Your involvement in GIMPS
 

Since GIMPS isn't just about finding record-size primes, I'm a little curious as to how everyone participates in the project. So, what kind of work do you do for GIMPS, if any?

The majority of my computers are currently set to fetch work that makes the most sense. However, PrimeNet must think my computers are outdated as I haven't gotten any LL tests in a very long time. As such, I'm doing mostly P-1 and TF right now.

Some of these options are not mutually exclusive.
Also the option "I am in it for the money" is regrettably absent.

I call :poop: on Batalov's poll answer...:bs meter:

[QUOTE=ixfd64;304436]Since GIMPS isn't just about finding record-size primes, I'm a little curious as to how everyone participates in the project. So, what kind of work do you do for GIMPS, if any?

The majority of my computers are currently set to fetch work that makes the most sense. However, PrimeNet must think my computers are outdated as I haven't gotten any LL tests in a very long time. As such, I'm doing mostly P-1 and TF right now.[/QUOTE]
Since the "MPS" bit stands for "Mersenne Prime Search" I fail to see the source of your mystification. No mention of records, although this is always likely.

What TF/P-1 are you doing?
This may be useful or effectively the opposite.

I do virgin LL tests (P-1 if needed). The smaller the better, and if inadequately TFed, ask a guy with a GPU to do a few more bits.

David

[QUOTE=Batalov;304438]Some of these options are not mutually exclusive.
Also the option "I am in it for the money" is regrettably absent.[/QUOTE]

It's a little hard to deal with those considering that the forum software does not allow users to put in more than one set of questions per thread.

FYI: the "money" option would best go under "finding a new prime."

[QUOTE=Batalov;304438]
Also the option "I am in it for the money" is regrettably absent.[/QUOTE]
Albeit understandable

[QUOTE=c10ck3r;304439]I call :poop: on Batalov's poll answer...[/QUOTE]
Maybe because that is as much as you know, dude.
You probably also don't know the list of members who cleared more than 2,000,000 exponents (when it was still easy). I'll give you a hint: there was a few of us. ;-)

[SPOILER]I am quoting from an anecdote: When a certain scientist was asked for an opinion of the true modern genii in physics, he answered "Well, there's a few of us".[/SPOILER]

Here are my lifetime stats, normalized so that the lowest value = 1
TF 33699
P-1 139
LL 6665
LL-D 3569
ECM 1
ECM-F 29

[QUOTE=Uncwilly;304446]Here are my lifetime stats, normalized so that the lowest value = 1
TF 33699
P-1 139
LL 6665
LL-D 3569
ECM 1
ECM-F 29[/QUOTE]

How much of that TF is GPU?

[QUOTE=Batalov;304445]
[SPOILER]I am quoting from an anecdote: When a certain scientist was asked for an opinion of the true modern genii in physics, he answered "Well, there's a few of us".[/SPOILER][/QUOTE]
Give us a clue.
What era?

D

[QUOTE=davieddy;304454]Give us a clue.
What era?

D[/QUOTE]

Witten?

[QUOTE=Dubslow;304455]Witten?[/QUOTE]
Possible...
other options strike me as equally plausible.

Depends what you describe as [URL="http://www.youtube.com/watch?v=594WLzzb3JI"]Modern[/URL]

D

[QUOTE=Batalov;304445]Maybe because that is as much as you know, dude.
You probably also don't know the list of members who cleared more than 2,000,000 exponents (when it was still easy). I'll give you a hint: there was a few of us. ;-)

[SPOILER]I am quoting from an anecdote: When a certain scientist was asked for an opinion of the true modern genii in physics, he answered "Well, there's a few of us".[/SPOILER][/QUOTE]

The first part was what I [URL="http://mersenne.org/report_top_500_custom/?team_flag=0&type=1001&rank_lo=28&rank_hi=28&start_date=&end_date=&B1=Get+Report"]disagreed[/URL] with, Hrothgar...

[QUOTE=Dubslow;304452]How much of that TF is GPU?[/QUOTE] 0.0000% all CPU grunt work. Some of it is borg boxen.

[QUOTE=Uncwilly;304459]0.0000% all CPU grunt work. Some of it is borg boxen.[/QUOTE]
Just leave some low-hanging fruit to the GPUers.

CPUs are best for LL work these ol' days.

Power to the elbow.

D

Cheers

[QUOTE=c10ck3r;304458]The first part was what I [URL="http://mersenne.org/report_top_500_custom/?team_flag=0&type=1001&rank_lo=28&rank_hi=28&start_date=&end_date=&B1=Get+Report"]disagreed[/URL] with, Hrothgar...[/QUOTE]
That's just because you don't know how to use data.
Tried this? -

[QUOTE=davieddy;304454]Give us a clue.
What era?

D[/QUOTE]
The era of my father. He told me that story on very many occasions (he has short memory, ...and this appears to be hereditary*). Usually he would just say the punchline - "[B][I]Нас н[/I][/B][FONT=Verdana][SIZE=2][B][I][FONT=Times New Roman]é[/FONT]сколько[/I][/B]!"[/SIZE]
[/FONT]
[FONT=Verdana][SIZE=2]The whole story goes like this:[/SIZE][/FONT]
[QUOTE]Correspondent: "Кого бы Вы назвали из современных великих ученых?"
The scientist: "Нас несколько"[/QUOTE]
Could have been one of their lesser known to the world local nuclear physicists (only Sakharov is well-known from the former VNIIEF), or it could be an abstract caricature.

___
*yes, yes, I can joke at my own expense. Here's another hereditary joke. "Jones is a hereditary bachelor. His father, grandfather and grand-grandfather and grand[SUP]3[/SUP]-father were all bachelors."

I'm surprised that so many forum members (>75%) are active GIMPS contributors. Since many users are talking about factoring software (such as msieve and NFS@Home) as opposed to Prime95, I was expecting the percentage to be much lower. I guess I was wrong!

After all, nobody says one cannot join both GIMPS and another math project (although it's impossible to allocate 100% of the available computing resources to both). :smile:

Missing option: Current participant, completely unfocused.

(how can one be focused on finding primes? you have to clear a lot of exponents first, for this; I think this is like the "i am in for the fun"/"i am in for the money" pool, but all have to DIG first)

I decided "current, focussed on clearing exponents" was closest, though I do some LL too. My 365 day stats:
TF 12.2%
P-1 37.0%
LL 33.4%
LL-D 17.4%

I think a good rule would be as follows:

[B]You are focused on finding new primes if one or more of the following apply...[/B]

1. You joined GIMPS in hopes of earning a prize.
2. You mostly request first-time tests.
3. You often request record-size numbers to test.

[B]You are focused on "clearing" exponents if one or more of the following apply...[/B]
1. You mostly request work that "makes the most sense" (according to PrimeNet).
2. You do a lot of double-checking.
3. You do a lot of P-1 or TF on exponents of unknown status.
4. You don't care so much about winning a prize (although many of us would admit that it'd be nice) as much as cutting down on the number of non-DC'd exponents.

For the purpose of this poll, you are considered a GIMPS participant as long as you've returned a single work unit, no matter how small.

1. I joined GIMPS in hopes of earning a prize. - in fact that was the only reason I joined, and I was dreaming of finding the biggest prime in few weeks, take the money. There was a time I had hundreds of cores in the project, and after the EFF award was taken I let them die slowly (tired to run around the building for maintenance, and P95 lacking a mechanism of centralized command). I kept only the things around my desk/office/home, easier to maintain.
2. You mostly request first-time tests. - not anymore, but the only reason is that the hardware I have is more efficient for other tasks
3. You often request record-size numbers to test. - not anymore, I don't have the patience, and changing the hardware very often - that is the job, I make/test hardware - long term tasks won't really fit.

1. You mostly request work that "makes the most sense" (according to PrimeNet). - no, never. I like to use manual gear, all my life hated cars with automatic gear box, the car goes on fifth when you need third and viceversa. I manually specify the work units, according with the hardware and the estimation of the time the hardware will be available.
2. You do a lot of double-checking. - indeed, but the only reason is that I have few GPU's and I don't trust them running CudaLucas for the first time. For DC/TC I can check the result against something existent, and they also take less time.
3. You do a lot of P-1 or TF on exponents of unknown status. Indeed, same reason as above. Of course, I would like to test thousands of billion digits numbers, find a prime and take the money, then come back here and make fun of you all, but I am aware of the fact that such task is impossible to accomplish with the hardware and the time I have (except the making fun part :razz:).
4. You don't care so much about winning a prize (although many of us would admit that it'd be nice) as much as cutting down on the number of non-DC'd exponents. - Oh really? C'mon! You find the prime, gimme the money! Who give a damn for cutting down exponents? :razz:
[ATTACH]8209[/ATTACH]

No wonder that Chris Cooper finds all the primes (and by now, he'd already found two more but holds on to them because [SPOILER]two is so pass[FONT=Calibri]é[/FONT] - he wants to submit three at the same time[/SPOILER]).
Everyone else is clearing composites. :-)

Of course, seriously, this is not true either, and this poll has a systematic bias - it polls the [I]forum members[/I]. 99% of the crunchers don't even know that the forum exists. Of them ~50% (or 30% or 90%?) may not even have a human behind them (I am referring to the Niesluhowski paradox - if one of his computers had deciphered the SETI signal he would never know! It could have been in a small classroom 50 miles away from him. He allegedly included SETI@Home + BOINC to CDs that were used to image all school district computers).

Sometimes this is how a new prime goes unnoticed for three months. (And the same nightmare is now in progress at PrimeGrid; there's a record prime, but no one can find the human who found it. The computer that reported the number stays cool under torture and never gave them the name. 8-S )