Sorry to my wingmen.
I lost a lot of Cuda tasks when windows updated.
Normally have auto update off don't know what happened.
Once again sorry.

EDIT: Checked some of the WU's and they were VLAR's 200 were auto killed.
They should be sent as 6.03

---

When all else fails...........and the kitties cannot get what they want for their kibble bowl............you will have lost one cruncher............

Mark,

I know the kitties expect the bowl to be kept completely full at all time, but as long as the bowl is not empty, they'll be able to eat.

It's only a problem when the kitties start to starve.

Also, please realize that a big part of this discussion is how to best keep food in the bowl -- and in every other bowl -- as much as possible.

-- Ned

Ned......don't try to coddle me..........
I only look after my kitties' best interests.........not yours or anybody else's/

Got it???

Sorry Ned.......that was rather rude.
But to the point.......I did not mean to be so brash.

The project should feed all the kitties, not just the ones who yeowl at the bowl. They can't cater to some "fat cats" and let the rest starve.

Got it?

The kitties that yeowl at the bowl speak for those who can or will not.......

I am the voice of kitties far and wide.....and speak for those who do not stand up for themselves. I am not always right, for sure. But at least I stand up and speak my peace.

Got it?

At the beginning of this thread you said:

When all else fails...........and the kitties cannot get what they want for their kibble bowl............you will have lost one cruncher............

In other words, if you don't get yours, you will quit.

In this post, you said:

I am the voice of kitties far and wide.....and speak for those who do not stand up for themselves. I am not always right, for sure. But at least I stand up and speak my peace.

Either you are demanding that your cache stay full and to heck with everyone else, or you are in favor of equitable distribution of work. Can't be both.

When we started a technical discussion of how to best keep everyone working, that was when you threatened to quit. You said you only care about your kitties, no one elses.

Maybe I missed something, but it sounds like you don't care how many kitties go hungry as long as your "fat cats" have bowls completely full at all time.

Then you say "I speak for those who can't speak for themselves" -- which seems to say that you're for equitable distribution -- kibble for everyone.

Please pick one position.

In the meantime, the technical discussion is about the best way to allocate work: how to efficiently and quickly keep at least some work in every queue without anyone going without.

OK......I'll back down....
I am a bit adamant that the kitties get what they want, but I guess this should be about what is best for the project.
I am very proud of what I have accomplished here, and would like it to remain so..
But I guess it's not all about me, now is it?

I have to stand down and realize that if the kitties are having trouble getting their kibble bowls full, it surely is affecting every other participant in the project as well.

Sorry I got so verbose.....
The kitties will share.

---

"Time is simply the mechanism that keeps everything from happening all at once."

In the meantime, the technical discussion is about the best way to allocate work: how to efficiently and quickly keep at least some work in every queue without anyone going without.

OK......I'll back down....
I am a bit adamant that the kitties get what they want, but I guess this should be about what is best for the project.
I am very proud of what I have accomplished here, and would like it to remain so..
But I guess it's not all about me, now is it?

I have to stand down and realize that if the kitties are having trouble getting their kibble bowls full, it surely is affecting every other participant in the project as well.

Sorry I got so verbose.....
The kitties will share.

My thinking is that, whenever possible, no kitty should ever have an empty bowl.

Now, they may want the bowl full, and that is probably possible most of the time.

If we work from a "no queue is ever left empty" instead of "top up a queue at every opportunity" that should work better when things are slow, and end up with full queues most of the time.

At least it seems like that'd work.

---

In the meantime, the technical discussion is about the best way to allocate work: how to efficiently and quickly keep at least some work in every queue without anyone going without.

OK......I'll back down....
I am a bit adamant that the kitties get what they want, but I guess this should be about what is best for the project.
I am very proud of what I have accomplished here, and would like it to remain so..
But I guess it's not all about me, now is it?

I have to stand down and realize that if the kitties are having trouble getting their kibble bowls full, it surely is affecting every other participant in the project as well.

Sorry I got so verbose.....
The kitties will share.

My thinking is that, whenever possible, no kitty should ever have an empty bowl.

Now, they may want the bowl full, and that is probably possible most of the time.

If we work from a "no queue is ever left empty" instead of "top up a queue at every opportunity" that should work better when things are slow, and end up with full queues most of the time.

At least it seems like that'd work.

Well...although the AP splitters have run out of data to split again, the kitties managed to get enough kibble in their bowls over the last couple of days to keep them happy for a bit. Maybe they can snag a few resends here and there to help a little.

If it really comes down to it, I'll just have to adjust their diet to include a portion of MB work rather than switch to another project...

And, as the situation is not likely to change as far as I can tell, I would suggest that new users default to MB only unless they manually choose to add AP work to their preferences. Leave the AP work for those who want it and know that their rigs are fast enough to process it effectively.

---

"Time is simply the mechanism that keeps everything from happening all at once."

My thinking is that, whenever possible, no kitty should ever have an empty bowl.

Now, they may want the bowl full, and that is probably possible most of the time.

If we work from a "no queue is ever left empty" instead of "top up a queue at every opportunity" that should work better when things are slow, and end up with full queues most of the time.

At least it seems like that'd work.

Well...although the AP splitters have run out of data to split again, the kitties managed to get enough kibble in their bowls over the last couple of days to keep them happy for a bit. Maybe they can snag a few resends here and there to help a little.

If it really comes down to it, I'll just have to adjust their diet to include a portion of MB work rather than switch to another project...

And, as the situation is not likely to change as far as I can tell, I would suggest that new users default to MB only unless they manually choose to add AP work to their preferences. Leave the AP work for those who want it and know that their rigs are fast enough to process it effectively.

The Lunatics unified installer is pretty cool. I'm using it to set up my crunchers here, and they're set up for AP and Multibeam.

My preferences are for Astropulse -- and to take whatever is available.

So my main cruncher has nearly all Multibeam right now, but hey, all work needs to be done, and the project never promised us there would always be work.

---

My thinking is that, whenever possible, no kitty should ever have an empty bowl.

Now, they may want the bowl full, and that is probably possible most of the time.

If we work from a "no queue is ever left empty" instead of "top up a queue at every opportunity" that should work better when things are slow, and end up with full queues most of the time.

At least it seems like that'd work.

Well...although the AP splitters have run out of data to split again, the kitties managed to get enough kibble in their bowls over the last couple of days to keep them happy for a bit. Maybe they can snag a few resends here and there to help a little.

If it really comes down to it, I'll just have to adjust their diet to include a portion of MB work rather than switch to another project...

And, as the situation is not likely to change as far as I can tell, I would suggest that new users default to MB only unless they manually choose to add AP work to their preferences. Leave the AP work for those who want it and know that their rigs are fast enough to process it effectively.

The Lunatics unified installer is pretty cool. I'm using it to set up my crunchers here, and they're set up for AP and Multibeam.

My preferences are for Astropulse -- and to take whatever is available.

So my main cruncher has nearly all Multibeam right now, but hey, all work needs to be done, and the project never promised us there would always be work.

My rigs still have the opti MB app installed, so if I have to do some MB work, all I have to do is edit my preferences to allow Boinc to fetch it.....

---

"Time is simply the mechanism that keeps everything from happening all at once."

FYI Eric has been questioned on raising MB credit values and last I remember said no, so it is a moot point.

---

FYI Eric has been questioned on raising MB credit values and last I remember said no, so it is a moot point.

Not surprising that he would not want to rock the boat....
It was just a thought.

---

"Time is simply the mechanism that keeps everything from happening all at once."

FYI Eric has been questioned on raising MB credit values and last I remember said no, so it is a moot point.

I think we're forgetting Eric's server-side credit multiplier.

The problem is that credit is defined in terms of benchmark * time, but calculated based on FLOPs.

His script samples recently returned work, finds a median work unit, calculates the FLOPs credit and the Benchmark * Time credit and adjusts a multiplier based on a 30 day average.

That should tie credit back to the definition of a cobblestone, and it should be about right -- eventually.

It will never be perfect because some processors do some instructions faster than others -- and it's different with each new chip.

---

Not possible to upload at the moment.

I was just 'panicing' :)

One look at http://fragment1.berkeley.edu/newcricket/grapher.cgi?target=%2Frouter-interfaces%2Finr-250%2Fgigabitethernet2_3;view=Octets;ranges=d shows network traffic pegged...

---

Clk2HlpSetiCty:::PayIt4ward

Two AP splitters up and running. Yea

Data Distribution State ......SETI@home # Astropulse # As of*
Results ready to send.......... 86,113 .............5 .........0m

Edit" totals
Edit 2: oops, one dropped off, now only one.

---

well i just update software (.31), new hardware (gtx 260) for cuda, and up my cashe to 4 days. so far

6/4/2009 8:29:08 PM SETI@home Reporting 1 completed tasks, requesting new tasks
6/4/2009 8:29:13 PM SETI@home Scheduler request completed: got 0 new tasks
6/4/2009 8:29:13 PM SETI@home Message from server: No work sent
6/4/2009 8:29:28 PM SETI@home Sending scheduler request: To fetch work.
6/4/2009 8:29:28 PM SETI@home Requesting new tasks
6/4/2009 8:29:33 PM SETI@home Scheduler request completed: got 0 new tasks
6/4/2009 8:29:33 PM SETI@home Message from server: (Project has no jobs available)
6/4/2009 8:30:48 PM SETI@home Sending scheduler request: To fetch work.
6/4/2009 8:30:48 PM SETI@home Requesting new tasks
6/4/2009 8:30:53 PM SETI@home Scheduler request completed: got 0 new tasks
6/4/2009 8:30:53 PM SETI@home Message from server: (Project has no jobs available)

---

Network traffic has slumped again, and getting plenty of
5/06/2009 13:07:42 SETI@home Message from server: (Project has no jobs available)
messages and no work.

---

Grant
Darwin NT

Network traffic has slumped again, and getting plenty of
5/06/2009 13:07:42 SETI@home Message from server: (Project has no jobs available)
messages and no work.

Finally got some work after all the "no jobs available messages"
Now getting
5/06/2009 13:54:28 SETI@home Scheduler request failed: Couldn't connect to server


EDIT- now getting work again & no Scheduler connection failure messages. Just checked the network taffic again. Downloads were still dead but inbound traffic had surged (hence the problems contacting the scheduler).
Outbound traffic has now picked up again, inbound reducing. Looks like things have settled down again.

---

Grant
Darwin NT

It is very clear the last couple of weeks, when AP work becomes available, the bandwidth becomes almost maxed out. And when no more AP work left, it drops almost immediately.

The end effect is that the AP clients are getting starved. When the work are availabe, most of the time my crunchers can't even upload finished work, or didn't get through to ask for work. And when connection is restored, the AP WUs are gone.

Seems like there are too many "AP only" crunchers who are not accepting MBs? If the project thinks that we need to temporarily convert all clients to MB to clear the backlog, please tell us so.

---

I will agree with AP being an opt-in sort of deal, but after looking at one of the causes of all of my pending credits/tasks.. AP should not be allowed on hosts that have less than a 250 RAC. That requirement makes sure they're going to actually try to do some crunching. 90% of the wingmen I get have less than 5 tasks, are brand new rigs, or have RACs of less than 10..because of one/both of the previous two reasons.

Which will eliminate at least 1 of my rigs crunching AP permanently.

Now, if you want to tell me that someone needs to have returned at least 10 MB's or something, I don't have an issue with that. But there are plenty of people out there with rigs in the 100-200 rac that are MORE then capable of crunching AP.

Now, if you want to tell me that someone needs to have returned at least 10 MB's or something, I don't have an issue with that. But there are plenty of people out there with rigs in the 100-200 rac that are MORE then capable of crunching AP.

Even with a 10 Multibeam limit there is going to be a problem with new people getting hit with APs for the first time. With the estimated time to completion inflated so badly at first the new cruncher takes one look at that and figures he will never be able to complete them so he just deletes BOINC and is never heard from again leaving whatever work was on their machine to time out.

It seems to me about the best way to handle this is to make Astropulse work strictly opt-in so that you actually have to come into the forums and find out what it is all about before you get any. This way, any of the set and forget crowd will be happily crunching away on MBs and wont even know what they are missing. Those that actually look at their results will eventually start to wonder why some of their wingmen are getting so much more credit and come into the forums to find out.

---

PROUD MEMBER OF Team Starfire World BOINC

...there are plenty of people out there with rigs in the 100-200 rac that are MORE then capable of crunching AP.

And there are hosts out there with S@H RAC of 3 or less which are more than capable of crunching AP, low RAC may simply mean actively working for many projects. If such a host gets an AP WU it finishes within deadline, but will build LTD and not ask for work here for quite some time.

                                                                 Joe