It was because I find increasingly difficult to justify running 3-4 computers 24hrs/day when a) there is no feedback to what my contribution has delivered and b) ...

A partial answer is the new feature Matt just released showing some of the results ...

Still more work to be done, but, some information is coming out ...

. . . yep - see here in Science: First Signs of the NTPCkr

---

BOINC Wiki . . .

Science Status Page . . .

............possibly because of too much tinkering with the front end?

---

Sorry The Gas Giant for getting off topic.

Ty Dr C.E.T.I for your post about the changes to the science status page. Funny... I looked yesterday and there was no chart. I really like seeing data represented in a graphic form. BRAVO MATT!! I can't wait to see what the millions of results start producing!!
BTW Matt, if you read this, is there a way to get the file splitter stats for my flash project? Is there a different XML file I'm unaware about? I really want to include them in my Flash project. Even cooler, an XML that has the data for the NTPckr? I know you have a 'full plate' but it would be nice to add either or both to my project so it can continue to grow.

About the auto updates of clients.. yes.. it has been my experience that if you leave the door open, the spammers will march right in BUT by using the LONG user ID, which only the seti database and client knows, an update could only be triggered when the two IDs match solving the issue of 'outsiders' some how tricking your system. As far as the platforms, It would be a fair amount of work but all the different platform clients could me collated in to a single database. By adding a data item describing what platform they use to the database that contains the subscribers information, tasks, computers and such, that issue would be solved. I'm pretty sure there is a data item for what client version the user is running anyway.

As far as my Flash project.. Thanks all for the 'Atta Boy's
I have no idea why Adobe has not supported 64 bit yet seeing how they claim 98% of all computers are running Flash. I'm sure it's in the works. I chose Flash because it's a nice environment to work in and some what easy to learn. I think doing an ulternate page for non-flash users is a great idea and when I get this one done, I will need to 'bone up' on my PHP abilities.... lol I know C quite well and visual basic is a breez but PHP is a very new PL for me.

About the 50k cost to SETI... I use to own a machine shop that at one point employed just over 50 employees. Stating that it cost twice as much as the employee's earnings is not quite correct. The insurance was 240.00/month which I paid out of pocket.. social security, medicare, the rail road tax (<- a WA state tax), dep of labor and industries (<-65 cents for every man hour worked) was held out of the employees check along with federal taxes. I paid for ALL required safety equipment including steel toe boots. I found the overhead net cost to be about 10 to 15% depending on that employees work habits/skill. The cost of having the payroll done by ADP was 125.00 a week for processing. $125/50 = 2.50 per employee.
The average pay was $18.00/hr.. or around 2880.00/month gross. FYI.. 18.00x2080(hrs in a yr)=$37440.00

2880.00(insurance/yr)+325.00(boots twice a year)+130.00(payroll processing/yr)+1440.00(7 days/holiday+3 sick days/yr)=$4775.00
$4775.00/$37440.00=7.84% $37440.00+$4775.00=$42,215.00
The other 2 to 7% additional overhead came from employee mistakes. At a 7% cost/pay scrap rate, that employee would no longer be employed by my company...

oops... way off topic and toooooo much rambling.. but just trying to enlighten you.

---

Not Ready Reading BRAIN. Abort/Retry/Fail?

Well 5% or less of the people that joined on the same date as me are stil crunching for Seti.

Instead of just asking for more people to join. How about trying to get the project at a place where those who join actually stay. Remember if someone has a bad experiene they'll tell 10 people whilst if it's good they'll tell 2 maybe 3 at most.

I only stay because I like the atmosphere on the boards and I believe that the project has a purpose.

But for most casual crunchers coming to the project new, it looks like the guys are constantly having problems, there is a extreme lack of financial support from the various sources that claim to support it. The Seti system is basically used as a testing pen for Boinc. And there is still no information regarding what is happening to the data that we are all crunching.

Thankfully, due the introduction of NTPKR we can see something relating to what we are all doing.

I would suggest concentrating on keeping those of us that are already crunching. I have said before and i'll say it again. If there was a huge increase in cruncher numbers the system would not cope in it's current state.

---

Well 5% or less of the people that joined on the same date as me are stil crunching for Seti.

Instead of just asking for more people to join. How about trying to get the project at a place where those who join actually stay. Remember if someone has a bad experiene they'll tell 10 people whilst if it's good they'll tell 2 maybe 3 at most.

I only stay because I like the atmosphere on the boards and I believe that the project has a purpose.

But for most casual crunchers coming to the project new, it looks like the guys are constantly having problems, there is a extreme lack of financial support from the various sources that claim to support it. The Seti system is basically used as a testing pen for Boinc. And there is still no information regarding what is happening to the data that we are all crunching.

Thankfully, due the introduction of NTPKR we can see something relating to what we are all doing.

I would suggest concentrating on keeping those of us that are already crunching. I have said before and i'll say it again. If there was a huge increase in cruncher numbers the system would not cope in it's current state.

Perhaps you missed Matt's view on the world: "...our current data flow bottleneck is the amount of data we are able process, so the more computers the better. Tell your friends!

To me, that's the horse's mouth (Sorry Matt).

F.

---

Fred that is what i meant, maybe i should have worded it better.

When the system does go backside over breast, there is never enough work to keep the project going.

---

Fred that is what i meant, maybe i should have worded it better.

When the system does go backside over breast, there is never enough work to keep the project going.

Satan,

First of all I agree with your premise - it is a marketing truism that it costs 10x as much to recruit a new customer as to retain an existing one (actually the factor might be greater than 10 but you get the drift).

I, too, enjoy the community of these boards and that is what has kept SETI as my primary project over the past 10 months since I discovered BOINC.

But I read your post as saying "Please don't recruit too many new crunchers as the system won't be able to cope" which is diametrically opposite the Matt's plea for as many new crunchers as can be recruited in order to process the data that are available. Perhaps I am being more than usually dim this morning?

F.

---

One thing that worries me is that I'm not sure Matt has really got his head round the issue of 'shorty' (VHAR) workunits.

At the moment, we're seeing very few of these, and the download data graph is relatively flat at about 40 Mbit/sec. We know that the old router used to max out at about 60 Mbit/sec: we don't know the maximum for the new router, because the next bottleneck in the system is the nominal 100 Mbit/sec copper connection 'down the hill': it tends to max out at a practical bandwidth of around 95 Mbit/sec.

So we have a working bandwidth limitation of less than 2.5 times the 'steady-state' consumption of full-length workunits.

'Shorties' require roughly six times the bandwidth of full-length WUs. (They crunch in one-sixth of the time, so you download six of them to keep warm for as long as one long one). So if we fall over a batch of tapes which are all VHAR (as happened between Christmas and New Year), we would already be at least 2x overcommitted on the download pipe.

I've tried to make this point to Matt in the technical area, but he's never actually answered it - he keeps talking about 'noisy' WUs and the radar blanking problem, which manifests itself as -9 overflows. Worse, we seem to be processing blocks of adjacent recordings together. As I type, six of the splitters are working on recordings from 25 or 26 March. The main Arecibo oberving project on those dates was A2010, so things will be quiet for a while longer (mid-range ARs). But watch out for tapes from the second half of April: 11 of the last 15 days in April had observing sessions for A2060, which by my reckoning generates 'shorty' ARs. Things may not look so rosy for the return-invitees unless Matt consciously loads a mixture of recordings, so that there are long WUs issued among the shorties.

Richard,

If they can get Astropulse on-line then that will slow the processing time quite a bit especially if the project does not allow an "opt-out" option for work.

As for the whole point of the thread, well, this is back to the point I beat to death in the Politics forum with the many reasons people leave, not just SaH, but all BOINC projects.

The lack of understandable results is just one of the many reasons ...

---

One thing that worries me is that I'm not sure Matt has really got his head round the issue of 'shorty' (VHAR) workunits.

At the moment, we're seeing very few of these, and the download data graph is relatively flat at about 40 Mbit/sec. We know that the old router used to max out at about 60 Mbit/sec: we don't know the maximum for the new router, because the next bottleneck in the system is the nominal 100 Mbit/sec copper connection 'down the hill': it tends to max out at a practical bandwidth of around 95 Mbit/sec.

So we have a working bandwidth limitation of less than 2.5 times the 'steady-state' consumption of full-length workunits.

'Shorties' require roughly six times the bandwidth of full-length WUs. (They crunch in one-sixth of the time, so you download six of them to keep warm for as long as one long one). So if we fall over a batch of tapes which are all VHAR (as happened between Christmas and New Year), we would already be at least 2x overcommitted on the download pipe.

I've tried to make this point to Matt in the technical area, but he's never actually answered it - he keeps talking about 'noisy' WUs and the radar blanking problem, which manifests itself as -9 overflows. Worse, we seem to be processing blocks of adjacent recordings together. As I type, six of the splitters are working on recordings from 25 or 26 March. The main Arecibo oberving project on those dates was A2010, so things will be quiet for a while longer (mid-range ARs). But watch out for tapes from the second half of April: 11 of the last 15 days in April had observing sessions for A2060, which by my reckoning generates 'shorty' ARs. Things may not look so rosy for the return-invitees unless Matt consciously loads a mixture of recordings, so that there are long WUs issued among the shorties.

So, at least at this time, we know that we either haven't gotten another long run of shorties, or there may be some plan in-place that successfully shuffles tapes and prevents a run of shorties.

It seems to me that the project can live in fear of angering crunchers when things go poorly, or they can do the best they can and live up to the promises they've made (the promise that sometimes things will go well).

I'll also submit that the average cruncher probably hasn't looked at BOINCMGR once in the past month, and wouldn't notice if the project was down, or struggling.

---

As for the whole point of the thread, well, this is back to the point I beat to death in the Politics forum with the many reasons people leave, not just SaH, but all BOINC projects.

It seems IOTTMCO that the biggest issue is apathy.

How many people download BOINC, join a project, and then just forget about it?

Their E-Mail address changes, BOINC is running in the background, E-Mail from the project goes unreceived and unread, and if BOINC quits, they wouldn't even notice.

If all of the active crunchers were deeply interested in the project, we'd have 800,000 active posters in the forum. I suspect the actual number is less than 8,000.

---

One thing that worries me is that I'm not sure Matt has really got his head round the issue of 'shorty' (VHAR) workunits.

At the moment, we're seeing very few of these, and the download data graph is relatively flat at about 40 Mbit/sec. We know that the old router used to max out at about 60 Mbit/sec: we don't know the maximum for the new router, because the next bottleneck in the system is the nominal 100 Mbit/sec copper connection 'down the hill': it tends to max out at a practical bandwidth of around 95 Mbit/sec.

So we have a working bandwidth limitation of less than 2.5 times the 'steady-state' consumption of full-length workunits.

'Shorties' require roughly six times the bandwidth of full-length WUs. (They crunch in one-sixth of the time, so you download six of them to keep warm for as long as one long one). So if we fall over a batch of tapes which are all VHAR (as happened between Christmas and New Year), we would already be at least 2x overcommitted on the download pipe.

I agree with your concern, but...

The old estimates for 'Shorties' were that they would take one-sixth the time of midrange work, but it was never true. The current estimate is about 1/3.27 and seems to be reasonable on average. Still, 3.27 times 40 Mbit/sec means that the Scheduler would be issuing tasks needing download bandwidth of 130 Mbit/sec if we got into an 'all shorties' situation.

I've tried to make this point to Matt in the technical area, but he's never actually answered it - he keeps talking about 'noisy' WUs and the radar blanking problem, which manifests itself as -9 overflows. Worse, we seem to be processing blocks of adjacent recordings together. As I type, six of the splitters are working on recordings from 25 or 26 March. The main Arecibo oberving project on those dates was A2010, so things will be quiet for a while longer (mid-range ARs). But watch out for tapes from the second half of April: 11 of the last 15 days in April had observing sessions for A2060, which by my reckoning generates 'shorty' ARs. Things may not look so rosy for the return-invitees unless Matt consciously loads a mixture of recordings, so that there are long WUs issued among the shorties.

On all of the April days when A2060 is active, there's also A2010 and/or A2048 to provide mid-range work. The A2060 observations occupy 2.5 hours so should end up in two of our 50 GB chunks and be handled by two of the splitter processes while the other seven splitter processes do other data. Statistics and Murphy say that there will be times when there's an overload, but it shouldn't last too long.

In May when A2186 (Galfacts) starts, there will still be leavening from A2010 and/or A2048, so for what's on the Arecibo schedule so far I think there won't be major issues. Unfortunately the ALFALFA (A2010) Spring 2008 observing schedule runs out the end of May...

Even short-term issues do tend to discourage participants, though. It would be good if the project could do some forward thinking, it would be nice if they had enough funding to relieve their need to scramble every day just keeping the project running.

                                                                 Joe

Even short-term issues do tend to discourage participants, though. It would be good if the project could do some forward thinking, it would be nice if they had enough funding to relieve their need to scramble every day just keeping the project running.

                                                                 Joe

... which has been my theme for a while.

I've been here watching BOINC on my machine when "things are bad" and while I think there is a missed opportunity for the project to tell BOINC clients to "cool it a bit" when the load is heavy, BOINC (client and server) does survive, and recover, without intervention.

It seems to me that participants (those who watch, that is) have an unreasonable expectation of perfect, when "good enough" really is good enough.

Who cares if one computer has to try more than once to talk to another. If BOINC has work, and eventually delivers work, nothing is lost.

---

Who cares if one computer has to try more than once to talk to another. If BOINC has work, and eventually delivers work, nothing is lost.

Yep exactly. If the servers are overloaded and wont send me work besause of a bandwidth issue due to hardware issues or shorties or whatevver, then I dont really care. I have a 4 day cache and so far thats lasted long enough to get me through the most severe slowdowns so far this year.

If I see a bunch of shorties happening again I might think about increasing the cache after the worst is through and then only connect every 24hrs or so to upload & get more work if I can.

So far the servers have done extremely well under load, and they should handle the new app in their stride as well - I cant see it making a huge difference as not that many will be using it initially.

---

I only have a three day cache and I have had no problems keeping the farm working although, It did come close once.


Not sure who said it but they hit the nail square on the head as to why people join up and shortly there after, uninstall boinc or ignore it. LACK OF UNDERSTANDABLE SCIENCE!! Look at the science stats page for example. Matt has added a cool graphic to it.. call me stupid if you want but.. ermmm...

ok.. we have 650,000,000 plus pulses and this and that..
I know what it means.. but what about newbies? The Seti staff is too tied up with other issues to deal with the maintenance of the pages here.
I tell you, If I could do a graphics for the science stats, Id add some bar graphs showing what was discovered today, over the week and over the last year.
More over, Id have a map showing what areas in the sky are being crunched.
I have said this before.. the overall feel of this site needs a face lift.

I wonder if there is a way to form a "Seti Site Design Team", recrute people with very good knowledge in php/mySQL/html/css/java or individuals in each respective language (that will donatate their time) and get to work on building a mirror site on a test server. I think it would solve issues if a second group called the "Site Approval Team" was formed to decide on issues like the general 'motif' or 'theme' of the new pages and assign the 'design team' the tasks to get the the pages done. Something like.. the design team decides on a overhead, rollover, drop down menu system and tells the 'design team' to get to work on it' Ok so they get to work but have an argument over say... John thinks his menu is better than Michelle's. It would be the approval teams responsibility to study and select the best 'menu' based on its 'feel' I.E ease of use and general looks. END OF STORY.. Dont like it.. you dont have to be part of the design project if you can't deal with a rejected idea!! That makes me think also that a basic working set of rules needs to be created for both the 'Approval Team' and the 'Design Team'
I think the staff.. I.E. Matt, Eric and the rest, should NOT be a member of either team BUT will have the ABSOLUTE say on weather or not a page is exceptable. The prime directive being.. 'NO objectionable content' "EXTREMLY KID FRIENDLY!" The Seti Staff will need to participate in helping sort things out, like how to get to some of the data or may be asked to export the data via XML for use on the new pages. Basicly, they would be acting as 'contributers'. There are also SECURITY ISSUES that will have to be delt with. For example, the data containing user info.. passwords... the forums data sets as well as the science data. I recommend setting up a second set of data bases with the same structures as the originals but with just enough fictitious data to work with.. Also.. the mirror forums could be used wisely by the design team, approval team and the Seti Staff to discus design issues while testing prior to publication of the new site.
When the design team, approval team and the Seti staff says 'It's ready to publish' the new Site will just appear all at once...
....A hush fell apond the crowd as they gasp in amazement over the outstanding ease of use, functionality and the brilliant use of colors of the new Seti@home!!

If the burden of maintaining the site was handed off to some responsable programmers and others, it would be one less burden on the Seti staff.
Thank you for reading...
Mike

---

Not Ready Reading BRAIN. Abort/Retry/Fail?

Well 5% or less of the people that joined on the same date as me are stil crunching for Seti.

Instead of just asking for more people to join. How about trying to get the project at a place where those who join actually stay. Remember if someone has a bad experiene they'll tell 10 people whilst if it's good they'll tell 2 maybe 3 at most.

I only stay because I like the atmosphere on the boards and I believe that the project has a purpose.

But for most casual crunchers coming to the project new, it looks like the guys are constantly having problems, there is a extreme lack of financial support from the various sources that claim to support it. The Seti system is basically used as a testing pen for Boinc. And there is still no information regarding what is happening to the data that we are all crunching.

Thankfully, due the introduction of NTPKR we can see something relating to what we are all doing.

I would suggest concentrating on keeping those of us that are already crunching. I have said before and i'll say it again. If there was a huge increase in cruncher numbers the system would not cope in it's current state.

If you look at the Top 100 multi-project BOINC participants on BOINC URL, you will see that many do not crunch for SETI........maybe someone should ask them why.

BOINC V5.2.5 and earlier don't give you correct credit..
Since when BOINC V5.2.6 is available.. since days? ;-)

So I think it's time to ban BOINC V5.2.5 and earlier! :-)

There are platforms (not Windows or Linux) that do not have 5.x clients available.

Which do you excatly mean ? - There are BOINC clients 5.x for the most recent Unixes available. The only one I know which is not 5.x, is HPUX 10.20...
And HPUX 10.20 is out of support since several years. But users still use it.

---

I will probably leave SETI eventually because on a recent TV show they said and I'm paraphrasing here that Seti@Home probably won't ever be successful because once radio signals get about a lightyear out they dissipate into the ether.

I'm too lazy too change any of my clients right now and search for a project that would better benefit humanity so I will just hang out here for a while.

---

If you look at the Top 100 multi-project BOINC participants on BOINC URL, you will see that many do not crunch for SETI........maybe someone should ask them why.

OK ... according to that list i'm in the top 100 and seem to be qualified to reply ...BOINC TOP LIST

88. Crunch3r is contributing 260 GFLOPS.
Country: Germany; Team: SETI.USA


1. credits suck
2. continuing server issues
3. see jeffusa's post

---

Join BOINC United now!

I will probably leave SETI eventually because on a recent TV show they said and I'm paraphrasing here that Seti@Home probably won't ever be successful because once radio signals get about a lightyear out they dissipate into the ether.

I'm too lazy too change any of my clients right now and search for a project that would better benefit humanity so I will just hang out here for a while.

They were talking about signals like TV broadcasts. Not systems that use directional antenna.
If a UHF transmitter is connected to a 60 foot diameter parabolic reflector antenna the signal is focused into a beam less that 1 degree wide and it would effectively increase the transmitters output by about 40,000 times, but only within the beam.
Just the same as a torch or car headlight.