Is there any optimized einstein app APPROVED for S5 workunits

ErichZann
ErichZann
Joined: 11 Feb 05
Posts: 120
Credit: 81582
RAC: 0

RE: The upshot of all this

Message 41059 in response to message 41056

Quote:

The upshot of all this being that if it is a serious problem, we have to start S5 again, with changes so that results from the existing apps cannot get in.

Think that wont happen.
The whole Boinc thing is open source.
So anyone can change anything to the apps, thats one reason why the results have to be validated, so anyone doing crap with his app doesnt get credits for it (and the result isnt taken serious).
But if someone changes anything and it still gives valid results thats normaly no problem, but it seems things have changed...

Crunch3r
Crunch3r
Joined: 22 Jan 05
Posts: 90
Credit: 30237616
RAC: 0

RE: RE: The upshot of

Message 41060 in response to message 41059

Quote:
Quote:

The upshot of all this being that if it is a serious problem, we have to start S5 again, with changes so that results from the existing apps cannot get in.

But if someone changes anything and it still gives valid results thats normaly no problem, but it seems things have changed...

As long as both std. app and the optimized app. produces exactly the SAME results, i don't see any reason to discard the opt. app.

Both are producing the SAME result as i can see it now and why should i use a slower app that does the same sience in 30% more time ?

Barrie
Barrie
Joined: 23 Mar 05
Posts: 219
Credit: 21449106
RAC: 0

RE: The whole Boinc thing

Message 41061 in response to message 41059

Quote:

The whole Boinc thing is open source.
So anyone can change anything to the apps, thats one reason why the results have to be validated, so anyone doing crap with his app doesnt get credits for it (and the result isnt taken serious).
But if someone changes anything and it still gives valid results thats normaly no problem, but it seems things have changed...

BOINC is open-source, but not the apps, I think. Akos had to back-engineer them. I suspect you are right, no action will be taken about the valid but unofficial results, and it really will turn out to be a storm in a teacup. Having said that, I still think that for the present, no-one has any business submitting results from unofficial apps, for reasons I stated earlier. It's like throwing eggs at someone's house and saying it isn't damage because it will wash off.

Dead men don't get the baby washed. HTH

Digger
Digger
Joined: 24 Mar 05
Posts: 84
Credit: 27421
RAC: 0

In all honesty folks, I


In all honesty folks,

I normally just don't give a rat's behind what anybody does. If people are only interested in credit, then I think that's great because a lot of science is still getting done. Everyone chooses to crunch for their own reasons. And when I see computers that are much slower than mine but are claiming twice the credit on other projects, I just don't care because the science is still getting done. I know that there are people who cheat the system for their own personal benefit, but frankly I still don't care because the science is still getting done. It's not our job to try and police our co-workers.

But if there's even just a small chance that Akos' optimized apps are messing with the science, and people continue to use them, well then I think that's kinda poopy. Those people have lost the right to claim that they have any interest in the science at all, because if they did, they would not accept even the smallest possibility that it was being trashed. And who knows... maybe it'll turn out that the stable patches really aren't hurting the science... but for right now we just don't know and that should be enough to stop using them.

I always used Crunch3r's apps over at SETI and was very grateful for the speedup. But when he asked us to stop using them, I deleted them from my system out of respect for his wishes, in fact I detached from the project altogether and moved over to Beta. He's the one who developed the apps and he should have the right to call the shots. I wonder why we are not extending the same courteousy to Akos.

It doesn't take a rocket scientist to see that Akos probably got into some hot water over releasing these unofficial patches to the public. He has recalled them now and we should respect that. Continuing to use them is only slapping him in the face even more. Bruce and the rest of the project team are highly professional, and as such, they are most likely giving Akos a chance to clean this up on his own before they have to step in. I know that's what I would do.

The guy is over a barrel here. Let's give him a break, okay?

Dig

Digger
Digger
Joined: 24 Mar 05
Posts: 84
Credit: 27421
RAC: 0

P.S. I'm sorry that post got


P.S. I'm sorry that post got so long. I didn't mean to write a novel. LOL

Desti
Desti
Joined: 20 Aug 05
Posts: 117
Credit: 23762214
RAC: 0

RE: Think that wont

Message 41064 in response to message 41061

Quote:

Think that wont happen.
The whole Boinc thing is open source.
So anyone can change anything to the apps, thats one reason why the results have to be validated, so anyone doing crap with his app doesnt get credits for it (and the result isnt taken serious).
But if someone changes anything and it still gives valid results thats normaly no problem, but it seems things have changed...

But when one WU is send to two crap apps and to one good, you got two crap results returned and they got valid while the good results become invalid, because the crap results have the majority.

The BOINC validation system is only designed to detect minoritiy faults.

Brian
Brian
Joined: 25 Mar 06
Posts: 22
Credit: 80237
RAC: 0

RE: As long as both std.

Message 41065 in response to message 41060

Quote:


As long as both std. app and the optimized app. produces exactly the SAME results, i don't see any reason to discard the opt. app.

Both are producing the SAME result as i can see it now and why should i use a slower app that does the same sience in 30% more time ?

Just because the numbers are the same doesn't mean the results are the same. If you've ever attempted to be published in a scientific journal, you'd know the QA rigors that the judges will put you through. With a major scientific discovery, you can expect the scrutiny to be many times greater. You can't attempt to publish a gravity wave discovery when some of the results were calculated by an unofficial program, regardless of the similarity of the results. It may seem stupid to you, but that's how it is. If you don't like it and think you're smart and cool for refusing to comply with the wishes of the project owners, you need to realize that you're not; you're only wasting their time.

If you love credits and nothing but credits, I guess that's your deal and I shouldn't judge. If you halfway pretend to be interested in the science, you need to get with the program.

Jord
Joined: 26 Jan 05
Posts: 2952
Credit: 5893653
RAC: 76

RE: Sorry about that one

Message 41066 in response to message 41057

Quote:

Sorry about that one Ageless,

It was just funny to see reading the S@H board after weeks and seeing you disapear and H. poping up out of the dust right where you left ...


I was having a long quiet nap until someone took that tag away from me. :-)
It was more interfering than helping.

And what I said at the time to you on the forums, was without thinking about that tag on my shoulder. ;-)
My late apologies. Well, as a(n) (ex) mod then. :)

Quote:
Just courious that no one wondered that one guy with less than 1k credit is now a mod over there...


last I heard it's someone with a lot more to his regular name (and farm), but since the forums do not allow anyone to post with zero credits, he had to crunch.

ErichZann
ErichZann
Joined: 11 Feb 05
Posts: 120
Credit: 81582
RAC: 0

RE: RE: As long as both

Message 41067 in response to message 41065

Quote:
Quote:


As long as both std. app and the optimized app. produces exactly the SAME results, i don't see any reason to discard the opt. app.

Both are producing the SAME result as i can see it now and why should i use a slower app that does the same sience in 30% more time ?

Just because the numbers are the same doesn't mean the results are the same. If you've ever attempted to be published in a scientific journal, you'd know the QA rigors that the judges will put you through. With a major scientific discovery, you can expect the scrutiny to be many times greater. You can't attempt to publish a gravity wave discovery when some of the results were calculated by an unofficial program, regardless of the similarity of the results. It may seem stupid to you, but that's how it is. If you don't like it and think you're smart and cool for refusing to comply with the wishes of the project owners, you need to realize that you're not; you're only wasting their time.

If you love credits and nothing but credits, I guess that's your deal and I shouldn't judge. If you halfway pretend to be interested in the science, you need to get with the program.

So... why dont they finaly start to check that program and think about making it official? I think the time that could be saved there is much much much bigger than the time an intensive test of that version would need. But they dont even say anything official to it...

Stick
Stick
Joined: 24 Feb 05
Posts: 790
Credit: 33130322
RAC: 1161

RE: Well again, the only

Message 41068 in response to message 41058

Quote:

Well again, the only issue i see there is the "bad result" send back by a minor of the patched apps. and that's the issue.

resending them shouldn't be an issue to the project, although drecreasing the quorum to 2 results might have raised that issue in the first place ...

Going back to 3 valid results would solve those issues.

I think Crunch3r is right about this. Going to the "2 result" quorum was the "developer's" way of optimizing. Obviously using a "3 result" quorum would take the project about 6 months longer to finish. Unfortunately, they didn't anticipate the potential problem of "two bad results" matching up - although they certainly should have, given the wide usage of Akos' apps on S4. So now it's back to the drawing boards. Hopefully, something good will come out of this. By that I mean: I hope a more formal process can be developed that will allow work like Akos' to be incorporated in future projects. BOINC projects need to encourage such contributions - not stifle them.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.