while the indices are rebuilding, and if the indices of the "active" table get corrupted it will only take an hour or two to regenerate them.

Of course making this change will require some down time, but at least it will be scheduled down time.  And such major changes to the server architecture will not be fun. 

It's unfortunate that none of our volunteer Informix consultants foresaw this possibility.  But at least I'm glad we can continue to operate without the indices on the spike table (530 million entries at last count) should they fail.  Regenerating them would take much much much longer.
Eric

-zAmboni

Data Server Back Online!
The way it looks from the server status page, the servers are back up.  Unfortunately it looks like the servers are totally swamped from everyone trying to upload all their results, I cannot connect.  Soooooooo....It still looks like you still have time to fit in a benchmark for the benchmarking page ;-)  -zAmboni

Server Outage!
Looks like the Seti servers are down right now, and are going to be down for a while. Therefore, you might want to use your idle time running through the benchmarking unit- it can be handy in determining if your system is running as fast as it should. Not to mention, we don't have a lot of times from some systems, and can always use some more. Anyways, it's not like your CPU is doing anything better! -Rat

Server Outage a Major Problem?
Many of you probably have realized that the SETI data servers have been down for most of the day.  The problem with the server may be pretty serious (on the other hand I don't know anything about the servers and if it may be a simple problem :).  The length of time the server has been down can be an indication, and this post on the Tech News page doesn't sound promising:

SERVER OUTAGE: 12:30pm PDT - The data server is currently down due to unexpected problems with how the server communicates with our database. We have been debugging this since early this morning, and there is currently no definite end to this outage in sight. We'll update this message when a solution is nigh.

Hopefully the crew can get this squashed and back online soon.  An unexpected benefit of the downtime has been that the backlog of data is being pared down during this time.  The posted team totals are getting back to the actual team totals...but the backlog may grow back when the servers are placed back online and are sure to be overloaded.  You can either try to connect with the client or check the server status page to see if the servers are going or not.  -zAmboni

More Version 3.0 News
With the upcoming version 3.0 of the SETI client there are questions to be asked on just how they will handle the switchover, and how things will change with the SETI@Home website.  Here are a few of them that have been posted (and answered) in the past couple of days:

>How are the science results of the pulse detection going to be presented
>on the website? Will it be something more than the 'highscore' table for
>gaussians? Any chance those get revamped as well?

We've got lots of web site changes planned.  The first will be a major update to the gaussian high score table.  We'll be going through the gaussians top down examining them with redundancy checks and the like.  We'll eventually have two tables.  The "high-score" table with links to more info about the gaussian, and the "former top 20 gaussians" table, which will list the reason that we no longer consider that gaussian a candidate.  My guess is that the "former top 20 gaussians" table will have tens of thousands of entries, especially once we get the RFI disqualifiers working.

Then we'll be doing the same thing for the pulse detections

I have a feeling that they want to change the SETI page to show more of the progress in analyzing the results that they have received from members.  This next question was posed by me asking about the performance and rollout of the version 3.0 clients:

>1)  Are there any rough estimates on how much longer the version 3.0 client
>will take compared to 2.x (if it will take longer :-)

We're shooting for 15 to 25%, but it might end up as little at 10%.

>2)  How will the rollout of version 3.0 of the client be handled?  Will the
>grace period be about as long as the one for version 2.x?  I am asking
>because if many users wait a month (or so) until the deadline date for
>installing version 3.0, there would be alot of time spent analyzing with the
>version 2.0 clients instead of version 3.0 with the added science (possibly
>alot of lost information).

It will likely be a shorter grace period than 2.0, but it's exact duration isn't known yet.

Ug.......25%?  Lets hope that it is going to be closer to 10% :-).  I think I will do some calculations on how it will effect Team Lamb Chop an the other top teams on their daily production.  Stay Tuned.  The shorter grace period isn't really suprising.  Alot of science would be lost (i.e. pulse detection) if the grace period would be longer.  Most of the crunchers would probably wait until they HAD to upgrade their version just for the speed advantage of version 2.x.  The above two questions were answered by Eric Korpela....but Matt Lebofsky and Hiram added that the there was never a forced upgrade to the version 1.x clients for the Win/Mac GUI versions, but there was a forced upgrade to the CLI and unix clients. 

Another take on the "lost processing" was addressed easily in the next post.  There was a question if the SETI crew would go back and reanalyze the previous data crunched with the version 3.0 client.  That may not be necessary.  Because the sky will be scanned several times during the project the previous tapes looks as if they will be used to confirm something they find "interesting":

>>The assumption now is that we won't be going through the old tapes again.
>>We may eventually change our minds, but it's more likely that we'll just
>>continue on with new data...
>>
>
>If an interesting signal is detected by the new pulse detection code, will you
>be retrieving data from archived tapes to confirm the signal?

Yes.  It would be a fast way to do a confirmation without requiring use of a telescope.

The last question has to deal with the use of SETI add-ons and the new version.  This is also interesting to the current users of add-ons and if they would have to reconfigure their setups for the new version.  It appears that it may not be necessary:

> Eric or Matt,
>
> For the SETI Add-on developers - will the files used be the same?  In v2,
> the important files appear to be:
>
> wtemp.sah - WU being transmitted or broken transmission
> work_unit.sah - WU ready to process
> results.sah  - WU ready to transmit
> state.sah   - contains CPU and PROG lines to determine WU processing status
> stop_after_send.txt - tells SETI to transmit a completed WU and not download
> the next WU
>
> I may have forgotten a file here, but you get the idea.
>
> Also, will the command line client switches
> of -stop_after_(*), -proxy, -socks_(*) remain the same?

The changes between 2.x and 3.0 are mostly science, which means all the file names and command line options (or the ones you ask about) will be identical.

-zAmboni

New SETI Newsletter
I was checking the SETI@Home main page a bit ago and noticed that they have changed some things.  The main one was that they got rid of the News Summary link (they hadn't updated that page since Dec 29, 1999), and have integrated it into the Science Newsletter page.  Looks like they put up Science Newsletter #2 yesterday....you can go and check it out.  For a brief synopsis, there are some portions of the sky which have been scanned multiple times, and they have been looking for signals that have appeared during scans of the same area:

So far, we've identified two pairs of strong gaussian shaped signals, both seen on two different observations. We've also found three trio's of weaker gaussian signals, each signal seen on three different observations. We are examining these data further, and also examining some even weaker signals. But these multiple detections are almost certainly from noise, and not signals from another civilization. We are searching through a data base of 50 million gaussian signals, and a few of these 50 million signals will randomly occur at almost the same place and the almost the same frequency. We've examined thousands of these kinds of multiple detections in our 25 years of searching, and so far they've always turned out to be noise or radio pollution from terrestrials, not extraterrestrials.

There is also some information about the next version, and the pulse signal detection that is going to be included:

The new pulse searching code adds several hours of computation to each work unit. However, we've also been optimizing version 2.04 analysis software, and when we combine the extra computational burden from the pulse search and the speed up from the optimization, the new version 3.0 doesn't take that much longer than version 2.04, and will allow a search for a whole new class of signals.

Hopefully version 3.0 will not be *that* much slower than the current versions out now...but I guess it is worth the added information that will be provided.  -zAmboni

New SETI Project
Ok, this project isn't going to compete for SETI@Home members, but it is interesting none the less.  There is an article on space.com reporting an an "Optical SETI" project being developed.  This project is set to scan the sky looking for lasers pointing at the earth and detect the laser signals.  It is an interesting project since laser signals can contain a heck of a lot more information than radio signals can....BUT!  You may scratch your head about the project since the lasers basically need to be directed specifically at the earth, and is assuming that ET would be trying to contact US specifically.  I really doubt if you will get a stray laser signal hitting the earth...but ya never know.  This is unlike the radio telescope searches since radio signals aren't a directed signal, and if there is an ET radio signal, it is more likely to reach the earth than a laser signal.  -zAmboni

Humor via The Register
A couple of weeks ago The Register announced a "contest" to "find which combination of processor and operating system can munch its way through a typical SETI workload in the shortest time." Well it looks like they have found their winners, and some of their statements in the follow-up article may give you some laughs.  Here is an example:

Vying for the best overclocked box award was LKY's Intel Pentium III 550E @ 770MHz (140 fsb), using an Asus P2B-B M/B, 128MB Micron PC-133 -75B RAM, running Windows 98 SE, which averaged 6 Hours 15 Mins per work unit.

Hrmmmmm....I am sure we can come up with many better candidates from Rat Bastard's results table.  That "kick ass" system cant even come close to even my system:  PIII 600E @ 744MHz (124 FSB), Abit BX6 (rev2), 192 MB vanilla PC-100 RAM, Win 98.  Using the Version 2.0 CLI, I am averaging about 5:40 for a "normal" work unit, and with the benchmark work unit I turned a 5:06.  -zAmboni

-Front Page