LLRnet rally Apr. 30th to May 2nd

[gd_barnes]

Originally Posted by mdettweiler

I've added a table of primes found during the rally to the first post of this thread. So far it's just Lennart's two (one confirmed and one new), but I see jamers has found a (not yet reported) confirmed prime as well.

Edit: Whoops, I forgot about Gary's that I got a notification of earlier today (and which he just reported). I'll add that now.

Thanks, Max. Oh, and thanks for setting up that PRPnet server for Lennart and/or coordinating that with AMDave on such short notice. I'm sure it took a little while.

I went ahead and "reported" jamers' confirmed prime and showed it in the first post here.

[mdettweiler]

Originally Posted by gd_barnes

Thanks, Max. Oh, and thanks for setting up that PRPnet server for Lennart and/or coordinating that with AMDave on such short notice. I'm sure it took a little while.

No problem--it did take a little while, but it was somewhat of an eventuality that would have to be done when I eventually got around to testing PRPnet 3.2.5. Speaking of which, I'm quite pleased with how 3.2.5 is working so far--I think we may have a winner! After the rally is over I plan to switch my personal server over to it so I can further get the hang of it, and later switch the CRUS port 1300 server.

[mdettweiler]

Well, the rally officially ended about an hour ago, weighing in at a total of 4 new primes and 2 confirmed. We didn't quite make it to our goal of n=770K on port 3000, though combined with Lennart's efforts on port 9000 we did quite a bit of work. And we're still close enough to 770K that I estimate we can reach it later today if we maintain our current rate of processing.

Even though the rally is over, we of course welcome everyone involved to stick around! I for one plan to keep my quad on port 3000 for the next few days--hopefully I can find a prime before I move back to my personal work.

Thanks to all who participated for a great rally!

[gamer007]

I'll keep my quad on for another day. After that, I'll be going back to the mini-drive which seems like I'm doing alone now.

[gd_barnes]

Originally Posted by mdettweiler

Well, the rally officially ended about an hour ago, weighing in at a total of 4 new primes and 2 confirmed. We didn't quite make it to our goal of n=770K on port 3000, though combined with Lennart's efforts on port 9000 we did quite a bit of work. And we're still close enough to 770K that I estimate we can reach it later today if we maintain our current rate of processing.

Even though the rally is over, we of course welcome everyone involved to stick around! I for one plan to keep my quad on port 3000 for the next few days--hopefully I can find a prime before I move back to my personal work.

Thanks to all who participated for a great rally!

I will second all of Max's comments here. Thanks to all who participated and I'd encourage people to stick around for a while if they'd like. Primes at this level come in in the top 1000 of all time.

I'll leave almost all of my machines on until later Sunday night and then start pulling off but will maintain a higher level on this than before. I'll leave 5 quads (vs. 3 prior to the rally) here to fill in the gap up to n=780K, which I think can be done in 7-10 days with some help from others, assuming no fftlen change.

I'd like to run one more rally before we get into the major heat in the middle of the summer; perhaps over Memorial Day weekend or the weekend after. Although it's somewhat a long shot, it would be nice to hit our goal of n=1M on the 5th drive by the end of this year. It will likely take several rallies and a couple of new "regulars" to accomplish it. If we can get one of the real heavy hitters or several smaller players from AMD users, we might be able to create a good rivalry between the ROLP and AMD users teams.


Gary

[kar_bon]

I'll stay at Drive #5 with 12 cores to fill the gap and to reach that goal.

With about 3500 pairs/day (this rally lowered a bit the amount) to reach the goal till end of the year, it's much work to do!

Originally Posted by gd_barnes

I'll leave 5 quads (vs. 3 prior to the rally) here to fill in the gap up to n=780K, which I think can be done in 7-10 days with some help from others, assuming no fftlen change.

It's just around the corner: fftlen from 49152 to 57344 at n~769100 (at higher end of k-range).

[gd_barnes]

Originally Posted by Mini-Geek

Fortunately, you shouldn't do that.
The expectations deal with unknowns. If you expect 2.299 in the whole range, but know one of them to be prime, you still expect 2.299 in the whole unsearched range (minus some small amount for the removal of the one known-prime candidate; in some cases of ridiculously small numbers this might actually be significant).
It's more like this:
For each number, what fraction of the expected number of primes should we expect to find there?
Or another way:
For untested numbers, you should expect 1*(expected number, ignoring other factors). For single-tested numbers, assuming a 0.1% error rate, you should expect 0.001*(expected). For double-tested numbers, assume a 2^-64 chance of a false negative, so 2^-64*(expected). But for known primes, you can assume a 0 chance that it's an undiscovered prime.
Or to think of it a different way:
What if we happened to accidentally not include 405*2^765777-1 in our file. We would still expect 2.299 primes.
To take it to the extreme:
Say we find 5 primes in the first n=1000. Now how many do you expect in the rest of the range? About 2.299-5 or around 2.

(to put it in slightly more technical terms: whether a number is prime is independent of whether any other number that happens to be near it is prime; they are independent events, not dependent events, so past events do not influence future events)

It would be more accurate to remove that one prime from the equation (so have one less candidate in the file...not a big difference ) and, for any others that we know or think to have been checked once, (e.g. k=405 in the range seems pretty likely) divide their chance of a prime by, say, 1000, and for any we know or think to have been checked twice, divide their chance of a prime by, say, 2^64.

Boy, I sure said the same thing in a lot of ways!

And I'll second all of Tim's comments here. :-)

To put it a "little bit" more simply and to give an exact estimate of the expected # of already-known and new primes for the rally range:

1. Let's say that 5% of all k's have already been searched, i.e. 5 out the 100 k's in this drive and that the k's are of average weight. (This should be pretty close to what it actually is.)

2. Find the # of expected primes. Max found that to be 2.299 from where we started the rally to n=770K.

3. Multiply the expected # of primes times the percentage of k's already searched, i.e. 5%. This will give you the expected # of already-known primes that there "should" be in the range.

4. Multiply the expected # of primes times the perctage of k's not yet searched, i.e. 95%. This will give you the expected # of NEW primes that there should be in the range.

The fact that we already KNOW that there is one already-known prime is irrelavent. Here is what the true expectation should be:

New primes: 2.299 * 95% = 2.184
Already-known (confirmed) primes: 2.299 * 5% = .115

Looking at actual, we got 2 new and 1 confirmed prime so right within expectation. Had the rally started 2 hours sooner, we would have gotten 3 new primes for the range, also within the normal expectation.

This applies to ALL n-ranges on this drive, assuming all have already had 5 average-weight k's searched. Most of the time, there will be a fraction of an already-known expected # of primes but the TRUE # of already-known primes will be zero, as would be the case for n=770K-780K. It just so happens that the rally range had 1.

We didn't calculate the # of expected new/confirmed primes for Lennart's range but it's very clear that he came in well above expectation so we can say that we got somewhat lucky during the rally. Without doing any calculating, I think for Lennart's range, it would have had an expectation of 0.75-1 new and .05-.1 already-known prime. Actual came in at 2 new and 1 already-known prime so a somewhat "lucky" range there whereas the main rally range was about normal.


Gary

[vaughan]

Has port = 3000 run dry or is the problem at my end (again).

I couldn't see if there is any work waiting on any ports on the noprimeleftbehind index page.

My machines all report "sleeping".

[gd_barnes]

Originally Posted by vaughan

Has port = 3000 run dry or is the problem at my end (again).

I couldn't see if there is any work waiting on any ports on the noprimeleftbehind index page.

My machines all report "sleeping".

Hum. Are you sure about that? There are no problems on this end and I checked the hourly totals and yours showed no drop in processed pairs returned to the server.

[vaughan]

Yeah sorry, must be my computers being flaky. bok's stats show I'm still running.

Having some dns issues back home on the ranch :)
It seems some of the computers on ethernet won't showup on LogMeIn but the wifi connected ones are showing. Sort of weird as usually ethernet is more stable.

Still this page shows 0 pairs left

http://www.noprimeleftbehind.net/stats/

[kar_bon]

Originally Posted by vaughan

Still this page shows 0 pairs left

http://www.noprimeleftbehind.net/stats/

The port status is not working properly for Gxxx-servers (so now all of them after Gary's farm now holds all servers).

Better use the Progress Report page for such listings or the Recent results one.

Best one the Status of all servers!