Faster credit: 4 copies of each WU initially being sent out now


log in

Advanced search

Message boards : Number crunching : Faster credit: 4 copies of each WU initially being sent out now

Previous · 1 · 2 · 3 · 4 · 5 · Next
Author Message
JAF
Avatar
Send message
Joined: 9 Aug 00
Posts: 289
Credit: 163,203
RAC: 0
United States
Message 58202 - Posted: 29 Dec 2004, 13:24:44 UTC - in response to Message 58197.
Last modified: 29 Dec 2004, 13:27:09 UTC

> I do agree with you - I really don't care about the credits. For me it's more
> like a progress indicator, nothing else! They can remove the credits
> completely since I use "View Results" in "My Account".
>
But isn't "View Results" in "My Account" also a form of credit (similar to Seti Classic)? I don't quite see the point.

Also, using results as a measurement really distorts the cross platform "picture". One result in another Boinc project may equal 500 Seti results. I thought the credit system we are using now was supposed to attempt to show our participation in whatever project(s) we choose.

[edit] since I'm only participating in Seti at this time, my perspective may be way off. I'm sure I will be informed of this shortly (lol).....
____________
<img src='http://www.boincsynergy.com/images/stats/comb-912.jpg'>

Scott Brown
Send message
Joined: 5 Sep 00
Posts: 110
Credit: 59,739
RAC: 0
United States
Message 58213 - Posted: 29 Dec 2004, 14:29:04 UTC - in response to Message 58194.

> > No credit? No way. I think most people wouldn't be willing to do work
> for
> > berkeley if they got no credit at all...
>
> Actually, I would guess that most people don't really care. It is just that
> the ones that seem to care are very noisy about their, ahem, dissatisfaction.
>

On most things I would agree with you Paul, but here I think you are incorrect. Research on volunteering and altruism show pretty conclusively that most people need to feel appreciated for their efforts. This can come in the form of accomplishment (e.g., credits or other rewards) but could also have other forms such as positive feedback (which the small staff clearly have no time to do) and awards (e.g., the printable certificates from SETI classic). Even those who appear to have the purest most altruistic motives, when given detailed examination of their contribution, still have some identifiable form of feedback (even if it is nothing more than a generalized feeling of being part of a solution, etc.)

Profile Paul D. Buck
Volunteer tester
Send message
Joined: 19 Jul 00
Posts: 3898
Credit: 1,158,042
RAC: 0
United States
Message 58217 - Posted: 29 Dec 2004, 15:39:36 UTC - in response to Message 58213.

Scott,

> On most things I would agree with you Paul, but here I think you are
> incorrect. Research on volunteering and altruism show pretty conclusively
> that most people need to feel appreciated for their efforts. This can come in
> the form of accomplishment (e.g., credits or other rewards) but could also
> have other forms such as positive feedback (which the small staff clearly have
> no time to do) and awards (e.g., the printable certificates from SETI
> classic). Even those who appear to have the purest most altruistic motives,
> when given detailed examination of their contribution, still have some
> identifiable form of feedback (even if it is nothing more than a generalized
> feeling of being part of a solution, etc.)

You may be right. I may have stressed the "at all" part a bit. I just mean that most people would look at it like their gas guage. Yes it is rising ... cool ... and then move on with life. When you look in the forums you see that the vast majority of participants never show up at all ... we have about 60,000 participants; yet the vast majority of complaints are from a very vocal minority.

I saw the same thing in the SETI@Home Classic message boards in the past. And it is likely that it is going on even now. Endless debate on the fact that participant "A" is getting more than participant "B" ...
____________

Scott Brown
Send message
Joined: 5 Sep 00
Posts: 110
Credit: 59,739
RAC: 0
United States
Message 58230 - Posted: 29 Dec 2004, 17:14:13 UTC - in response to Message 58217.

> Scott,
>
> You may be right. I may have stressed the "at all" part a bit. I just mean
> that most people would look at it like their gas guage. Yes it is rising ...
> cool ... and then move on with life. When you look in the forums you see that
> the vast majority of participants never show up at all ... we have about
> 60,000 participants; yet the vast majority of complaints are from a very vocal
> minority.
>
> I saw the same thing in the SETI@Home Classic message boards in the past. And
> it is likely that it is going on even now. Endless debate on the fact that
> participant "A" is getting more than participant "B" ...
>

You are certainly right about this being an ongoing issue from even before BOINC. But I would say that the importance of the credit system has been vastly underestimated by the Berkeley staff. It would seem that credit, competition, etc. has been an important motivator for not only participation but also the level of participation of the user base (e.g., all the SETI-only farm building, etc.). I would consider it more than a cosmetic issue to be resolved after everything is running perfectly (though it of course was and should have been lower priority than getting the basic system up and running).

As for the silent masses of the 60,000 participants, none of us knows for sure. Of course a quick e-mailed survey from Berkeley (maybe find a social sicence or marketing grad student looking for a master's project) could answer much of this and provide smoother sailing all around. The credit system may become even more of an issue as all of the other projects get back in gear providing competition for the user base. For example, CPDN has already increased their fixed per-trickle credit (from about 78 to about 95 I think) due to complaints that the same work for other projects got more credit. I wonder what would happen if there were another fixed credit project like CPDN without the huge workunit size and faster processor requirements?

BTW Paul, it is nice to see you back from vacation and very active on the boards again...your wisdom was missed :)

Ned Slider
Send message
Joined: 12 Oct 01
Posts: 668
Credit: 4,375,315
RAC: 0
United Kingdom
Message 58241 - Posted: 29 Dec 2004, 19:56:05 UTC

December 28, 2004
More on redundancy parameters. After some discussion we have decided to send out 4 copies of each workunit and continue with a quorum of 3 for validation. Validation should still proceed at a good pace because a single result in error will not mean distributing an additional workunit copy (and waiting for it's result) in order to reach a quorum.


I think the above is a step in the right direction. Presumably the reason for experimenting with this is to speed up the credit process, meaning Berkeley can validate results faster and get them off the system and us users get our credit faster too. This will obviously help in situations where one result is not returned within the 14 day cut off period but won't affect awarded credit in the way the previous suggestion of validating only two results would have.

Good work guys,

Ned

____________
*** My Guide to Compiling Optimised BOINC and SETI Clients ***
*** Download Optimised BOINC and SETI Clients for Linux Here ***

N/A
Volunteer tester
Send message
Joined: 18 May 01
Posts: 3718
Credit: 93,649
RAC: 0
Message 58247 - Posted: 29 Dec 2004, 21:07:16 UTC - in response to Message 58217.

...we have about 60,000 participants; yet the vast majority of complaints are from a very vocal minority.
The squeaky wheels get the grease.

I want to be able to convince everyone who's got a computer to join in on BOINC regardless of which project they decide to support. The only way that I can do that is to make comments on how things are right now and what would be convincing for potential users. Keep in mind, most people have idle cycles to spare for distributed computing, and no idea about how to use computers in the first place.

There are three kinds of potential users: Compu-aficionados who want to push their hardware to the limit, people with a vested interest in already existing projects like SETI or E@H, and (at the risk of sounding snobbish) the compu-phobes and -illiterate. The first two would jump in to support BOINC, but I want to be able to convince the latter to join in on the fun - Yes, I said fun.

I saw the same thing in the SETI@Home Classic message boards in the past. And it is likely that it is going on even now. Endless debate on the fact that participant "A" is getting more than participant "B" ...
The only way I can convince the last group of potential users to join is by pushing the "credit" angle. And if people are going to screw around with the credits and such, then I can't pitch the notion of joining BOINC in my users groups.

FWIW, if I were a better programmer, I'd have made the changes in the source code myself and made the alterations public.
____________

B-Roy
Send message
Joined: 4 May 03
Posts: 220
Credit: 169,888
RAC: 67
Austria
Message 58260 - Posted: 29 Dec 2004, 21:52:42 UTC

If I understood it right, 4 copies of every wu are sent out now. however, my result page of recently received wus shows that all of them have only been sent 3 times. any guess?
An example is here , which has been sent today.
____________

Profile Keck_Komputers
Volunteer tester
Avatar
Send message
Joined: 4 Jul 99
Posts: 1575
Credit: 1,704,721
RAC: 754
United States
Message 58261 - Posted: 29 Dec 2004, 21:54:36 UTC - in response to Message 58260.

> If I understood it right, 4 copies of every wu are sent out now. however, my
> result page of recently received wus shows that all of them have only been
> sent 3 times. any guess?
> An example is <a> href="http://setiweb.ssl.berkeley.edu/workunit.php?wuid=6978732">here[/url] ,
> which has been sent today.
>
This only applies to work generated after the announcement. There will be 2 and 3 workunit types floating around for a month most likely until they clear out.
____________
BOINC WIKI

BOINCing since 2002/12/8

Tom Gutman
Send message
Joined: 20 Jul 00
Posts: 48
Credit: 219,500
RAC: 0
United States
Message 58264 - Posted: 29 Dec 2004, 21:58:18 UTC - in response to Message 58154.

> News - Technical News - Server Status
>
> December 28, 2004
> More on redundancy parameters. After some discussion we have decided to send
> out 4 copies of each workunit and continue with a quorum of 3 for validation.
> Validation should still proceed at a good pace because a single result in
> error will not mean distributing an additional workunit copy (and waiting for
> it's result) in order to reach a quorum.
>
Ah, the arrogance of excessive processing power. We are now going from inflating the processing requirements by 50% (over that required for the science) to inflating it by 100%. That's wasting 50% of the available CPU power in order to try to make the credit system semi-reasonable (no CPU based credit system is going to be consistent) and fast.

At this point I'd rather have the unit sent out to two machines (the minimum needed for the science) initially and give all users the credit they claimed (if the unit is valid). Let the cheaters cheat -- so what? Along with the other figures, publish the average credit per work unit. Since over the long term variations in work units will average out, the cheaters will be very obvious, when they show up as claiming ten times more credit per work unit than the average.
____________

------- Tom Gutman

Profile Paul D. Buck
Volunteer tester
Send message
Joined: 19 Jul 00
Posts: 3898
Credit: 1,158,042
RAC: 0
United States
Message 58265 - Posted: 29 Dec 2004, 22:05:08 UTC - in response to Message 58230.

> You are certainly right about this being an ongoing issue from even before
> BOINC.

¥eah, there was some forlorn hope that it might get better. But oh my, during Beta the complaints! Even though the credit was wiped, it was silly people moaning about how it was calculated.

> As for the silent masses of the 60,000 participants, none of us knows for
> sure. Of course a quick e-mailed survey from Berkeley (maybe find a social
> sicence or marketing grad student looking for a master's project) could answer
> much of this and provide smoother sailing all around.

Now there is an idea! :)

> BTW Paul, it is nice to see you back from vacation and very active on the
> boards again...your wisdom was missed :)

Thanks, I appreciate that. Sitll recovering I think. But the good news is that I am making progress to migrate the site, I got it down to only 3,011 broken links. Well, tomorrow is another day and I will try to get the Glossary done, and with REAL luck, the core of the Web Site Owner's Manual too ... (I did the FAQs today) ...

The cool thing is that I can do the headers and footers and then do the insides at my leasure (well, it seems to be working so far ...) ...


____________

N/A
Volunteer tester
Send message
Joined: 18 May 01
Posts: 3718
Credit: 93,649
RAC: 0
Message 58283 - Posted: 30 Dec 2004, 0:12:30 UTC - in response to Message 58264.

Along with the other figures, publish the average credit per work unit. Since over the long term variations in work units will average out, the cheaters will be very obvious, when they show up as claiming ten times more credit per work unit than the average.
Absolutely brilliant!
____________

Profile tpl
Avatar
Send message
Joined: 12 Nov 03
Posts: 364
Credit: 193,269,205
RAC: 12,451
Germany
Message 58305 - Posted: 30 Dec 2004, 3:04:42 UTC

I become the last credits Yesterday 18.20.....after this time
Units with 3 results are pending. :-((
____________

Profile RichaG
Volunteer tester
Avatar
Send message
Joined: 20 May 99
Posts: 1690
Credit: 3,631,246
RAC: 1,210
United States
Message 58306 - Posted: 30 Dec 2004, 3:15:48 UTC

Has anybody seen wus sent out to 4 hosts?
My last few wus at this time are still being only sent to 3 hosts.
____________
Red Bull Air Racing

Gas price by zip at Seti

Profile kinnison
Avatar
Send message
Joined: 23 Oct 02
Posts: 107
Credit: 682,151
RAC: 0
United Kingdom
Message 58310 - Posted: 30 Dec 2004, 4:27:21 UTC

Along with the other figures, publish the average credit per work unit. Since over the long term variations in work units will average out, the cheaters will be very obvious, when they show up as claiming ten times more credit per work unit than the average.

-------------
A very good idea!
I was just browsing thru a few results now and this one caught my eye:

http://setiweb.ssl.berkeley.edu/workunit.php?wuid=6030554

claiming (and getting) over 96 credits for a single WU. That *has* to be wrong!!

I myself am a lowly person with only 1 pc getting some 200 credits a day, but looking at the team I'm in, the UK overclockers, I'm sure that credit is a definite motivating factor for some of the top members of that team. They have created SETI "farms" as was mentioned earlier in this thread.

My main motivation is simply to utilise my computer's processing capability in some "useful" fashion, rather than it standing idle. SETI seems like a very good way to do that. So is CPDN. Some sort of scoring system is essential though, surely?

As far as the SETI scoring goes, I'd just like to get exactly the same number of credits as a 500Mhz pentium doing the same WU, or a hyper-threaded P4-3000. Whether that's 20 or 40 or 100, doesn't matter to me as long as its the same for everyone.


Chris.
____________
<img border="0" src="http://boinc.mundayweb.com/one/stats.php?userID=268&amp;prj=1&amp;trans=off" /><img border="0" src="http://boinc.mundayweb.com/one/stats.php?userID=268&amp;prj=4&amp;trans=off" />

Profile Benher
Volunteer developer
Volunteer tester
Send message
Joined: 25 Jul 99
Posts: 517
Credit: 465,152
RAC: 0
United States
Message 58329 - Posted: 30 Dec 2004, 5:52:32 UTC

>I was just browsing thru a few results now and this one caught my eye:
>http://setiweb.ssl.berkeley.edu/workunit.php?wuid=6030554

>Athlon XP 1800+
>Measured floating point speed 1254.9 million ops/sec
>Measured integer speed 3636.89 million ops/sec

Yea,
Thats an obvious person using a BOINC manager source code rewritten to multiply float and integer benchmarks by some number.

The odd thing is, they somehow processed a WU which was also handed out to another cheater.

Someone in some user group probably made the executable and is passing it around to buddies.

Check most of their pending results and I think you will find they virtually never get their inflated claims...and will not until a sizable percentage of hosts are cheating.


____________

Profile KWSN - MajorKong
Volunteer tester
Avatar
Send message
Joined: 5 Jan 00
Posts: 1406
Credit: 793,366
RAC: 0
United States
Message 58339 - Posted: 30 Dec 2004, 7:51:28 UTC - in response to Message 58329.

> >I was just browsing thru a few results now and this one caught my eye:
> >http://setiweb.ssl.berkeley.edu/workunit.php?wuid=6030554
>
> >Athlon XP 1800+
> >Measured floating point speed 1254.9 million ops/sec
> >Measured integer speed 3636.89 million ops/sec
>
> Yea,
> Thats an obvious person using a BOINC manager source code rewritten to
> multiply float and integer benchmarks by some number.
>
> The odd thing is, they somehow processed a WU which was also handed out to
> another cheater.
>
> Someone in some user group probably made the executable and is passing it
> around to buddies.
>
> Check most of their pending results and I think you will find they virtually
> never get their inflated claims...and will not until a sizable percentage of
> hosts are cheating.
>
>
>

Umm... I beg to differ about that host. Just what is going on is not all that obvious. I also have an Athlon-XP1800+ running Win2k, and his benchmarks are near what mine are (one is higher, one is lower). With that work unit, he is claiming about 2x what I claim, on average on that machine, but then his CPU time is about 2x what my average time is.

Without knowing his exact hardware setup on that machine, saying 'cheater' is a bit premature. My wife's computer, for instance, is an Athlon-XP 2400+, and has better benchmark scores than mine. So, one would expect that her computer would finish work units quicker than mine... Wrong! Her computer has a cheap motherboard with integrated video . End result: work units take close to an hour longer on her 2400+ than on my 1800+, therefore claiming WAY more credit (60ish vs. 40ish).

Another factor to consider is that all work units are not 'equally busy'. Some units, due to things like AngleRange and amount of 'signals' do a LOT more Gaussian/Pulse/Triplet searches than others.

It is my opinion that one will find (if one investigates) that either the host in question is not very '133+' or that work unit was 'busy', or both. I hate cheaters as much (and likely far more) as the next guy, but I don't see any conclusive evidence here.
____________
“The more corrupt the state, the more numerous the laws.”
― Tacitus, The Annals of Imperial Rome

“Truth is confirmed by inspection and delay; falsehood by haste and uncertainty.”
― Tacitus

Profile KWSN - MajorKong
Volunteer tester
Avatar
Send message
Joined: 5 Jan 00
Posts: 1406
Credit: 793,366
RAC: 0
United States
Message 58345 - Posted: 30 Dec 2004, 8:41:35 UTC - in response to Message 58264.


> Ah, the arrogance of excessive processing power. We are now going from
> inflating the processing requirements by 50% (over that required for the
> science) to inflating it by 100%. That's wasting 50% of the available CPU
> power in order to try to make the credit system semi-reasonable (no CPU based
> credit system is going to be consistent) and fast.
>
> At this point I'd rather have the unit sent out to two machines (the minimum
> needed for the science) initially and give all users the credit they claimed
> (if the unit is valid). Let the cheaters cheat -- so what? Along with the
> other figures, publish the average credit per work unit. Since over the long
> term variations in work units will average out, the cheaters will be very
> obvious, when they show up as claiming ten times more credit per work unit
> than the average.
>

I respectfully disagree with your statement that two is the minimum required 'for the science'. In the case of a disagreement in the results, how can you tell who is more correct?

Also, I do agree that initially sending out 4 copies is a waste. 3 is the number of initial copies I would send out. But, in any case, 4 is WAY fewer than S@H-Classic sends out. Analysis shows that Classic sent out each work unit AT LEAST 7 or 8 times, and I have heard Dr. Anderson say (In a video of a speach he gave) that it was 'at least a dozen times'. So, S@H/BOINC is 2 to 3 times more efficient than Classic.

I must say that I liked the old '3' setup very well. I have some credit still pending from August, yet you don't hear me whineing.

In my long experience with DC projects (started my 1st S@H account back in May 1999) I have found that there are 3 types of participants. Those in it for the science (and could care less about credits), those in it for the credits ONLY, and the largest group - those in it for varying degrees of both. I am in the last group (at about 80% science, and 20% credits). I find that the credits are a useful measure of my own progress in the project, but I do not get my male hormones all in an uproar because someone else has more credits than me. I am so anti-cheater because of the harm that cheating can do to the science, not because some cheating loser got their nickname up on top of a leader-board.

The credits system for BOINC projects was designed the way it was for two reasons. First, the old 'wu count' had to be dropped in favor of a numerical score. Different type work units on the same project, as well as different BOINC projects entirely, are not going to have anywhere near the same CPU requirements, and a 'count' would not be fair to people that get a bunch of 'long' ones. The 'credits' score makes it possible to combine work from multiple work unit types and even multiple BOINC projects into ONE total, overall score. See my friend Toby's stats site if you are curious about combined project stats. http://stats.kwsn.net Check the last line in the table.

The second reason that the BOINC credits system was designed this way is to make it MUCH harder to cheat. The credits claimed/granted system is in place to make it virtually impossible for someone to fake a result file and return it for credit the way that SOME were caught doing a few years ago in S@H-Classic. The result must be validated by comparing the returned data (NOT the 'claimed credits as some people have mistakenly believed since S@H/BOINC went 'live') BEFORE any credit is granted. The 'grant the middle value of the three claimed credits' feature is there to prevent people from jacking their benchmarks or cputime to inflate their credits. It USED to be (back during Beta) the LOWEST value, but quite a number of the beta testers started complaining about that because of a few Solaris hosts, due to a bug, claiming very low credit (~0.03) on large numbers of work units, chopping down everyone else that did the same units as they did.

All in all, I am very happy with the credits system in place since beta->live, and I wish they wouldn't change it...
____________
“The more corrupt the state, the more numerous the laws.”
― Tacitus, The Annals of Imperial Rome

“Truth is confirmed by inspection and delay; falsehood by haste and uncertainty.”
― Tacitus

Profile Roks
Send message
Joined: 20 Dec 02
Posts: 55
Credit: 137,776
RAC: 0
Slovenia
Message 58351 - Posted: 30 Dec 2004, 9:31:20 UTC

I think I have a better idea. Instead of taking away top and lowest claimed credit and then grant the middle claimed credit, calculate the average of all three results. For example: 20, 30, and 60 credits are claimed. With system seti has now, all three would get 30 credits. But with "my" system, they would get 36,7 granted credit. I think that it's more fair, because my laptop for example (amd xp 2800+) almost always claims around 70 credits. I'm not cheating, it also does around 7 hours/wu of work. And I never get even near consearning the claimed credits... BOINC, take the average credit of all claimed.
____________
<img src="http://www.boincstats.com/stats/banner.php?cpid=d2319b8f0ad14565556d0ba45b64e779">

N/A
Volunteer tester
Send message
Joined: 18 May 01
Posts: 3718
Credit: 93,649
RAC: 0
Message 58375 - Posted: 30 Dec 2004, 13:00:33 UTC - in response to Message 58351.
Last modified: 30 Dec 2004, 13:09:35 UTC

...yes, but let's say that there's a WU that's worth 1, 1, and 120. We stop the cheater by granting 40cr, but we're also inflating everyone else by the same amount. In your example of {20, 30, 60}, x&#x0305;=36.6&#x0305;cr, but {1, 1, 120} x&#x0305;=40.6&#x0305;cr. There's a huge difference between dropping a claim by &cong;&#x2153; and booting another's forty-fold.

OK, so that's an extreme case, I grant you, but I think I've made my point: plain averages won't thwart the problem in its entirety. I'm working on a weighted-average solution...

[EDIT] Fixed &#x; escapes...
____________

Profile Roks
Send message
Joined: 20 Dec 02
Posts: 55
Credit: 137,776
RAC: 0
Slovenia
Message 58379 - Posted: 30 Dec 2004, 13:11:41 UTC

Sure, but check out the wus, you'll see that there are no such big differences (at most around 40 credits). But look at that. My laptop claimed 70 credits, the other 2 were linux boxes and claimed 23 and 24 credits. My granted credit was 46 credits lower than claimed, even though I'm not a cheater. If the average was calculated, I'd get around 40 credits, much more acceptable amount.
____________
<img src="http://www.boincstats.com/stats/banner.php?cpid=d2319b8f0ad14565556d0ba45b64e779">

Previous · 1 · 2 · 3 · 4 · 5 · Next

Message boards : Number crunching : Faster credit: 4 copies of each WU initially being sent out now

Copyright © 2014 University of California