Turn on Pending credit!!

Message boards : Number crunching : Turn on Pending credit!!
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · Next

AuthorMessage
Profile Paul D. Buck
Volunteer tester

Send message
Joined: 19 Jul 00
Posts: 3898
Credit: 1,158,042
RAC: 0
United States
Message 28932 - Posted: 22 Sep 2004, 9:54:30 UTC - in response to Message 28780.  

> Surely we could please 90% of people by adding to the "your account" page a "#
> of results pending" figure? So, if machine "a" returns a result, this counter
> increases by one. When credit is granted/refused, it decreases. When credit is
> granted/refused, the host & user tables must be updated anyway, so cost of
> update = 0?
>
> This would surely please those who continually cry "I've returned one, where
> is it?". And would nicely sidestep the whole granted credit claimed credit
> issue?
>
> As a quick, easy (to an outsider) update whilst the system as a whole is
> capable of results browsing this seems a reasonable solution?

When results get posted to the results table, to do this we need to add a column to the users table and that will hold the number of results pending. Ok, so far, except we have now changed the update to a single record change in the results table to a change in the results table plus a change to the record in the user table. It may be that this is a minor increase in load and would not significantly change the I/O. I have not traced the update to the total credit value so I don't know it is updated when the result is posted, or using a different background process.

All of (well, almost all of) the proposals are reasonable. The problem is, will they degrade performance, or not, and if they do not, will they produce a sufficiently capable ability that satisfies all?

Digressing, the phone statistics component was developed by someone that is not on the BOINC Core development team. just like Janus has been doing some work on the web site's capabilities, this had zero impact on what the developers were doing at the time. So, like I provide a service to all by documenting, at zero cost to the development team, these are value added features that improve the overall capabilities of the system with no impact on the resources.

Next point, the growth of the user population on cp.net is a good thing, just like the almost immediate closure of the LHC@Homes opening up registration of new participants (now at 2,000). This is a movement towards realizining one of the core features of BOINC. When SETI@Home does not have enough work to issue the spare cycles are not wasted with make-work like they are in SETI@Home Classic. If and when Predictor@Home comes back with open registrations we should see good growth there too.

At this time, we have two different system 'stretching" projects available. cp.net required trickle reporting and long run times requiring a change to the scheduling mechanism. Which basically works right now. is it flawless, I don't think so. But for the moment it is good enough. SETI@Home stretches the throughput capacity of the BOINC system and has revealed problems, which are known, and I am sure that there is a background task towards fixing that problem.

Next point, cp.net, because of the massive amount of computing and the small user population (at this time) cannot afford to do redundent processing of most of the models that they are distributing. Besides, as I understant this project if there is an interesting model, the first run identifies it and they can re-run the analysis. For SETI@Home and most of the "normal" BOINC Projects the use of redundant computing is to ensure that the work unit has been tested and a viable conclusion can be assertained from the multiple tests.

Last point, any change to the system has to be figured out, then tested, then implemented. I suspect that with the reopening of the alpha and beta sites that we may begin to see more of these challenges being addressed.

If it were easy, anyone could play ...
<p>
For BOINC Documentaion: Click Me!


ID: 28932 · Report as offensive
PCZ
Volunteer tester

Send message
Joined: 3 Apr 99
Posts: 28
Credit: 20,603
RAC: 0
United Kingdom
Message 28970 - Posted: 22 Sep 2004, 12:50:15 UTC

As for the beta statement in the about box, this is for the Boinc software not SETI.

Seti is supposed to be Live but it has nearly all of the most desireable features disabled.
WU's returned take ages to be validated/credited if at all.
It doesn't have to be this way, those of us who have paticipated in the Predictor and LHC projects know how good it can be.

How about someone just telling the truth for once instead of stringing everyone along.

ID: 28970 · Report as offensive
Profile xi3piscium
Volunteer tester
Avatar

Send message
Joined: 17 Aug 99
Posts: 287
Credit: 26,674
RAC: 0
China
Message 28972 - Posted: 22 Sep 2004, 12:55:48 UTC

I would also like to see "pending turned on,
but it's out of my control, so I won't fret or sweat.
I have been using SetiSynergy for looking at stats, it
may not satify my "pending" jones, but Zain has done one
hell of job, enough info to keep the monkey off my back.
@Zain...if you read this reply, is there any way to incorp.
pending on your website? just a thought...

ID: 28972 · Report as offensive
Profile Bruno G. Olsen & ESEA @ greenholt
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 875
Credit: 4,386,984
RAC: 0
Denmark
Message 28973 - Posted: 22 Sep 2004, 12:55:57 UTC - in response to Message 28970.  

> How about someone just telling the truth for once instead of stringing
> everyone along.

What is the truth?


S@h Berkeley's Staff Friends Club ©

ID: 28973 · Report as offensive
Profile Nino Dal Borgo

Send message
Joined: 12 Dec 00
Posts: 8
Credit: 135,105,299
RAC: 0
Switzerland
Message 28982 - Posted: 22 Sep 2004, 13:11:40 UTC

IMHO, there would bye a solution for all these problems:

1) Give immediate credit to each one who has crunched a WU.
2) At the end, when 3 results are equal, revoke credits for the computers which didn't calculate the same result (if any).

Benefit:

1) I get immediate feedback my work has been counted.
2) If my credit decreases, I know something is faulty with one or more of my computers.
3) No pending credits stats.
4) The amount of processing and data to be kept should be the same as for the present solution.
ID: 28982 · Report as offensive
Grimm

Send message
Joined: 1 Sep 99
Posts: 15
Credit: 5,767,902
RAC: 2
United States
Message 28985 - Posted: 22 Sep 2004, 13:30:38 UTC - in response to Message 28698.  

Its not about getting instant credit - its about getting ANY credit. I set up a new Win98SE machine a week ago. Its been processing WUs in about 11 hours. I have yet to receive ANY credit on this machine. With the Pending Credits enabled, I would be able to see if the machine is returning bad results.


> It seems to me that the problem coming from Classic to BOINC is that Classic
> gave "instant credit" while BOINC can take some time.

ID: 28985 · Report as offensive
STE\/E
Volunteer tester

Send message
Joined: 29 Mar 03
Posts: 1137
Credit: 5,334,063
RAC: 0
United States
Message 28987 - Posted: 22 Sep 2004, 13:38:03 UTC
Last modified: 22 Sep 2004, 13:39:56 UTC

1) Give immediate credit to each one who has crunched a WU.
2) At the end, when 3 results are equal, revoke credits for the computers which didn't calculate the same result (if any).
==========

Ya right, then you would have everybody screaming and hollering that Credit was taken away from them...It would be more confusing that way than it is now with a now you have credit now you don't have credit approach...IMO

ID: 28987 · Report as offensive
Profile Papa Zito
Avatar

Send message
Joined: 7 Feb 03
Posts: 257
Credit: 624,881
RAC: 0
United States
Message 28999 - Posted: 22 Sep 2004, 14:28:56 UTC - in response to Message 28789.  


> I suppose you call being able to see your Stats on your Cell Phone Patching
> the Holes, I call it throwing Cheese to a bunch of Drowning Rats...

Gah. SETI didn't engineer that, someone else did. SETI just got the benefit of it. I agree that it's 100% pure fluff, but it wasn't SETI who wasted their time on it.

>
> It wasn't that long ago that there where only about 9000 Clients over at CPDN,
> now there is close to 20,000 over there according to the Project Stat's. That
> tells me there are a lot of People that simply don't like the Color of the
> Ship over here anyway.
>
> Myself, I've just about given up on Seti ever being a viable Project to run
> & have switched all my CPU time to CPDN like a lot of other people I know,
> I like the idea of all my Computers getting the Credit that they are due every
> day in a timely manor instead of never knowing what credit I am due and when
> the hell I'll ever get it.

And, you know, that's fine, because the whole point of BOINC is to create a common platform where people can contribute to different projects. I'm not arguing against people jumping ship (will this nautical nonsense ever stop?); you can do what you want. I'm arguing that the pending credit feature is less of a priority than some of the other problems the system has.

> I don't like the idea of my PC's not getting their Credit because Joe Blow
> decides to take a vacation and shuts his PC's off for a month & the WU's
> have to be sent back out again. And then the next Joe Blow that gets the Wu's
> decides something is wrong with his system or the way the Projects running on
> his PC so he resets the Project & the same WU's have to be sent out again
> & on & on until I see some of the WU's I used to get with 8's &
> 9's behind them & I wonder just how many times this WU is going to be sent
> out before 3 people actually finish it.

As I understand it, this is a BOINC thing, not a SETI thing. I could be wrong though.

Do you know of a better system that both distributes WUs in a timely, efficient manner AND makes sure they're 100% validated? Because there's no point in having ANY of these projects if all they're getting back is crap data.

> Also if your going to talk about
> doing it for the Science then I think theres a whole lot more to do for
> Science over there than running around looking for Little Green Men on the
> other side of the Moon...

I don't think I've ever used the "doing it for Science" line. But more to the point - I think the primary reason humanity has gotten where it has is by the desire to look outside itself, to push boundaries and to explore. While attempting to predict climate shifts and such is great, it's not nearly so culture-shifting as discovering that there is intelligent life in the universe outside of ourselves. It's the ability to look long-term that makes us so great, and SETI is looking very, very long-term.




------------------------------------


The game High/Low is played by tossing two nuclear warheads into the air. The one whose bomb explodes higher wins. This game is usually played by people of low intelligence, hence the name High/Low.
ID: 28999 · Report as offensive
PCZ
Volunteer tester

Send message
Joined: 3 Apr 99
Posts: 28
Credit: 20,603
RAC: 0
United Kingdom
Message 29000 - Posted: 22 Sep 2004, 14:28:59 UTC - in response to Message 28973.  

> > How about someone just telling the truth for once instead of stringing
> > everyone along.
>
> What is the truth?
>
>
> S@h
> Berkeley's Staff Friends Club ©


>
>

If you have nothing of value to add to this thread please refrain from posting
ID: 29000 · Report as offensive
Profile Bruno G. Olsen & ESEA @ greenholt
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 875
Credit: 4,386,984
RAC: 0
Denmark
Message 29025 - Posted: 22 Sep 2004, 15:49:23 UTC - in response to Message 29000.  

> If you have nothing of value to add to this thread please refrain from posting

You might not see the value, but ok, last post on these boards


S@h Berkeley's Staff Friends Club ©

ID: 29025 · Report as offensive
Profile Paul D. Buck
Volunteer tester

Send message
Joined: 19 Jul 00
Posts: 3898
Credit: 1,158,042
RAC: 0
United States
Message 29056 - Posted: 22 Sep 2004, 17:25:24 UTC - in response to Message 28970.  

> As for the beta statement in the about box, this is for the Boinc software not
> SETI.
>
> Seti is supposed to be Live but it has nearly all of the most desireable
> features disabled.
> WU's returned take ages to be validated/credited if at all.
> It doesn't have to be this way, those of us who have paticipated in the
> Predictor and LHC projects know how good it can be.
>
> How about someone just telling the truth for once instead of stringing
> everyone along.

Ok, re-read all my posts. Just because you do not like the answer does not mean that I did not tell the truth.

> I don't like the idea of my PC's not getting their Credit because Joe Blow
> decides to take a vacation and shuts his PC's off for a month & the WU's
> have to be sent back out again. And then the next Joe Blow that gets the Wu's
> decides something is wrong with his system or the way the Projects running on
> his PC so he resets the Project & the same WU's have to be sent out again
> & on & on until I see some of the WU's I used to get with 8's &
> 9's behind them & I wonder just how many times this WU is going to be sent
> out before 3 people actually finish it.

If that person does take a vacation, well, if the results are not returned by the deadline, the work unit is issued to someone else to process. When they are done the validaded results are processed and credit is granted.

Right now there are many things going on that are contributing to the "problem", one is the high rate of data being issued and returned. I am sure that there is work going on in the back ground to solve performance problems. What exactly is going on and when it is going to be delivered is unknown to all of us participants, and as one, I have no more visibility into the process and how it is going to turn out.

I do know that during the Beta test, the smaller alpha group knew things the larger beta group did not. In part, because the alpha group is/was hand selected for knowledge and their ability to cooperate with out causing turmoil.

> As I understand it, this is a BOINC thing, not a SETI thing. I could be wrong though.

It is a BOINC thing and a SETI@Home thing. :)

Ok, BOINC gives the project the ability to collect redundant results, with that number being set to a value of one to "n". In SETI@Home's case that number is 3. That means that SETI@Home must issue the WU a sufficient number of times to get 3 results that are "identical" (within the boundaries that they have set that indicates what "identical" means). This may reruire that the WU be issued 3 times (the minimum) to some number (I think it is 15) after which, if it gets this number of results and cannot come up with a quorum the WU is retired for additional analysis.

The cp.net project is using a value of 1, and because of the long run times of the models gives out small "trickels" of earned credit. The remaining projects currently available are all using some variation of the SETI@Home redundant computing process.

> Its not about getting instant credit - its about getting ANY credit. I set up a new Win98SE machine a
> week ago. Its been processing WUs in about 11 hours. I have yet to receive ANY credit on this
> machine. With the Pending Credits enabled, I would be able to see if the machine is returning bad
> results.

Yes, well, 11 hours is not that long of a time. Agreed, by the standards of SETI@Home classic with "increment the bean" accounting it is forever (at least it is not a nautical one). But, with a two week "window" which can extend to multiples of that window due to errors and failed compares, etc. it is no time at all. If you assume for starters that the process is one that takes a month, you have only done work for 1/60 of that time. Yes, you have not seen any credit yet.

If you "need" instant credit, do half SETI and half cp.net and there you will be getting trickles ...

As far as credit, once again I have to point to the external statistic sites that fulfill the "need" for credit numbers. By watching the site of your choice you can see your credit balance and watch it change ...

I grant that there are some things where greater visibility would be better, and I too miss that visibility, but I don't let this froth on my life's wake (sorry about that) bother me. Heck, BOINC View even gives you estimated credit in the reports ... So there is another way to see what is going on, including messages of the clients and other stuff I have not figured out how to use yet ...

<p>
For BOINC Documentaion: Click Me!


ID: 29056 · Report as offensive
PCZ
Volunteer tester

Send message
Joined: 3 Apr 99
Posts: 28
Credit: 20,603
RAC: 0
United Kingdom
Message 29066 - Posted: 22 Sep 2004, 18:19:04 UTC

Paul

This project went through alpha and beta testing and was officially launched.
You and many others keep mentioning beta tests, this project is out of beta.

The features that people are asking for are available in other Boinc projects and are working well.
Lets be honest here these features are disabled in Seti because the server infrastructure isn't man enough to cope.

Seti is the oldest and largest of the DC projects with thousands of active participants it therefore stands to reason that a lot of people will make the transition to Boinc/Seti.

Seti should have been a showcase for Boinc instead it is an embarrassment for all concerned.





ID: 29066 · Report as offensive
Profile Paul D. Buck
Volunteer tester

Send message
Joined: 19 Jul 00
Posts: 3898
Credit: 1,158,042
RAC: 0
United States
Message 29085 - Posted: 22 Sep 2004, 18:53:18 UTC - in response to Message 29066.  

PCZ,

> This project went through alpha and beta testing and was officially launched.
> You and many others keep mentioning beta tests, this project is out of beta.

Well, ok ... that depends on your definitions. When I look at the help dialog on the BOINC Work Manager it says "Beta". The link on the classic says that there is a new version:

[quote]A new version of SETI@home based on BOINC, is now available. We'll be transitioning to this new version (details are here). For now, you can use the new version or stick with SETI@home Classic. [end quote]

Neither states that BOINC is here, classic is gone. Even if we accept that the system was launched, and I don't mind talking from that perspective either, then it is a functional version. And if you have been reading my posts, and my documentation I have never denied that there are problems and serious lacks. As I posted in another thread, the glass is neither as the critics say, empty, nor as some might argue that it is full (and those people are hard to come by) and complete.

What most of the people (sometimes described as "cheerleaders", and I have no shame to be counted within that group) are saying is that it is not as bad as you might think. And even better there are alternatives to painting the sky black and yelling. Yes there are issues with the program crashing, yes there is no cross-platform GUI, yes there are issues with the benchmark, yes there are back end server issues, etc. as long as you would like ...

But, I have BOINC up and running on 6 computers. Two different Operating Systems (WIndows XP Pro and Macintosh OS X 10.3.5) and am contemplating adding one or two Linux versions on my two slowest boxes to give me experience with that stuff so I can improve my written material on those areas.

Back to the point. I am attached to 5 projects on 6 computers and the software is stable enough that I can let it run for as long as I like (24/7) with no apparant problems. I will then point to Windows (of any version) and note that what some people believe is a world class software development company has similar issues with their products (I am not, by the way, one of those that believes that Microsoft is, in fact, world class, but I digress again ...).

> The features that people are asking for are available in other Boinc projects
> and are working well.
> Lets be honest here these features are disabled in Seti because the server
> infrastructure isn't man enough to cope.

As I have consistantly point out, none of the other projects have unrestricted account creation. LHC is currently at 2,000 accounts, Predictor is at 5,707 if I recall correctly, and cp.net I think said something about being up to 11,449 ... (source is SETI Synergy.

To be honest, I have no Idea of how many users are active on what projects and at what rate they are returning results. But, I do know that you cannot easily compare apples and oranges (well, actually some physsicsts did develop an expiriment that did do just that, but I digress again ... my mind is a dusty closet) because SETI@Home is dooing redundant comparisons of the work returned, cp.net is not. You send in a trickle, and it is counted. No testing is done, no redundant running of the models because the science is different. Only when Predictor@Home or LHC@Home, or some other project goes live will we know for sure that it is the SETI@Home staff or equipment that is the long pole in the tent.

> Seti is the oldest and largest of the DC projects with thousands of active
> participants it therefore stands to reason that a lot of people will make the
> transition to Boinc/Seti.

Agreed.

> Seti should have been a showcase for Boinc instead it is an embarrassment for
> all concerned.

In a thread awhile ago, I pointed out that even if you assert that SETI@Home and BOINC went live in (July was it?) that you cannot make a vaild comparison to the current SETI@Home Classic as they are at very different points in their life-cycle. If you go back to the technical news pages on the old site, or my recopied versions on my site, you will see that life back then was anything but flawless.

Heck not too long ago we had a causalty that knocked SETI off line and those of us doing multiple projects did not even get that excited. Why? Because we have severeal projects in the can and it is not likely that they will all go down at the same time. And if they do, it is likely to take the whole internet with it. Even then, I have cp.net and have at least one WU there on all my machines so...

"What, me worry?"
<p>
For BOINC Documentaion: Click Me!


ID: 29085 · Report as offensive
Profile xi3piscium
Volunteer tester
Avatar

Send message
Joined: 17 Aug 99
Posts: 287
Credit: 26,674
RAC: 0
China
Message 29128 - Posted: 22 Sep 2004, 21:37:05 UTC - in response to Message 28973.  

> > How about someone just telling the truth for once instead of stringing
> > everyone along.
>
> What is the truth?

As we know,
There are known knowns.
There are things we know we know.
We also know
There are known unknowns.
That is to say
We know there are some things
We do not know.
But there are also unknown unknowns,
The ones we don't know
We don't know.

Donald Rumsfeld
Feb. 12, 2002, Boinc news briefing :)

Just trying to add a little humour...it's
good for the soul.



ID: 29128 · Report as offensive
Profile xi3piscium
Volunteer tester
Avatar

Send message
Joined: 17 Aug 99
Posts: 287
Credit: 26,674
RAC: 0
China
Message 29143 - Posted: 22 Sep 2004, 22:28:12 UTC - in response to Message 29066.  
Last modified: 22 Sep 2004, 22:29:10 UTC

> Seti should have been a showcase for Boinc instead it is an embarrassment for
> all concerned.

@PCZ...your points are valid
ID: 29143 · Report as offensive
STE\/E
Volunteer tester

Send message
Joined: 29 Mar 03
Posts: 1137
Credit: 5,334,063
RAC: 0
United States
Message 29178 - Posted: 23 Sep 2004, 0:53:34 UTC
Last modified: 23 Sep 2004, 0:53:53 UTC

> If you have nothing of value to add to this thread please refrain from posting

You might not see the value, but ok, last post on these boards
=========

I think I would have gone down with a little more fight than that Bruno, a more appropriate response would have been like Kiss me where the sun don't shine...IMO

ID: 29178 · Report as offensive
Profile xi3piscium
Volunteer tester
Avatar

Send message
Joined: 17 Aug 99
Posts: 287
Credit: 26,674
RAC: 0
China
Message 29213 - Posted: 23 Sep 2004, 2:24:13 UTC

Hey guys...don't mind the rumsfeld thing,
just trying to lighten things up a bit.
My apologies to all...

Sincerely,

Joe
ID: 29213 · Report as offensive
Profile Kevin N. Shapley
Volunteer tester
Avatar

Send message
Joined: 1 Jan 00
Posts: 100
Credit: 2,539,295
RAC: 0
United States
Message 29231 - Posted: 23 Sep 2004, 3:11:58 UTC - in response to Message 29213.  

> Hey guys...don't mind the rumsfeld thing,
> just trying to lighten things up a bit.
> My apologies to all...
>
> Sincerely,
>
> Joe
>
-
-
I thought it was very funny, my kind of humour...ROTFLMAO

Thanks.


<a> [/url]
-
Oderint dum metuant
ID: 29231 · Report as offensive
Profile Papa Zito
Avatar

Send message
Joined: 7 Feb 03
Posts: 257
Credit: 624,881
RAC: 0
United States
Message 29238 - Posted: 23 Sep 2004, 3:35:04 UTC - in response to Message 29056.  


> > As I understand it, this is a BOINC thing, not a SETI thing. I could be
> wrong though.
>
> It is a BOINC thing and a SETI@Home thing. :)
>
> Ok, BOINC gives the project the ability to collect redundant results, with
> that number being set to a value of one to "n". In SETI@Home's case that
> number is 3. That means that SETI@Home must issue the WU a sufficient number
> of times to get 3 results that are "identical" (within the boundaries that
> they have set that indicates what "identical" means). This may reruire that
> the WU be issued 3 times (the minimum) to some number (I think it is 15) after
> which, if it gets this number of results and cannot come up with a quorum the
> WU is retired for additional analysis.
>
> The cp.net project is using a value of 1, and because of the long run times of
> the models gives out small "trickels" of earned credit. The remaining
> projects currently available are all using some variation of the SETI@Home
> redundant computing process.

Rock on. I hear you. One must wonder, though, if CPDN's choice to forego redundant checking will bite them at some point.




------------------------------------


The game High/Low is played by tossing two nuclear warheads into the air. The one whose bomb explodes higher wins. This game is usually played by people of low intelligence, hence the name High/Low.
ID: 29238 · Report as offensive
STE\/E
Volunteer tester

Send message
Joined: 29 Mar 03
Posts: 1137
Credit: 5,334,063
RAC: 0
United States
Message 29296 - Posted: 23 Sep 2004, 9:14:51 UTC

Rock on. I hear you. One must wonder, though, if CPDN's choice to forego redundant checking will bite them at some point.
==========

The main reason they probably are skipping the Redundant Checking is because of the length of the Work Units. My P3 850 has an estimated finish Time of 66-67 days & that would be running it 24/7...Running mutiple Projects or any down time with the PC and it will only increase the length of time it will take to complete...

They would be waiting around forever for 3 results to be turned in on most of the Work Units I would think...



ID: 29296 · Report as offensive
Previous · 1 · 2 · 3 · 4 · Next

Message boards : Number crunching : Turn on Pending credit!!


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.