We're Here To Crush The Strike

---

You don't care because you don't have a RAC worth talking about.

Outrageous pomposity! Does higher RAC make you a better person in any way? Shame!

Am I the only one who feels this deserves redress??? In the recent past such crass comments as his would have been blown out of the water by dozens. Step up people!

Did I say this ? I never said nor claimed that those with lower daily credits are less valuable to the project. I said "You don't care because you don't have a RAC worth talking about" does this mean that you are less important - NO, does this mean that I am more important, NO - what it means is simply - I CARE ABOUT MY CREDITs, those with lower credits may not.

Strange double talk here. Your statement: "You don't care because you don't have a RAC worth talking about." You said that this person doesn't care because they have a low RAC. Those were your very own words. Then you try to reinforce/redirect this statement by saying that your high RAC reflects your care for the project substantiated by your financial investments in gear and donations. Immediately after that you state that a low RAC contributer may not care. Sorry, dude, but you're busted. Better to apologise now.

Any talk of the thousands you've invested in crunching machines does nothing to mitigate your implication that high RAC equals care for the project and low RAC equals less or no care (than you) for the project.

The very foundation of this project is that anyone with a few spare cycles can make a worthy contribution to this effort. If a person has only a single, old, computer, they contribute with as much validity as a person who has many cutting edge rigs. That person with a single aged computer can, and usually does, care as much or more than you do about their contribution.

I'm reminded of the old story about the rich man who drops a bag of gold in the church donation box. Later, a poor man drops in a single penny - all he has. The rich man sees himself as the more worthy contributer. He is soon shamed since his bag of gold is a small portion of his wealth and the poor man's single penny is a proprtionately huge amount to give away. Which man cared more?

Or are you saying that if a person cared as much as you do for the project, they'd build a bunch of rigs as you have? Is the amount of financial investment your measure of caring?

Like I said, you're busted, dude.

Amen, I agree totally.
I am one of the people he is talking about. I am fairly new here but only because recently is the first time that I have had an internet connection since this project started. I am retired and living off of a disability check so I don't have the money to invest in even one "decent" crunching computer. However I do care about this project. Sure I may not have a high rac or many credits here. That is because I do care about this project and have been doing most of my crunching in the beta project to try to help get this enhanced application ready to release. And this person's assumption that if someone cares about the project they would spend thousands of dollars on crunching machines is absurd. *I can't afford even one expensive computer!* If I could afford it I would, but I'm not about to take food out of my or my wife's mouth to feed this project, no matter how much I care about it. I'm lucky to be able to afford the internet connection. And as far as computers I'm using an old p3 500mhz computer which has been put together mostly from donated parts.
However I do care about this project just as much as anyone and I resent the implication that since I don't have an expensive crunching farm that I don't care.

---

Jim

Some people plan their life out and look back at the wealth they've had.
Others live life day by day and look back at the wealth of experiences and enjoyment they've had.

We're Here To Crush The Strike

Ha, Ha. Museum pieces, It's a real screa'em. ;)

---

Savoir-Faire is everywhere!
The T1 Trust, T1 Class 4-4-4-4 #5550, America's First HST

Amen, I agree totally.
I am one of the people he is talking about. I am fairly new here but only because recently is the first time that I have had an internet connection since this project started. I am retired and living off of a disability check so I don't have the money to invest in even one "decent" crunching computer. However I do care about this project. Sure I may not have a high rac or many credits here. That is because I do care about this project and have been doing most of my crunching in the beta project to try to help get this enhanced application ready to release. And this person's assumption that if someone cares about the project they would spend thousands of dollars on crunching machines is absurd. *I can't afford even one expensive computer!* If I could afford it I would, but I'm not about to take food out of my or my wife's mouth to feed this project, no matter how much I care about it. I'm lucky to be able to afford the internet connection. And as far as computers I'm using an old p3 500mhz computer which has been put together mostly from donated parts.
However I do care about this project just as much as anyone and I resent the implication that since I don't have an expensive crunching farm that I don't care.


I TOTALY agree with you Jim-R, I'm in it for the project as well and to see the crap being said on the Forums or where ever you look makes me sick (SO MUCH FOR SCIENCE)

Stargate

---

Amen, I agree totally.
I am one of the people he is talking about. I am fairly new here but only because recently is the first time that I have had an internet connection since this project started. I am retired and living off of a disability check so I don't have the money to invest in even one "decent" crunching computer. However I do care about this project. Sure I may not have a high rac or many credits here. That is because I do care about this project and have been doing most of my crunching in the beta project to try to help get this enhanced application ready to release. And this person's assumption that if someone cares about the project they would spend thousands of dollars on crunching machines is absurd. *I can't afford even one expensive computer!* If I could afford it I would, but I'm not about to take food out of my or my wife's mouth to feed this project, no matter how much I care about it. I'm lucky to be able to afford the internet connection. And as far as computers I'm using an old p3 500mhz computer which has been put together mostly from donated parts.
However I do care about this project just as much as anyone and I resent the implication that since I don't have an expensive crunching farm that I don't care.


I TOTALY agree with you Jim-R, I'm in it for the project as well and to see the crap being said on the Forums or where ever you look makes me sick (SO MUCH FOR SCIENCE)

Stargate

No spending money on PCs is not that important, I spend money when I can only afford It and never more than I can afford, As I get SSI. Although due to an inheritance from My late Aunts Trust fund that She and Her 2nd late Husband set up, I'll be able to afford some computer expansion in October or November, More likely November 2006. And more importantly than that I'll be able to buy some land and setup a house free of any mortgage and all for about $66,000 too(The house and land that is). :D

---

Savoir-Faire is everywhere!
The T1 Trust, T1 Class 4-4-4-4 #5550, America's First HST

You don't care because you don't have a RAC worth talking about.

Outrageous pomposity! Does higher RAC make you a better person in any way? Shame!

Am I the only one who feels this deserves redress??? In the recent past such crass comments as his would have been blown out of the water by dozens. Step up people!

Did I say this ? I never said nor claimed that those with lower daily credits are less valuable to the project. I said "You don't care because you don't have a RAC worth talking about" does this mean that you are less important - NO, does this mean that I am more important, NO - what it means is simply - I CARE ABOUT MY CREDITs, those with lower credits may not.

Strange double talk here. Your statement: "You don't care because you don't have a RAC worth talking about." You said that this person doesn't care because they have a low RAC. Those were your very own words. Then you try to reinforce/redirect this statement by saying that your high RAC reflects your care for the project substantiated by your financial investments in gear and donations. Immediately after that you state that a low RAC contributer may not care. Sorry, dude, but you're busted. Better to apologise now.

Any talk of the thousands you've invested in crunching machines does nothing to mitigate your implication that high RAC equals care for the project and low RAC equals less or no care (than you) for the project.

The very foundation of this project is that anyone with a few spare cycles can make a worthy contribution to this effort. If a person has only a single, old, computer, they contribute with as much validity as a person who has many cutting edge rigs. That person with a single aged computer can, and usually does, care as much or more than you do about their contribution.

I'm reminded of the old story about the rich man who drops a bag of gold in the church donation box. Later, a poor man drops in a single penny - all he has. The rich man sees himself as the more worthy contributer. He is soon shamed since his bag of gold is a small portion of his wealth and the poor man's single penny is a proprtionately huge amount to give away. Which man cared more?

Or are you saying that if a person cared as much as you do for the project, they'd build a bunch of rigs as you have? Is the amount of financial investment your measure of caring?

Like I said, you're busted, dude.

He never said they did not care about the project. Where do you get off changing and adding words to his statement?? People with a low RAC do not care about the concerns of those who have a much higher RAC, They have said so on many occassions. That is the "care" he was talking about. They think we are "credit mongers" and can care less about what we think or what our issues are. So now who is busted?? Quit twisting the words to suit your agenda bub!!

---

Any more of this "you said" ... "no, you said" 'communication' and this whole thing will never go away.

I ask of everyone to stop pointing fingers at the other and to communicate in a decent way. So far the moderation has not moderated much in threads like these, but maybe it is time we started to.

So please, if you have something to say, try doing it within reason and let bygones be bygones. One day you'll all have to learn to live together on these forums, come what may. So why not start today?

Why don't Team USA go back to there team board, and toast the progress they made during the good times. Then after a few beers, think up some good legal, no-controversial, idea's to stay on top of whatever pile they think is worthwhile.

Andy

I will take a whack at this Eric...if Berkeley would come up with a rule that only "approved" clients are allowed to be used it might go a long way towards ending this. "Approved" can mean several things..from only Berkeley apps to people like Crunch3r, Trux, TMR, etc, sending their app to Berkeley to get it approved BEFORE they distribute it. Berkeley could then modify the code slightly when non "approved" verions of the software come out nullifing any credits they would receive.

I would hate to go there, because a large number of people use the anonymous platform mechanisms on platforms we don't support or to make their own optimized builds. If I need to sign a binary everytime someone wanted to compile SETI@home, I'd be more busy than I am. The vast majority of the compiled versions available seem to do the right thing. The rest, we rely on validation and the median of three results credit allocation mechanism to remove the incentive to over-report credits.

But if Berkeley were to require that no optimized app be released before being peer reviewed then it may have negated this whole situation we are now in. I agree it is more work, but maybe if you did a conference call with the optimizers and gave them some guidelines and had them peer review it among themselves BEFORE releasing it to the public, it might help. Tony seems to think that Crunch3r's app is "cheating" in the way it ask for credit, if the app had been peer reviewed he would not have a leg to stand on and be seen as nothing but a "whiner". It also would have pointed out any errors Crunch3r had that needed tweaking. I think Berkeley needs to make a decision to PUBLICLY either support or not support, the optimized apps that various people are releasing. No comment leaves the subject open to interpretation and leads us to where we are today, IMO.
I totally agree with and understand the rest of your comments so have deleted them.

---

I'm lucky to be able to afford the internet connection. And as far as computers I'm using an old p3 500mhz computer which has been put together mostly from donated parts.
However I do care about this project just as much as anyone and I resent the implication that since I don't have an expensive crunching farm that I don't care.

JimR where in NC do you live?

---

All this silly bickering... And with ManBearPig out there roaming around unchecked.

---

SETI.USA - Uniting the United States stats!

But if Berkeley were to require that no optimized app be released before being peer reviewed then it may have negated this whole situation we are now in. I agree it is more work, but maybe if you did a conference call with the optimizers and gave them some guidelines and had them peer review it among themselves BEFORE releasing it to the public, it might help.

Mikey, this sounds like something I emailed the "Boinc Alpha Mail list" about on December 2 2005: NOTE: the only answer was from Dr. A who states that this new Fpops system will address the inequities)

Boinc Alpha work, It's been suggested that since we don't really need to test out the seti app, that we might do something else. Also, I see the Optimization of both the Client and the Application reaching all new levels of Xtremes.

History
The optimizations of the core client first started to be done to try and equalize claimed credit between windows based hosts and Linux based hosts. Then optimizations of the applications were done. The optimization of the applications caused more work to be done in less time and this was good for science. However, it gave the users the feeling that they were being cheated out of claimed credit, so core client optimizations were made to raise the benchmarks and thereby try to even out the claimed credit. On the surface this is a good thing, but I don't know if all the things that are currently happening were thought of at the time.

The Problems with optimized applications
1. They claim less credit than the regular applications because of decreased cpu time.
2. When they first came out there use was limited and the decreased claims of the few who used it were weeded out by the granting credit process, but now with increased use, there is the increased chance that those not using one will be cheated out of credit ( I.E when two or more opt apps are used in granting credit).
3. The optimizations aren't regulated or checked by seti. They could be:
A. returning bad results
B. returning results that don't "strongly agree", this is even more of an issue as more and more authors produce these versions.
4. They can just be done improperly and trash work

The Problems with optimized core clients
1. They are being changed to allow user control over the quantity, and reporting of work. This removes project control of how much traffic and frequency of RPC calls for work uploading and reporting.
2. They can be twisted into a device to just plain attempt to cheat, even if only by a few tenths or even whole amounts of cobblestones per wu. (example, one user with an AMD64 3200 OCed just reported a benchmark of Measured floating point speed 3760 million ops/sec Measured integer speed 11460.0 million ops/sec, my AMD64 3700 reports 2k/4k with a stock boinc client. My app does a wu in 1 hour and unless his does one in 20 minutes then this user is cheating).
3. They were originally intended to compensate for reduce claime credit, by enhancing the benchmark for those using an optimized application. However, we have:
A. Users just using the optimized boinc to claim more without the use of an optimized app.
B. People using an optimzed boinc for other projects as well. Example
a. Rosetta doesn't use redundancy in wu and grants what a client asks, these optimized boinc users can be easily requesting and getting double what they should be getting. Dr Baker is even considering using redundancy to fight this, even though it reducing the quantity of science done by a factor of 3 or 4.
b. Other projects with redundancy will even feel the hit as more and more users adopt these clients.
4. Changes to caching work and even benchmark can cause the loss of work due to "over no reply" late reporting of work.

My thoughts
There will come a time (if it's not here now), that something will have to be done to help keep both projects and users happy (if not just satisfied)

Idea for a solution (part 1)
Have all boinc versions and applications digitally signed by berkeley or project owners. Have this signature unique and secure from hacking so only officially sanctioned versions can be used to do work for the projects. This will raise quite a ruckous if left as it stands without a reasonable process in place to allow the more creative coders out there to be able to get their versions approved and working.

I suggest we use the existing boinc alpha group to be used to test these optimized applications and if they work properly and return valid "strongly similar" results that meet the needs of the projects, then we gather stats on efficiciency and effectiveness, have them digitally signed and returned to the author to distribute as he/she sees fit.

We start a new group or maybe convert the beta group to test the Core client versions that get submitted for testing.

Ideas for alternate solutions
<please insert your alterations as needed here> or where ever you choose

thank you for your time

tony

Reply Reply to all Forward





mmciastro to boinc_alpha, boinc_dev, David
More options 12/2/05

I just had another thought, insert some form of code to not allow optimized core clients to work with more than just Seti(or other future optimized app) and label the client clearly and visibly as such.
- Show quoted text -



On 12/2/05, mmciastro <xxxxxxxx> wrote:
Ideas for alternate solutions
<please insert your alterations as needed here> or where ever you choose

thank you for your time

tony



Reply Reply to all Forward





David Anderson to me, boinc_alpha, boinc_dev
More options 12/3/05

The recently added
extern void boinc_ops_per_cpu_sec(double fp, double integer);
extern void boinc_ops_cumulative(double fp, double integer);

calls eliminate the use of the benchmark in the core client.
I believe that "SETI@home enhanced" will use boinc_ops_cumulative,
which in theory will give everyone exactly the same credit
for a given workunit.

It also would have pointed out any errors Crunch3r had that needed tweaking.

It was more the other way round. Crunch3r had a possible solution (seems to solve prob on my machines>almost never have error units) and it is i think generally the case that the optimisers help the developers with the bugs in their app on occasion.

Also i thnk your just going to alienate some optimizers with that idea of getting it vertified. They go to great lengths making sure it meets and o occasionally beats the quality standard.
Having to send it to get checked is unlikely to work because the developers are complaining theres too busy as it is and your likely to make optimizers not interested as they where at the task

It also would have pointed out any errors Crunch3r had that needed tweaking.

It was more the other way round. Crunch3r had a possible solution (seems to solve prob on my machines>almost never have error units) and it is i think generally the case that the optimisers help the developers with the bugs in their app on occasion.

Also i thnk your just going to alienate some optimizers with that idea of getting it vertified. They go to great lengths making sure it meets and o occasionally beats the quality standard.
Having to send it to get checked is unlikely to work because the developers are complaining theres too busy as it is and your likely to make optimizers not interested as they where at the task

Also theres the problem that i think added to the reason crunch3rs offer to an optimized general app was turned down. Crunch3r and some others might use a more uptodate compiler and the project would have to buy a copy of that exspensive compiler before they can really go throught it.

They would have had to rely on crunch3r to continue to develop that app. And they don't seems to like to lose any contol

It also would have pointed out any errors Crunch3r had that needed tweaking.

It was more the other way round. Crunch3r had a possible solution (seems to solve prob on my machines>almost never have error units) and it is i think generally the case that the optimisers help the developers with the bugs in their app on occasion.

Also i thnk your just going to alienate some optimizers with that idea of getting it vertified. They go to great lengths making sure it meets and o occasionally beats the quality standard.
Having to send it to get checked is unlikely to work because the developers are complaining theres too busy as it is and your likely to make optimizers not interested as they where at the task

As reported in other threads the problems with the units that didn't error out were splitter BOINC problems not Seti. And the other problem of the win9x timer has actually been there since day one, it was just not noticed as much with the old apps. This is another problem that in future is going to be solved in BOINC, which should be responsible for timimg, not the app.
Crunch3r just got lucky.
Andy

Isn't this getting a bit over complicated? Finding solutions that is?

The new credit system is suppose to and almost always does ensure almost identicle credit claims. Since the vast majority use standard app it shouuld be easy to spot odd claims.

Right now tho its made difficult because of the 'old clients' problem. When thats finaly sorted and we are all on the same flops counting system then spotting the VERY FEW odd claims would be very easy.
Maybe would have been a better idea to put that multiplier aspect (just the muultiplier) on the server side. That would also make it easy to modify the multiplier if neccisary in the future as it should affect everyone EVENLY and IMMEDIATLY.

Isn't this getting a bit over complicated? Finding solutions that is?

The new credit system is suppose to and almost always does ensure almost identicle credit claims. Since the vast majority use standard app it shouuld be easy to spot odd claims.

Right now tho its made difficult because of the 'old clients' problem. When thats finaly sorted and we are all on the same flops counting system then spotting the VERY FEW odd claims would be very easy.
Maybe would have been a better idea to put that multiplier aspect (just the muultiplier) on the server side. That would also make it easy to modify the multiplier if neccisary in the future as it should affect everyone EVENLY and IMMEDIATLY.

That maybe a good idea if the servers collected the AR and fpops count info and then calculated the credit.

Andy

As reported in other threads the problems with the units that didn't error out were splitter BOINC problems not Seti. And the other problem of the win9x timer has actually been there since day one, it was just not noticed as much with the old apps. This is another problem that in future is going to be solved in BOINC, which should be responsible for timimg, not the app.
Crunch3r just got lucky.
Andy

I was talking about the probs in the standard app that are causing some quite a bit of bother. Or maybe i'm just not understanding your post?

My main points is that if you force the optimizers to get their work validated like it is bieng sugested. WE( people who use them) make very sure to check it working correctly and doing the science well. If the app has any probs that wastes work we wouldn't use it as is evident as i say i wil NOT go back to the standard app anytime soon because it just watses too may units. 1 in evry 10-20 at best guestimate but now its prety much never. Just when the unit is bad.

Isn't this getting a bit over complicated? Finding solutions that is?

The new credit system is suppose to and almost always does ensure almost identicle credit claims. Since the vast majority use standard app it shouuld be easy to spot odd claims.

Right now tho its made difficult because of the 'old clients' problem. When thats finaly sorted and we are all on the same flops counting system then spotting the VERY FEW odd claims would be very easy.
Maybe would have been a better idea to put that multiplier aspect (just the muultiplier) on the server side. That would also make it easy to modify the multiplier if neccisary in the future as it should affect everyone EVENLY and IMMEDIATLY.

That maybe a good idea if the servers collected the AR and fpops count info and then calculated the credit.

Andy

Is this going off topic??
sorry if so but it seems like a better idea to me.

I imagine it would be harder for anyone that would try to be naughty to modify the app to give false flops and AR(?) than to just switch the multiplier which can easily get messed up by even a simple inocent typo

As reported in other threads the problems with the units that didn't error out were splitter BOINC problems not Seti. And the other problem of the win9x timer has actually been there since day one, it was just not noticed as much with the old apps. This is another problem that in future is going to be solved in BOINC, which should be responsible for timimg, not the app.
Crunch3r just got lucky.
Andy

I was talking about the probs in the standard app that are causing some quite a bit of bother. Or maybe i'm just not understanding your post?

My main points is that if you force the optimizers to get their work validated like it is bieng sugested. WE( people who use them) make very sure to check it working correctly and doing the science well. If the app has any probs that wastes work we wouldn't use it as is evident as i say i wil NOT go back to the standard app anytime soon because it just watses too may units. 1 in evry 10-20 and best geustimate but know its prety muhc never. Just when the unit is bad.

What I said previously was to do with the standard app not doing an error or timing out. It was a BOINC splitter problem, not Seti app.

This should be sorted now see Eric K's post in forum_thread.php?id=30475#328637

Andy