I think I'm being misunderstood. I care a lot more about the science than the rankings. It's just that my first machine, and the one I've been running on for the past 4 years, was a laptop with an Intel Celeron 900 2.2 GHz single-core processor and integrated graphics. Having a decent rank, even in the top 50%, is significant to me because it's like making a very high score on a benchmark. I'm still having a hard time believing I have a machine this powerful after watching people build ones like it for so long. My motivation for running GIMPS is purely scientific. I want to be an electrical engineer. But if those rankings give me a little extra motivation, I'm not going to feel shitty about myself as long as I keep them in perspective, which I do. If people want to make it a bit of a fun competition that spurs scientific research, that's great. Isn't that how the commercial science industry works anyways? I know GIMPS isn't commercial, but it still works for the industry, and I think it could work to come extent on the smaller scale.

[QUOTE=Red Raven;384480]...GIMPS isn't commercial, but it still works for the industry, ...[/QUOTE]
No, it doesn't.
It is purely for entertainment. (And for promotion of scientific education, as a byproduct.)

There is nothing wrong with entertainment. It doesn't help anyone to have false reasons for simply having fun.

[QUOTE=Batalov;384486]No, it doesn't.
It is purely for entertainment. (And for promotion of scientific education, as a byproduct.)

There is nothing wrong with entertainment. It doesn't help anyone to have false reasons for simply having fun.[/QUOTE]

I meant that competition works for industry, and it can work to a lesser extent for GIMPS. The comparison is only true on a shallow level though.

[QUOTE=Red Raven;384488]I meant that competition works for industry, and it can work to a lesser extent for GIMPS. The comparison is only true on a shallow level though.[/QUOTE]

Your meaning was quite clear.

Heck, what are the XPrizes all about?

(Ummm... Competition?)

[QUOTE=Red Raven;384488]I meant that competition works for industry, and it can work to a [B]lesser[/B] extent for GIMPS. The comparison is only true on a shallow level though.[/QUOTE]
I agree with all three of these statements.
(I've read too fast and picked the wrong antecedent. Sorry! you are right.)

Competition is natural to most humans, and in some cases if the virtual "employees" are willing to compete among themselves on who is doing more work* or faster than others and take pride solely in that, then the "employer" wins (if all measures are taken against cheating; if not, then this is where pressure/competition hurts science/industry; it's not hard to find and read about the horrors of [I]some[/I] scientific sweatshops**). "Gamification" is all about the same, too; probably, with even more fun. The job gets done, the science arguably benefits (even if a little) from the results (the participants may learn no science but get a warm and fuzzy feeling that they've "won", or "earned some credits", etc).

It is important that if this is not all just a game of credits to [I]even a fraction[/I] of the participants (and they go on and learn something about what they are participating in), then GIMPS' true role is achieved.

__________
[SIZE="1"]*they don't even have to understand what that [I]work[/I] is.

**and even more horrors in industry, e.g. in pharma; think clinical trials... Dallas Buyers Club, The Constant Gardener, that sort of thing...[/SIZE]

[QUOTE=Red Raven;384445] 1538 GHz Days (can someone explain that metric please?)[/QUOTE]

I'm glad you're having fun!

This metric is quite old but we still use it. Specifically, it is "Intel Core 2" GHz-Days, and a GHz-Day is what one core of an Intel Core 2 processor running at 1 GHz could produce in one day.

So, if you're running a Core 2 Quad @ 2 GHz for 3 days, you get 4 cores x 2 GHz x 3 Days = 24 GHz-Days.


Every job is standardized to the amount of time a Core 2 processor would take to do the job. You get that amount of credit regardless of the actual time it took you; [I]one core[/I] of my i5-3570k running at 4.5 GHz produces roughly 12 GHz-Days, if memory serves.

This is because the architecture of the newer CPUs is such that everything is run much more efficiently even on a per-clock basis. Think of a 30 year old car against a new car; same RPM, but a LOT more power (and less fuel consumed, to boot).


The metric is very unfairly biased toward GPU Trial Factoring as you have likely witnessed. The architecture of a GPU is such that Trial Factoring is ridiculously fast; my GPUs both crank through 260-290 GHz (that's 100 times more than one core of a Core 2 processor @ 2.6 GHz).

There has been talk of having a separate metric for TF because it skews overall competition, but as was pointed out by Silverman, nobody really cares. I just use it as a benchmark, too.

Hm.. A new guy is coming, fresh meat. He has resources (computing power), he wants to learn the routine, he doesn't care to pay for his own electricity, and most of all he is enthusiastic (as we all were, in some point of the near and/or far past). What do you do? Hit him in the face... Shame on you!

Bad management. Bad!

Hey Raven, I am here for credit, and I am not ashamed to say it. Don't look to RDS, except when he is talking math, when you have to listen, because he is really good to it, the rest of the time you can ignore him :razz:. If you want to compete, let's compete, you make a [URL="http://www.gpu72.com/reports/workers/"]gpu72[/URL] account, install [URL="http://www.mersenneforum.org/misfit/"]misfit[/URL], select "let gpu72 decide" in misfit's work scheduler, and you are ready to go. Let's see who finds the most factors in the following month.

(This is the spirit! :razz: Why doesn't god give us 5000 guys, all with 295x2 and willing to compete for the credit? Hehe...)

[QUOTE=LaurV;384529]Hm.. A new guy is coming, fresh meat. He has resources (computing power), he wants to learn the routine, he doesn't care to pay for his own electricity, and most of all he is enthusiastic (as we all were, in some point of the near and/or far past). What do you do? Hit him in the face... Shame on you!

Bad management. Bad!

Hey Raven, I am here for credit, and I am not ashamed to say it. Don't look to RDS, except when he is talking math, when you have to listen, because he is really good to it, the rest of the time you can ignore him :razz:. If you want to compete, let's compete, you make a [URL="http://www.gpu72.com/reports/workers/"]gpu72[/URL] account, install [URL="http://www.mersenneforum.org/misfit/"]misfit[/URL], select "let gpu72 decide" in misfit's work scheduler, and you are ready to go. Let's see who finds the most factors in the following month.

(This is the spirit! :razz: Why doesn't god give us 5000 guys, all with 295x2 and willing to compete for the credit? Hehe...)[/QUOTE]
Sounds like fun! It's even more interesting because it's by factors found, not numbers factored. I downloaded the GPU to 72 version of MISFIT and I have a GPU72 account. Before I try to get it running, how exactly does the primenet>GPU72>MISFIT>mfakto chain work, and does the results reporting chain work in the exact opposite order?

"My motivation for running GIMPS is purely scientific"
 

[QUOTE=Red Raven;384584]Sounds like fun! [/QUOTE]
Sounds like you found a match.

[QUOTE=Red Raven;384584]It's even more interesting because it's by factors found, not numbers factored. [/QUOTE]
Say what? Next thing, you know, you will start searching for additional factors for the candidates with already known factors (just to get your "score" higher in this artificial competition); not that it's something new - you can join the club of people already running this. There are people who search for more factors than necessary -- for a tiny chance to have the input number 'completely factored' (with a disclaimer) - but this satellite project is not GIMPS . When is doubt, search the forum; don't expect that all you need to read is in this thread. "Learn, learn, and learn", to quote from one nowadays unpopular "classic".

Did you say that your motivation was purely scientific, or someone twisted your arm to say this, hmm? If science is on your mind, listen: GIMPS is only interested in one factor per candidate (because it eliminates the candidate from the pool of potential primes); for GIMPS, there is no difference "by factors found, not numbers factored". But of course, between the two of you, you can compete on anything your heart would desire, maybe the number of found coca-cola caps or something even more exciting. :razz:

Anyway, when either of you will find the first 2 million factors, wake me up, I will maybe get interested in this little competition.
[SPOILER]Yes, I've already found 2,056,000+ factors. It's all downhill from there, as they say. Life lost all its meaning... not.[/SPOILER]

[QUOTE=Red Raven;384584]Before I try to get it running, how exactly does the primenet>GPU72>MISFIT>mfakto chain work, and does the results reporting chain work in the exact opposite order?[/QUOTE]
Were these questions all answered in the post #2 or were they not?
Why is LaurV's post suddenly such a revelation? :-)
_________________

P.S. Final remarks:
1. Let me compare GIMPS to a charity that delivers Christmas gifts. And you are a volunteer with a car (or a truck, or maybe a generous taxicab company owner who gives a command to his fleet to run some deliveries in free time or along the way) and you work part time for this charity. The volunteering help is useful, [B]no matter how little[/B]. Of course, you are free to take any route to deliver a package, maybe all the way around town (maybe you want to enjoy the sunset on your way). You can even help when all you own is a bicycle, not a car. Of course you are free to stop for coffee, or even take a week off. It's a charity! And a kind of a charity that doesn't have "projections for the 3rd quarter of the year, and when they are not met, the corporation's stock crumbles on NYSE". No! You are free to deliver packages while standing on your head if you can (and like it). It is only up to you if you care to help GIMPS more by doing what is really wanted (the P95 program and GPU72+bridges will fully take care of what is currently wanted), or something partially wanted but what you like (for whatever reason), or something totally useless but what you like even more (but then it has nothing to do with GIMPS, you have to realize that).
2. I have no role in the "Bad management" of GIMPS. That "Bad management" only exists in your head, LaurV. Even George's yearly bonus doesn't depend on how many factors were found or that /god damn it/ no prime was found this year again (no prime, no bonus! ...wait, what bonus?). ;-)

[QUOTE=Batalov;384593]

Wrote some stuff[/QUOTE]

Hey Red Raven,

you can safely ignore anything the thinkpol writes...as you can RDS. I'm certainly here for the credits and have been for the last 18 years..

[QUOTE=Batalov;384593]Sounds like you found a match.


Say what? Next thing, you know, you will start searching for additional factors for the candidates with already known factors (just to get your "score" higher in this artificial competition); not that it's something new - you can join the club of people already running this. There are people who search for more factors than necessary -- for a tiny chance to have the input number 'completely factored' (with a disclaimer) - but this satellite project is not GIMPS . When is doubt, search the forum; don't expect that all you need to read is in this thread. "Learn, learn, and learn", to quote from one nowadays unpopular "classic".

Did you say that your motivation was purely scientific, or someone twisted your arm to say this, hmm? If science is on your mind, listen: GIMPS is only interested in one factor per candidate (because it eliminates the candidate from the pool of potential primes); for GIMPS, there is no difference "by factors found, not numbers factored". But of course, between the two of you, you can compete on anything your heart would desire, maybe the number of found coca-cola caps or something even more exciting. :razz:

Anyway, when either of you will find the first 2 million factors, wake me up, I will maybe get interested in this little competition.
[SPOILER]Yes, I've already found 2,056,000+ factors. It's all downhill from there, as they say. Life lost all its meaning... not.[/SPOILER]


Were these questions all answered in the post #2 or were they not?
Why is LaurV's post suddenly such a revelation? :-)
_________________

P.S. Final remarks:
1. Let me compare GIMPS to a charity that delivers Christmas gifts. And you are a volunteer with a car (or a truck, or maybe a generous taxicab company owner who gives a command to his fleet to run some deliveries in free time or along the way) and you work part time for this charity. The volunteering help is useful, [B]no matter how little[/B]. Of course, you are free to take any route to deliver a package, maybe all the way around town (maybe you want to enjoy the sunset on your way). You can even help when all you own is a bicycle, not a car. Of course you are free to stop for coffee, or even take a week off. It's a charity! And a kind of a charity that doesn't have "projections for the 3rd quarter of the year, and when they are not met, the corporation's stock crumbles on NYSE". No! You are free to deliver packages while standing on your head if you can (and like it). It is only up to you if you care to help GIMPS more by doing what is really wanted (the P95 program and GPU72+bridges will fully take care of what is currently wanted), or something partially wanted but what you like (for whatever reason), or something totally useless but what you like even more (but then it has nothing to do with GIMPS, you have to realize that).
2. I have no role in the "Bad management" of GIMPS. That "Bad management" only exists in your head, LaurV. Even George's yearly bonus doesn't depend on how many factors were found or that /god damn it/ no prime was found this year again (no prime, no bonus! ...wait, what bonus?). ;-)[/QUOTE]

I only plan on doing work that helps GIMPS find Mersenne primes. However, I am curious as to why there isn't a database that contains every known prime ever discovered period. For completeness' sake, it should have lists of every factor of every number found by GIMPS and other projects. In other words, is there a database that contains all known data on primes and factors? Also, does GIMPS use data from other projects to check numbers off of the list? And no, no one twisted my arm. I know that GIMPS has a limited scope, but it's still science, it seems like one of the easier projects to get involved in, and I think the results of your personal efforts in GIMPS are more apparent than in other projects.