If you force the client to poll exactly at every cooldown period, you'll never get any work because you always hit the scheduler a little bit too soon. The cooldown is measured from the completion of one request to the start of the next one (as the server sees them), so you have to allow some time for the server to process the request.

Perhaps even too many and that leads to the current situation...

The problem is that Boinc prefers handing any new tasks to those hosts that already have work.

Hosts that have work do scheduler requests as often as the cooldown allows. Empty ones end up having several hour backoffs between requests. So the empty hosts that could return the results almost immediately never get them.

So why not set NNT and allow the rest of us the chance to get a few?

---

CAPT Siran d'Vel'nahr - L L & P _\\//
Winders 11 OS? "What a piece of junk!" - L. Skywalker
"Logic is the cement of our civilization with which we ascend from chaos using reason as our guide." - T'Plana-hath

So why not set NNT and allow the rest of us the chance to get a few?

I suppose everyone want to "crunch the last WU"

Does not really bother me, I leave my 2 Windows machines on and get what I get. As I use them both during the day and am a believer in not powering them on and off they will be on anyway.

So why not set NNT and allow the rest of us the chance to get a few?

I suppose everyone want to "crunch the last WU"

Well 2 out of my last 5 Tasks were USEFUL, because they helped to complete Workunits that still needed Validation

What is the point of crunching work on a super fast GPU that has already been VALIDATED days ago by 2 or 3 slower hosts ?

I would estimate that around 90% of Mr 1337's tasks In Progress are already validated !

My final Pending https://setiathome.berkeley.edu/workunit.php?wuid=3953270075 was one of 7 that had him as the original wingman. The other 6 are already validated days ago, but the 1337 guy still hasn't got around to them yet !

So why not set NNT and allow the rest of us the chance to get a few?

I suppose everyone want to "crunch the last WU"

Well 2 out of my last 5 Tasks were USEFUL, because they helped to complete Workunits that still needed Validation

What is the point of crunching work on a super fast GPU that has already been VALIDATED days ago by 2 or 3 slower hosts ?

I would estimate that around 90% of Mr 1337's tasks are already validated !

My final Pending https://setiathome.berkeley.edu/workunit.php?wuid=3953270075 was one of 7 that had him as the original wingman. The other 6 are already validated days ago, but the 1337 guy still hasn't got around to them yet !

Hi Keith,

I looked at the hosts of several of the top 10 heavy hitters and very few of them had very few tasks on their hosts. None of them had what Mr. 1337 has. Some people just think they are better and more deserving than all others and have no compassion at all. What can you say? Par for the course.

Have a great day! :)

Siran

---

CAPT Siran d'Vel'nahr - L L & P _\\//
Winders 11 OS? "What a piece of junk!" - L. Skywalker
"Logic is the cement of our civilization with which we ascend from chaos using reason as our guide." - T'Plana-hath

What is the point of crunching work on a super fast GPU that has already been VALIDATED days ago by 2 or 3 slower hosts ?

The host returns them when it returns them. Do you really expect me to click through all the tasks on the web site one by one to figure out what task have a filled quorum and what tasks do not?

I looked at the hosts of several of the top 10 heavy hitters and very few of them had very few tasks on their hosts. None of them had what Mr. 1337 has. Some people just think they are better and more deserving than all others and have no compassion at all. What can you say? Par for the course.

Lol I guess was then good for your blood pressure that you did not check their hosts couple of weeks ago ;)

What is the point of crunching work on a super fast GPU that has already been VALIDATED days ago by 2 or 3 slower hosts ?

The host returns them when it returns them. Do you really expect me to click through all the tasks on the web site one by one to figure out what task have a filled quorum and what tasks do not?

If this project was able to cancel Unnecessary, Unstarted Tasks then you wouldn't need to.
I can see that it is impossible to do that manually.

I was trying to be considerate of people's electrical bills, when I suggested that a few days ago.
Then the NEEDED results would be returned more quickly.

Then the NEEDED results would be returned more quickly.

If that is what they rely need is simple. Send the task for the high reliable hosts (they have the list of them) and all work will be returned ASAP. But instead of that they decided to send several resends of the same WU for a random number of hosts.
If you look just the top 20 reliable hosts has the capacity of crunching around 100 K WU per day.
And Before the thread could be ignited. I'm not telling anything elitists, just saying if they want the WU returned fast they need to use this fasts & reliable hosts. Only 1 of the ian hosts could crunch >10 K per day.
I was talking just about crunching speed and a way to garante the last few WU to be returned fast so they could close the science DB for good, nothing else, no rants please.
Anyway the total number of not rely validated WU (those who are rely waiting for another wingmen) at this time must be very low maybe less than 300K.

---

Lol I guess was then good for your blood pressure that you did not check their hosts couple of weeks ago ;)

I agree, sure not good for the blood pressure do that, only my host has > 130K WU 3 weeks ago (all served & returned BTW) and there where others who has a lot more. Not forgetting the anonymous who still has >40K Wu today! That sure could be a cause for a heart attack! LOL

---

At this rate soon they will pass the Results out in the field too. LOL

I think that will probably be at least 24 hours at the current rate.

Average turnaround times are now more than 7 days for both V8 and AstroPulse, they were below 7 days for the last few days.

"Results waiting for db purging" is now ahead of both "Results out in the field" and "Results returned and awaiting validation".

There are now probably between 2 and 3 In Progress tasks for each Workunit, with between 1 and 2 Tasks awaiting validation for each Workunit.

I've got several like this one - looks like it could be the middle of May before that one's cooked.

I've got several like this one - looks like it could be the middle of May before that one's cooked.

Both the_0 and_3 hosts have contacted the servers today, although it looks like the 0 had a batch of ghosts.
I wouldn't hold out much hope for the _1

Then the NEEDED results would be returned more quickly.

If that is what they rely need is simple. Send the task for the high reliable hosts (they have the list of them) and all work will be returned ASAP. But instead of that they decided to send several resends of the same WU for a random number of hosts.
If you look just the top 20 reliable hosts has the capacity of crunching around 100 K WU per day...

Good comment.

Note that an important part of the experiment that is 's@h and boinc' is to test the feasibility of using "untrusted" computers in a distributed computer system. Part of that experiment requires distributing the tasks randomly between all the volunteer clients.

Ok, so that random distribution is skewed somewhat by the number of WUs that the fast systems have gobbled through...

Another aspect is that we have been processing WUs for over two decades. The research work will continue for some years yet. Hence there will be no need for any rush to finish the last few WUs.


All very good science.

And some great participant fun along the way!

Enjoy!!

Keep searchin',
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

Hence there will be no need for any rush to finish the last few WUs.

If was in the past i say you are right, but that is not what as posted by Eric on the FB.
Unless i understand something wrong they ask to return the remaining WU ASAP to close the Science DB.
That is one of the reasons why they are generating extra wingmen for all the reaming WU's.

---

https://www.facebook.com/groups/317689381579505/permalink/3463142420367503/

Eric's reply is 5th or 6th down.

https://www.facebook.com/groups/317689381579505/permalink/3463142420367503/

Eric's reply is 5th or 6th down.

. . That link only shows the original post by Mr Cook. No comments are viewable including Eric's :(

Stephen

:(

https://www.facebook.com/groups/317689381579505/permalink/3463142420367503/

Eric's reply is 5th or 6th down.

. . That link only shows the original post by Mr Cook. No comments are viewable including Eric's :(

Stephen

:(

This is his post:

Eric Korpela We're resending a lot these days to try to get the outstanding ones validated and into the science database so we have a "final complete set".

You need to click where you see 9 Commentaries to see the rest.

<edit> @Stephen
Finally we have the same number of WU In progress.... zero! (0). LOL

---

It might not load comments if you’re not a Facebook user. Just a guess.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

It might not load comments if you’re not a Facebook user. Just a guess.

Strange. Why i could see here? I'm not a FB user. Not even have one FB account AFAIK. LOL

---