AP vs MB, credits and all that jazz........

Message boards : Number crunching : AP vs MB, credits and all that jazz........
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 901552 - Posted: 30 May 2009, 16:06:56 UTC

I really think someone has to make a statement about all of this..........

So I will.

I am in this for the science.

BUT...

Neither I, nor the project, nor the optimizers, nor the community as a whole, would have advanced to this point if it were not for the FUN of credits........

That is the entire point of the credit game.....to give users something fun to do along with the science.

Do you think for a moment that if I could not engage in the battle for supremacy (LOL) in the credit wars that I would have gone to the extent that I have simply to better the science??

I am a believer, but I am not THAT pure...LOL again.

If this Boinc/Seti thingy were simply a cold, hard science project, I think participation would be a fraction of what it is today.

Soooooooooo....
Now here is a possible new thought.....
If Matt and Eric need to balance the current work being done...AP vs MB....

How about a temporary boost in MB credits....
Offer the users a 2 week 30% increase in MB credit...
Or 4 week, or 40%, or whatever....

I think even the kitties might get enthused about that.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 901552 · Report as offensive
Profile elbea64

Send message
Joined: 16 Aug 99
Posts: 114
Credit: 6,352,198
RAC: 0
Germany
Message 901556 - Posted: 30 May 2009, 16:18:51 UTC - in response to Message 901552.  
Last modified: 30 May 2009, 16:20:37 UTC

Credit is based on the stock apps and is, at least tried, to be the same over all projects. To change credits for promotion reasons would break the complete system.

The only way to equalize the credit between AP and MB would be to more optimize the AP stock app, which would mean to reduce the credit boost from the optimized AP app as it doesn't do the work that much faster as the stock app anymore.
ID: 901556 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 901562 - Posted: 30 May 2009, 16:24:10 UTC - in response to Message 901556.  

Credit is based on the stock apps and is, at least tried, to be the same over all projects. To change credits for promotion reasons would break the complete system.

The only way to equalize the credit between AP and MB would be to more optimize the AP stock app, which would mean to reduce the credit boost from the optimized AP app as it doesn't do the work that much faster as the stock app anymore.

I was merely thinking out loud about a way for the project to boost the amount of MB work being done vs AP....
Most projects do not have this conundrum.......they only have one app, be it stock or optimized.
They do not have the users split between what version of the work they wish to do.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 901562 · Report as offensive
Chelski
Avatar

Send message
Joined: 3 Jan 00
Posts: 121
Credit: 8,979,050
RAC: 0
Malaysia
Message 901568 - Posted: 30 May 2009, 16:33:36 UTC
Last modified: 30 May 2009, 16:36:41 UTC

I think the last time we had a discussion like this the message was that Seti, being the benchmark project for BOINC credits, will be ill advised to use credit to as a reason to get users to run one project or rather one science application rather than the other.

Unless they should remove the theoretical link (or in actuality, the rather fictional link) between credits and an total amount of computation done.

Alternative will be a free market model. Something like give work to those who are willing to do it best - therefore you bid a certain amount of credit for the work done and if you bid low enough you'll get it. But of course it will not work.

Edit: Perhaps the best thing to do is to wait for another "Reboot" of AP and hopefully things will be closer to parity.
ID: 901568 · Report as offensive
Andy Williams
Volunteer tester
Avatar

Send message
Joined: 11 May 01
Posts: 187
Credit: 112,464,820
RAC: 0
United States
Message 901570 - Posted: 30 May 2009, 16:36:36 UTC - in response to Message 901556.  

Credit is based on the stock apps and is, at least tried, to be the same over all projects. To change credits for promotion reasons would break the complete system.

The only way to equalize the credit between AP and MB would be to more optimize the AP stock app, which would mean to reduce the credit boost from the optimized AP app as it doesn't do the work that much faster as the stock app anymore.


Whatever the initial intent, the credit system across projects now is no longer the same. There are projects that grant far more credit (and some that grant less) per flop compared to SETI. There is a list in one of the threads floating around the board.

Optimizing the stock app further has limits. The stock app is designed to run on anything, be it AMD, Intel, Pentium II or i7. The optimized apps take advantage of instruction sets that the stock app cannot.
--
Classic 82353 WU / 400979 h
ID: 901570 · Report as offensive
Profile Clyde C. Phillips, III

Send message
Joined: 2 Aug 00
Posts: 1851
Credit: 5,955,047
RAC: 0
United States
Message 901576 - Posted: 30 May 2009, 16:42:50 UTC

I've noticed a hit to my Multibeam credits the past couple months or so and I don't think it's all due to doin' Einstein when MB credits are unavailable. I've been awfully busy doin' other projects lately and don't have the time to fool around with trying to install Astropulse applications, modifying their appinfo files, running into trouble, etc. I'd probably do AP provided it were just two or three mouse clicks away.
ID: 901576 · Report as offensive
Profile Labbie
Avatar

Send message
Joined: 19 Jun 06
Posts: 4083
Credit: 5,930,102
RAC: 0
United States
Message 901584 - Posted: 30 May 2009, 16:57:35 UTC - in response to Message 901576.  

I've noticed a hit to my Multibeam credits the past couple months or so and I don't think it's all due to doin' Einstein when MB credits are unavailable. I've been awfully busy doin' other projects lately and don't have the time to fool around with trying to install Astropulse applications, modifying their appinfo files, running into trouble, etc. I'd probably do AP provided it were just two or three mouse clicks away.


Try the unified installer, it works great.



Calm Chaos Forum...Join Calm Chaos Now
ID: 901584 · Report as offensive
archae86

Send message
Joined: 31 Aug 99
Posts: 909
Credit: 1,582,816
RAC: 0
United States
Message 901589 - Posted: 30 May 2009, 17:05:48 UTC

There are imbalances at the moment, to be sure, but the clear, massive imbalance is that optimized AP gives enormously excessive credit per unit of CPU effort, whether compared to optimized SETI MB, project standard AP, project standard MB, standard (and currently only) Einstein, or any other BOINC project that is even close to honoring cross-project parity.

So the good thing, I think, would be for a project release AP to be issued with most of the important improvements of the current optimized AP, with AP credit appropriately reduced to be in reasonable balance.
ID: 901589 · Report as offensive
OzzFan Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Apr 02
Posts: 15691
Credit: 84,761,841
RAC: 28
United States
Message 901590 - Posted: 30 May 2009, 17:05:57 UTC - in response to Message 901552.  
Last modified: 30 May 2009, 17:13:12 UTC

Neither I, nor the project, nor the optimizers, nor the community as a whole, would have advanced to this point if it were not for the FUN of credits........

That is the entire point of the credit game.....to give users something fun to do along with the science.


While we're at it: we all know that American football is a great sport, but if it were not for the FUN of scoring touchdowns, the players would not have any incentive to play the game...

So, in light of that, I propose we increase the score applied when a players scores a touchdown so that everyone on the other team tries harder just to keep up (after all, it will be the same if they score a touchdown).


Does that adequately portray the fallacy of the logic? I mean, don't we all understand why increases are not made across the board for all aspects of life just to make things/people happier all around? Don't we understand that a great amount of effort is made to make things fair and balanced everywhere?

Do you think for a moment that if I could not engage in the battle for supremacy (LOL) in the credit wars that I would have gone to the extent that I have simply to better the science??

I am a believer, but I am not THAT pure...LOL again.


It would seem that through all of my previous arguing, that I have been labeled as some sort of "purist" who despises credit hounds, but that's not entirely accurate.

I simply believe that raising credits to encourage crunching is the wrong answer, just like I believe that increasing one's wages is an improper way to give people more money to bolster the economy, and it consistently amazes me that people keep visiting this topic and trying to convince everyone that higher pay is the way to go.

Everyone has a motivated self-interest in increasing credits so that our scores climb even higher. I enjoy collecting credit just as much as the next person, but then I have to ask: "where does it stop? When we become dissatisfied with our progress we ask for another increase?" Sure, credits are meaningless, so we can just keep giving them away, right? It doesn't really matter after all, right? You want more participants? Increase your pay and people will have more fun.

I know I'd be happier with a bigger paycheck every two weeks, and I'd definitely have more fun with all the extra money, but where does this all lead to? Happiness only lasts so long before we need to look for another "boost" to get the same high.

If this Boinc/Seti thingy were simply a cold, hard science project, I think participation would be a fraction of what it is today.


Perhaps, but the playing field should be kept level across the board. Inflating the scores so that people are happy is not the answer no matter which way you slice it.

How about a temporary boost in MB credits....
Offer the users a 2 week 30% increase in MB credit...
Or 4 week, or 40%, or whatever....

I think even the kitties might get enthused about that.


Sure, that may garner some interest, but then what do those inflated credits mean compared to all the other credits one has earned? You could argue that credits are already uneven due to previous adjustments, but at least in the long run the credits are roughly comparable.
ID: 901590 · Report as offensive
OzzFan Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Apr 02
Posts: 15691
Credit: 84,761,841
RAC: 28
United States
Message 901591 - Posted: 30 May 2009, 17:06:49 UTC - in response to Message 901589.  

So the good thing, I think, would be for a project release AP to be issued with most of the important improvements of the current optimized AP, with AP credit appropriately reduced to be in reasonable balance.


I would support this idea 100%.
ID: 901591 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 901594 - Posted: 30 May 2009, 17:16:44 UTC - in response to Message 901589.  

There are imbalances at the moment, to be sure, but the clear, massive imbalance is that optimized AP gives enormously excessive credit per unit of CPU effort, whether compared to optimized SETI MB, project standard AP, project standard MB, standard (and currently only) Einstein, or any other BOINC project that is even close to honoring cross-project parity.

So the good thing, I think, would be for a project release AP to be issued with most of the important improvements of the current optimized AP, with AP credit appropriately reduced to be in reasonable balance.

Just keep in mind that opti should be kept out of the equation...........

STOCK issue MB vs STOCK issue AP......

Opti only increases a computer's ability to process work, not the amount of credit it receives per unit of work.

If some of the opti improvements can be incorporated into the stock app across all platforms, that is the win-win for both the project and the optimizers.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 901594 · Report as offensive
Profile elbea64

Send message
Joined: 16 Aug 99
Posts: 114
Credit: 6,352,198
RAC: 0
Germany
Message 901604 - Posted: 30 May 2009, 17:46:07 UTC - in response to Message 901594.  

Just keep in mind that opti should be kept out of the equation..........

But exactly the opt apps are "the problem", you get less credit for MB than for AP only because of opt apps and that's the reason why AP is so popular. I don't exactly know, but believe that you get equal credit with stock apps for MB and AP. So the only reason for your initial request are the opt apps that give unequal credit and therefore MB needs some boost.

I suggest to integrate MB and AP apps to solve "the problem"
ID: 901604 · Report as offensive
OzzFan Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Apr 02
Posts: 15691
Credit: 84,761,841
RAC: 28
United States
Message 901608 - Posted: 30 May 2009, 17:54:58 UTC - in response to Message 901604.  
Last modified: 30 May 2009, 18:11:12 UTC

I don't exactly know, but believe that you get equal credit with stock apps for MB and AP.


Not true. AP has a higher credit multiplier than MB. The application no longer controls the credit multiplier; the project does through its validation process, so it remains true regardless of optimized or stock. Otherwise, we'd consistently see every optimized app claiming more than stock apps, but we do not because they claim the same. The project is actually paying more for AP than MB because of the length of time it takes to crunch an AP with the stock app.

The only advantage to running optimized apps is more credit per hour, which is reflected in your RAC. The problem is not with the optimized apps since they are not overclaiming, the problem is that the credits are not equal between the two, and the disparity is more apparent when looking at one's RAC when comparing between optimized and stock.

Therefore the answer is to balance things out, and since they are supposed to be conforming to the standard "Cobblestone", they should be ajdusted toward that standard (in this case, downward). However, people with self-motivated interests always want to see more rather than less, so they always try to convince everyone that more is needed get garner more interest and participation (and it would, but at the detriment of fairness).
ID: 901608 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 901615 - Posted: 30 May 2009, 18:13:15 UTC - in response to Message 901608.  


Therefore the answer is to balance things out, and since they are supposed to be conforming to the standard "Cobblestone", they should be ajdusted toward that standard (in this case, downward). However, people with self-motivated interests always want to see more rather than less, so they always try to convince everyone that more is needed get garner more interest and participation (and it would, but at the detriment of fairness).


Again......it is NOT the opti apps that create the difference.......any more than faster rigs create the difference.

The 'go faster' apps do not modify the credits per unit of work done. Any more than the Cuda apps do.......

And yes, I and the kitties do have a self-motivated interest, as voiced in my first post in this thread.

"It's the science that got us here........It's the credits that keep us going....faster and faster and faster..."
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 901615 · Report as offensive
Profile elbea64

Send message
Joined: 16 Aug 99
Posts: 114
Credit: 6,352,198
RAC: 0
Germany
Message 901618 - Posted: 30 May 2009, 18:17:13 UTC

so the stock AP RAC is higher than stock MB RAC on the same PC?

in this case MB credit has to be increased or AP lowered
ID: 901618 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 901619 - Posted: 30 May 2009, 18:18:40 UTC - in response to Message 901618.  

so the stock AP RAC is higher than stock MB RAC on the same PC?

in this case MB credit has to be increased or AP lowered

Dunno......have been running opti for years now.
What the score is on stock is not my kibble.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 901619 · Report as offensive
Profile elbea64

Send message
Joined: 16 Aug 99
Posts: 114
Credit: 6,352,198
RAC: 0
Germany
Message 901626 - Posted: 30 May 2009, 18:33:34 UTC - in response to Message 901619.  

LOL, like me, couldn't remember doing the stock apps, at least i've never done AP stock.

But the stock apps are the base for calculating credit and modification to credit could/should only be done to equalize credits(RAC) between stock apps. Any difference in credits(RAC) between opt apps is not relevant as credit is applied to work done and not how fast it is done. And how much work is done with one WU is defined by stock apps on a defined system.
ID: 901626 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 901630 - Posted: 30 May 2009, 18:40:31 UTC - in response to Message 901626.  

LOL, like me, couldn't remember doing the stock apps, at least i've never done AP stock.

But the stock apps are the base for calculating credit and modification to credit could/should only be done to equalize credits(RAC) between stock apps. Any difference in credits(RAC) between opt apps is not relevant as credit is applied to work done and not how fast it is done. And how much work is done with one WU is defined by stock apps on a defined system.

I guess us little, vocal, peons in the system have little impact.

So I suppose our whining here does little.........but I think Eric does read the forums.......


I am OK for now.......the kibble bowls have bean filled......
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 901630 · Report as offensive
Profile perryjay
Volunteer tester
Avatar

Send message
Joined: 20 Aug 02
Posts: 3377
Credit: 20,676,751
RAC: 0
United States
Message 901632 - Posted: 30 May 2009, 18:41:29 UTC - in response to Message 901626.  

Guys, I believe Cuda is the problem child causing all the confusion. Optimized apps, whether AP or MB claim the same as stock. The only difference is the optimized apps do the work faster so you do more work units. Cuda on the other hand has the problem with over claiming. I think the problem is being worked on but they can't quite get a fix on why it does it.


PROUD MEMBER OF Team Starfire World BOINC
ID: 901632 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 901637 - Posted: 30 May 2009, 18:52:11 UTC

I have asked for this thread to be closed....

We are not coming up with any answers.......

And I do not wish for this thread to be a lightning rod.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 901637 · Report as offensive
1 · 2 · Next

Message boards : Number crunching : AP vs MB, credits and all that jazz........


 
©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.