mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   News (https://www.mersenneforum.org/forumdisplay.php?f=151)
-   -   Merry Christmas and a prime! (https://www.mersenneforum.org/showthread.php?t=22837)

Prime95 2017-12-27 03:13

Merry Christmas and a prime!
 
A day after Christmas a computer has reported a new Mersenne prime. Unlike the numerous CudaLucas false positives we've had over the last few years, this one shows promise. It was assigned December 18, no errors during the run. This particular computer has reported 57 previous results including some successful double-checks.

Madpoo's running an LL test now. T-41 hours.

I'll reach out by email for the last save file, but with the holidays I may not get a response before Madpoo's run completes.

Fingers crossed!!

axn 2017-12-27 03:16

Nice! Here's hoping... *fingers crossed*

BTW, does the client itself do a quick selfcheck re-run from the last savefile in case of prime?

Prime95 2017-12-27 03:19

[QUOTE=axn;474944]BTW, does the client itself do a quick selfcheck re-run from the last savefile in case of prime?[/QUOTE]

No, its still in my long wish list.

axn 2017-12-27 03:22

[QUOTE=Prime95;474945]No, its still in my long wish list.[/QUOTE]

If Gerbicz error check becomes widespread, it will be a moot point.

Prime95 2017-12-27 03:28

Prime95 version 28.9 was used. Thus, no Jacobi checking during the run or at the end of the run.

gophne 2017-12-27 03:29

Wow...great news...All success.

Uncwilly 2017-12-27 05:16

Is a PRP being run on it too (in case it may finish first) as a check?

Did the e-mail system work? Did the residue masking work? (And is there a clue somewhere?) Can we assume that this is a WR or should we assume that this is a Double Check? Also, I would assume, based upon the times your are talking about, this is not in the 100M digit range.

Madpoo 2017-12-27 05:18

[QUOTE=Prime95;474943]...

Madpoo's running an LL test now. T-41 hours.
...[/QUOTE]

32 hours to go now (I setup that system to use 20 cores on it instead of the usual dual 14-core workers it runs).

I don't know if George has already reached out to anyone with a fast GPU system to get another verifying test going on it in parallel, just in case? What's the fastest GPU out there now for LL testing... Titan of some flavor? Anyone with one of those willing to lend some cycles for a couple days on the chance this is a new prime?

Otherwise, my verification won't technically count as a double-check since it's another Prime95 run and we'd still want one with another program (mprime, cudalucas, cllucas, etc). My run is just to know whether or not to go on to that next step. We've run into a lot of false positives over the past couple years, but this one has more confidence behind it. :smile:

Batalov 2017-12-27 05:22

Time to run mlucas on a C5 instance?

Madpoo 2017-12-27 06:09

[QUOTE=Uncwilly;474956]Is a PRP being run on it too (in case it may finish first) as a check?

Did the e-mail system work? Did the residue masking work? (And is there a clue somewhere?) Can we assume that this is a WR or should we assume that this is a Double Check? Also, I would assume, based upon the times your are talking about, this is not in the 100M digit range.[/QUOTE]

PRP... well, it raises an interesting question about how a new prime found via PRP would be handled. I'd guess that a pair of LL tests would be run since LL is deterministic and not "merely" probabilistic?

The email did successfully go out when the result was reported in, and fortunately I saw it pretty quick and started a verification run.

Since we have had those false positives lately (mostly from cudalucas, recently), what I've been doing when a "prime" comes in is to look at the age of the assignment. Most false reports (especially from cudalucas) were turning in results within hours of being assigned, so you could be pretty sure it was wrong.

There were the exceptions to that though, like the false positives from xolotl that were from Prime95, so the next thing I look at is the overall history of the machine that turned in the result, and of the user in general. How many double-check successes do they have, any bad results, etc. Are they a new user? Is the machine overclocked, etc?

So, yeah, everything in this case seems to be looking pretty good. It doesn't look like the machine is overclocked, it has a good history of matching DC's including some recent ones, in addition to the first-time checks it does. The user him or herself has been around for a while so this isn't some new person with a weird system...

In summary, fingers crossed, but it's looking good on the surface of it.

a1call 2017-12-27 06:23

Is it 22M dd+?
I'd like to let people I know, know before it makes the headline news.
Thanks in advance and good luck.:smile:

ewmayer 2017-12-27 07:02

[QUOTE=Prime95;474943]A day after Christmas a computer has reported a new Mersenne prime. Unlike the numerous CudaLucas false positives we've had over the last few years, this one shows promise.[/QUOTE]

[QUOTE=Madpoo;474957]Otherwise, my verification won't technically count as a double-check since it's another Prime95 run and we'd still want one with another program (mprime, cudalucas, cllucas, etc).[/QUOTE]

When I read George's OP his wording led me to understand that the prime-reporting run was done using CuLu ... but Aaron, you say it was in fact a Prime95 (or mprime) run?

@Serge: I don't recall if Gord Palameta (a.k.a. GP2) did timings for 16-or-more-core C5 instances, and if so, how those compared to the EC2-instance timings you got for your verify run of the last prime 2 years ago. I'll try to dig up some numbers tomorrow.

a1call 2017-12-27 07:27

A positive PRP test result for the candidate you are testing, would be a relativity quick test with a very strong indication that you do not have a false-positive.
Would it not?

ewmayer 2017-12-27 07:48

[QUOTE=a1call;474966]A positive PRP test result for the candidate you are testing, would be a relativity quick test with a very strong indication that you do not have a false-positive.
Would it not?[/QUOTE]

I don't see the advantage of using a PRP test in this context ... no faster than LL, and on the highly-reliable (typically with ECC RAM) systems usually used by George & Aaron for their 'first look' DCs the kinds of system glitches the Gerbicz error check is designed to catch are unlikely to come into play. Also, if the initial run used LL, using LL for the DC allows direct cross-comparison of intermediate Res64s, if the user's logfile is available. (At least I would hope it would work this way.)

Even if the original run didn't output interim residues in the manner I describe, when George/Aaron do their DC they set the flag to do so, which means if their run also says 'prime', then the official independent-software DCs can do such interim-residue cross-matching ... this is especially important for the GPU clients, which are very fast but error-prone, to be able to tell if the GPU DC has gone off the rails, in which case it is restarted from the most-recent interim savefile whose Res64 matches the Prime95 DC one.

axn 2017-12-27 10:00

[QUOTE=Madpoo;474957]What's the fastest GPU out there now for LL testing... Titan of some flavor?[/QUOTE]

Titan V. Probably will do this in under 18 hours. IIRC, TheJudger posted some benchies on it.

Prime95 2017-12-27 10:04

[QUOTE=Uncwilly;474956]Did the e-mail system work? [/quote]

Of course not. Chalk up yet another failure for this particular PHP code. I've asked Aaron and James to investigate.

What's weird is I copied that code for the recent PRP changes. I've successfully received several Mersenne-cofactor-is-PRP emails over the last few months.

[quote] Did the residue masking work? (And is there a clue somewhere?) [/quote]

The quick look around I did on the web site showed nothing of interest. Perhaps someone here can find a web page that doesn't mask the result as well as it should.

[quote] Can we assume that this is a WR or should we assume that this is a Double Check? Also, I would assume, based upon the times your are talking about, this is not in the 100M digit range.[/QUOTE]

You may assume anything you like :smile:

axn 2017-12-27 10:05

[QUOTE=TheJudger;473281]CUDA 9.0, CUDA driver 384.98, CUDALucas 2.05.1 (SVN rev. 99)

Benchmark FFT sizes './CUDALucas -cufftbench 2048 32768 20'
[CODE]Device Tesla V100-PCIE-16GB
Compatibility 7.0
clockRate (MHz) 1380
memClockRate (MHz) 877

fft max exp ms/iter
<snip>
4375 80897867 0.9595
4608 85111207 0.9649
5184 95507747 1.0124
[/CODE]

Oliver[/QUOTE]

Assuming 82M expo, Titan V should complete it in under 24 hrs.
EDIT:- The above benchmark is for a Tesla V100 which ought to be faster than Titan V.

Prime95 2017-12-27 10:08

@Ernst: Yes, the initial run was done with Prime95. Madpoo's unofficial DC is also with prime95.

An mlucas run and a CudaLucas run would be a good idea. Volunteers?

ET_ 2017-12-27 10:24

[QUOTE=Prime95;474947]Prime95 version 28.9 was used. Thus, no Jacobi checking during the run or at the end of the run.[/QUOTE]

Then I'm not the one who found it :sad:

As you know, I am maintaining the DoubleMersenne site. It would be nice to know the exponent in advance (once confirmed) to start the sieving (it should be tested up to at least 600T, or 210 CPU/days) and set up the updated pages.

Prime95 2017-12-27 10:29

[QUOTE=ET_;474975]Then I'm not the one who found it :sad:[/QUOTE]

So sorry, maybe next time.

I reached the potential discoverer by email. Of course, he/she is traveling and cannot send the last intermediate file to me. We have to wait for madpoo.

ATH 2017-12-27 11:11

[QUOTE=Prime95;474974]@Ernst: Yes, the initial run was done with Prime95. Madpoo's unofficial DC is also with prime95.

An mlucas run and a CudaLucas run would be a good idea. Volunteers?[/QUOTE]

I would like to start a Titan Black run and I could start a mlucas run on a C5 instance, I just need to compile mlucas on it.

The fastest GPU right now is the Tesla P100 and TheJudger might have one, as he has posted benchmarks.

TheJudger 2017-12-27 12:11

Fastest GPU today should be Tesla [B]V[/B]100 - TITAN V has "crippled" memory bandwidth and thus should be slower because CUDAlucas is memory bandwidth bound on this type of hardware.

Oliver

ixfd64 2017-12-27 13:40

In case anyone wants to do some sleuthing, the server lists one "success" under the name Jon Pace.

Edit: my prediction is M77196191.

ET_ 2017-12-27 14:09

[QUOTE=ixfd64;474980]In case anyone wants to do some sleuthing, the server lists one "success" under the name Jon Pace.

Edit: my prediction is M77196191.[/QUOTE]

That looks a bit too close to be possible... :rolleyes:

Batalov 2017-12-27 15:45

Let the Game of Guessing begin!

ATH 2017-12-27 15:57

My encrypted guess is: 63198095

GP2 2017-12-27 16:57

[QUOTE=TheJudger;474978]Fastest GPU today should be Tesla [B]V[/B]100 - TITAN V has "crippled" memory bandwidth and thus should be slower because CUDAlucas is memory bandwidth bound on this type of hardware.

Oliver[/QUOTE]

This GPU is available on AWS cloud as the p3.2xlarge instance type.

Prime95 2017-12-27 18:52

[QUOTE=ixfd64;474980]In case anyone wants to do some sleuthing, the server lists one "success" under the name Jon Pace.[/QUOTE]

This loophole is now closed for future discoveries. I also erased Xolotl's 6 false successes.

In the future, I'll have to manually update the SQL table after prime verification is complete.

Madpoo 2017-12-27 18:57

[QUOTE=GP2;474990]This GPU is available on AWS cloud as the p3.2xlarge instance type.[/QUOTE]

Just for giggles I looked up what it would cost me to run this on the Azure instance with 64 CPU cores (although I don't really know if those are physical or include HT cores, and how many sockets it's spread across). Best estimate though was it would cost me around $200-300 to test it out on that. LOL No thanks.

I haven't looked at the Azure GPU instances to see how they're priced but I'd guess it'd be somewhat similar to run a beefy system for a day or two.

Batalov 2017-12-27 19:05

[QUOTE=Madpoo;475002]Best estimate though was it would cost me around $200-300 to test it out on that. LOL No thanks.[/QUOTE]
Sheesh, that's steep. If one divined that this time this is a 'real one', then perhaps it would have been worth it. But not for ~10 false discoveries per year.

Last time I only [STRIKE]spent[/STRIKE] donated 2 x $20 for the two M49 verification runs with Mlucas (with different FFT sizes).

preda 2017-12-27 19:25

Does this new discovery (assuming confirmed), reduce by any amount the chance of finding another prime in the vicinity?

Madpoo 2017-12-27 19:51

[QUOTE=preda;475009]Does this new discovery (assuming confirmed), reduce by any amount the chance of finding another prime in the vicinity?[/QUOTE]

Impossible to say... I'm reminded of the (probably) 46th and 47th Mersenne primes that are within spitting distance of each other, for example.

Prime95 2017-12-27 20:43

[QUOTE=preda;475009]Does this new discovery (assuming confirmed), reduce by any amount the chance of finding another prime in the vicinity?[/QUOTE]

No, not that this is proven.

Prime95 2017-12-27 20:44

ATH has started a CudaLucas run.

I suggested an mlucas run be done by someone else. However, I think he's going to do it anyway.

GP2 2017-12-27 20:45

[QUOTE=Prime95;475001]This loophole is now closed for future discoveries. I also erased Xolotl's 6 false successes.

In the future, I'll have to manually update the SQL table after prime verification is complete.[/QUOTE]

Given the perennial failure of email notification, it might be worth leaving a few loopholes, just in case.

Since email didn't work this time, how exactly did anyone realize there was a new prime reported?

Madpoo 2017-12-27 21:37

[QUOTE=GP2;475033]Given the perennial failure of email notification, it might be worth leaving a few loopholes, just in case.

Since email didn't work this time, how exactly did anyone realize there was a new prime reported?[/QUOTE]

When I replied that the email notification worked, what I didn't realize was this was the "backup" email notification that we put in place as a secondary means. That got added after the "oops prime" incident and at least that worked out great.

Fortunately we (and really, it was James) figured out why the notifications weren't working... ended up being just a length limit in the SMTP protocol for the recipient list. :smile: Fixed!

Prime95 2017-12-27 21:37

[QUOTE=GP2;475033]Since email didn't work this time, how exactly did anyone realize there was a new prime reported?[/QUOTE]

The failsafe email worked.

The PHP code that sends an instantaneous email failed as it often does. James has looked it over and thinks it was due to too many recipients with no newlines in the recipient list.

The failsafe is a SQLServer stored procedure that runs daily and sends email to madpoo and myself. It looks for any reported primes in the last 45 days. It sends an email whether it finds a prime or not -- so that both madpoo and I can monitor whether the failsafe procedure has gone off the rails.

chalsall 2017-12-27 22:00

[QUOTE=Prime95;475039]The PHP code that sends an instantaneous email failed as it often does. James has looked it over and thinks it was due to too many recipients with no newlines in the recipient list.[/QUOTE]

Guys... Just wondering... and this is meant to be funny, and serious, at the same time.

Have you ever heard of testing your code?

(And this comes from the guy who coined the term Stupid Programmer Error (SPE). :smile:)

Madpoo 2017-12-27 23:04

[QUOTE=chalsall;475044]Guys... Just wondering... and this is meant to be funny, and serious, at the same time.

Have you ever heard of testing your code?

(And this comes from the guy who coined the term Stupid Programmer Error (SPE). :smile:)[/QUOTE]

I'm pretty sure it was tested, but as I think about it, I wonder if the test emails only went to a subset of the recipients. The limit on the recipient line was just a hair over, so it probably boiled down to a case of the test not exactly matching the real world use.

GP2 2017-12-27 23:06

[QUOTE=ixfd64;474980]In case anyone wants to do some sleuthing, the server lists one "success" under the name Jon Pace.

Edit: my prediction is [M]M77196191[/M].[/QUOTE]

There's certainly something weird about that exponent. There's an expired double-check assignment from 2016, but no record of any first-time test, other than the one that completed yesterday.

Prime95 2017-12-27 23:13

[QUOTE=Madpoo;475049]I'm pretty sure it was tested, but as I think about it, I wonder if the test emails only went to a subset of the recipients.[/QUOTE]

Yes, we do not blast the entire recipient list with annoying test emails.

Uncwilly 2017-12-27 23:21

[QUOTE=GP2;475051]There's certainly something weird about that exponent. There's an expired double-check assignment from 2016, but no record of any first-time test, other than the one that completed yesterday.[/QUOTE]But surely you don't want us to google the names of the 2 individuals that are on record for the LL and DC. :stirpot:

chalsall 2017-12-27 23:27

[QUOTE=Prime95;475053]Yes, we do not blast the entire recipient list with annoying test emails.[/QUOTE]

So, please correct me if I'm wrong...

The entire recipient list for notifications of new primes consists of you, James and Aaron.

Definitely grounds for claims of spamming.... :wink:

Prime95 2017-12-28 00:08

[QUOTE=chalsall;475056]So, please correct me if I'm wrong...

The entire recipient list for notifications of new primes consists of you, James and Aaron.

Definitely grounds for claims of spamming.... :wink:[/QUOTE]

I'd have to run a SQL query for the full list. It included discoverers of past Mersenne primes, Scott, myself, Aaron, Ernst, Chris Caldwell, Richard Crandall, and a few others. Just enough to go over 1023 characters.

rogue 2017-12-28 00:16

[QUOTE=Prime95;475057]I'd have to run a SQL query for the full list. It included discoverers of past Mersenne primes, Scott, myself, Aaron, Ernst, Chris Caldwell, Richard Crandall, and a few others. Just enough to go over 1023 characters.[/QUOTE]

Have you removed people that have died? Richard Crandall died five years ago.

ewmayer 2017-12-28 00:21

[QUOTE=rogue;475061]Have you removed people that have died? Richard Crandall died five years ago.[/QUOTE]

Sheesh, already 5 years on ... but agreed, "E's singin' in the choir invisible!" would seem reasonable grounds for removal from the distrib list.

Madpoo 2017-12-28 00:23

[QUOTE=rogue;475061]Have you removed people that have died? Richard Crandall died five years ago.[/QUOTE]

It definitely needs pruning. :smile:

Prime95 2017-12-28 00:43

[QUOTE=rogue;475061]Have you removed people that have died? Richard Crandall died five years ago.[/QUOTE]

I just looked. The answer is nope. I just did some cleanup, removing about 5 names, adding about 4. I'm missing the email addresses of some previous Mersenne discoverers.

Mark Rose 2017-12-28 00:52

I'm putting my money on ##,###,#17 ;)

chalsall 2017-12-28 02:29

[QUOTE=Prime95;475064]I just looked. The answer is nope. I just did some cleanup, removing about 5 names, adding about 4. I'm missing the email addresses of some previous Mersenne discoverers.[/QUOTE]

Did the resulting set pass the 2^10 character limit? :wink:

Prime95 2017-12-28 02:34

[QUOTE=chalsall;475068]Did the resulting set pass the 2^10 character limit? :wink:[/QUOTE]

I don't know. James says he fixed that. We need to find another prime to test that fix :)

I suppose we code try to code up a test case that sent it to several dummy addresses plus Aaron and myself.... Given our PHP prowess we'd probably mess that up too.

ewmayer 2017-12-28 02:44

[QUOTE=Prime95;475070]I don't know. James says he fixed that. We need to find another prime to test that fix :)

I suppose we code try to code up a test case that sent it to several dummy addresses plus Aaron and myself.... Given our PHP prowess we'd probably mess that up too.[/QUOTE]

Is there a way to trigger the system to try to send a test-message to the current recipient list? If so, could the system's e-mail access to the outside world be temporarily suspended, the test-message triggered, and the location where the system stashes messages whose delivery cannot be currently completed examined? If the message appears there it could be deleted and e-mail-access-to-outside-world restored. [Not an IT person, but ISTR this should work based on my long-ago IT-hackage on my old DEC OS X box.]

chalsall 2017-12-28 03:07

[QUOTE=ewmayer;475071]If so, could the system's e-mail access to the outside world be temporarily suspended, the test-message triggered, and the location where the system stashes messages whose delivery cannot be currently completed examined?[/QUOTE]

OMG! No!!! Don't mess with the MTA! Work "in situ".

Just set up a bunch of "throwaway" addresses, and trigger on those.

Mark Rose 2017-12-28 03:22

[QUOTE=chalsall;475072]OMG! No!!! Don't mess with the MTA! Work "in situ".

Just set up a bunch of "throwaway" addresses, and trigger on those.[/QUOTE]

Or if you use gmail, you can always do the [email]whoever+randomthing@gmail.com[/email] trick. Many other mail providers have a smilar feature, perhaps using a different character.

airsquirrels 2017-12-28 04:35

I’m just here to say hurrah for a (potential) new prime.

Alas I am both late to the party, and haven’t kept up with the latest overpriced NVIDIA cards so I can’t offer anything unique to the verification efforts. And since it sounds like the initial run was on CUDA, a non-GPU verification is probably best.

bsquared 2017-12-28 04:45

[QUOTE=Prime95;475070]I don't know. James says he fixed that. We need to find another prime to test that fix :)

I suppose we code try to code up a test case that sent it to several dummy addresses plus Aaron and myself.... Given our PHP prowess we'd probably mess that up too.[/QUOTE]

I don't know anything about PHP but couldn't you just send N separate emails? For loop?
No danger of exceeding a recipient list character limit there...

Madpoo 2017-12-28 05:30

[QUOTE=airsquirrels;475080]I’m just here to say hurrah for a (potential) new prime.

Alas I am both late to the party, and haven’t kept up with the latest overpriced NVIDIA cards so I can’t offer anything unique to the verification efforts. And since it sounds like the initial run was on CUDA, a non-GPU verification is probably best.[/QUOTE]

The original was a CPU run with Prime95 28.9 so there's my own verification run as a "just to make sure" but the official double-check would need to be a different program (i.e. mlucas, cudalucas, whatever).

ATH is doing a cudalucas run. I wouldn't personally be opposed to letting you run something if you were able to get it done quicker but at this point even a fast GPU might only finish at about the same time?

Wouldn't hurt to have someone run it with mlucas just as an extra validation. I'm not running any Linux distros at the moment and too lazy to set one up, otherwise I'd totally give that a shot.

ewmayer 2017-12-28 06:35

[QUOTE=Madpoo;475083]Wouldn't hurt to have someone run it with mlucas just as an extra validation. I'm not running any Linux distros at the moment and too lazy to set one up, otherwise I'd totally give that a shot.[/QUOTE]

ATH fired up an Mlucas run on an Amazon C5.18xlarge instance this morning, but hit avx-512 build issues with the 17.1 code and was forced to use an avx2 build as a result, got ~4 ms/iter before moving to a much cheaper C4 instance), since that also supports avx2. I sent him a patched v17.1 tarball just now, hoping to see appreciably better timings from use of avx-512, though even if so, I'm not sure he'd do the verify using that build due to the expense of running on C5.

I did some manythreaded timing tests using avx2 on David Stanfill's 32-core Xeon and the GIMPS KNL (the 2 machines I am currently using for side-by-side primality tests of F30 at FFT lengths 60M and 64M, respectively) ... getting just a smidge under 4 ms/iter there. Alas the KNL offers no faster alternative because

[1] It is severely underclocked relative to the Xeon to keep the massive die from melting;

[2] Even though the KNL has 64 physical cores and 256 logical ones vs the Xeon's 32/64, I cannot take advantage of the higher core count to make up for the lower clock speed because the parallel speedups peter out between 16 and 32 threads at the relatively small FFT length needed for the new-prime candidate. (OTOH, the large FFT lengths needed by my F30 runs take much better advantage of the high core/thread counts.)

Worst case we'll get an Mlucas verify within 72 hours - if Andreas gets some nice speedups from an avx-512 build on C5, that estimate will drop, even if I have to Paypal-bribe him the C5 run costs to get him to use that for the verify. :)

LaurV 2017-12-28 07:37

Yeaaa!

How did I miss this thread? (it looked like some xmas wishes, or what? I generally read all topics...)

I keep my fingers (and toes, tongue, eyes, testicles, everything) crossed!

If you need an extra CudaLucas test, give me a PM. (no, I have no idea about the exponent, I assume I joined the game too late this time).

ewmayer 2017-12-28 07:40

Update: GP2 did Mlucas 17.0 avx-512 build on c5.18xlarge, timing down to 2.9 ms/iter, we continue to play with core/threadcount params to try to reduce that further. Note this was not on the default instance of this type - in his words, a "pre-LTS (beta) version of Amazon Linux 2, which has gcc 7.2.1", compared to the gcc 6.4 of the default c5 instances, for which his timing of an avx-512 build at the same FFT length was 4 ms/iter. The latter is weird, because that's the timing ATH got using an avx2 build on the (IIRC) same instance type.

But timings weirdness seems to be the name of the game for lots of threads running on fairly exotic hardware- in my own avx2-build timing tests on David Stanfill's Xeon, I was able to get at best 4.8 ms/iter on the unloaded system, but just for giggles I redid all the timings with my big nice'd F30 run crunching away in the background, and voila! The LL-test timings dropped by 20%. Both jobs (new-prime verify at full priority, F30 test nice'd) have been running side-by-side for around 10 hours now, confirming the speedup-under-background-load is not a chimera - and even with the LL-DC consuming most of the FLOPS, the F30 run is still proceeding at ~1/3 of its normal speed. IIRC Serge Batalov ran into a similar phenomenon with his Mlucas verify of a then-new M-prime a few years back - he got permission to do it on a high-end server of the company he worked for at the time, meaning he had to put up with whatever 'real work' his colleagues were using the system for competing for cycles. And like with my run, his run was fastest when other jobs were running. It's like the NBA player who can only shoot straight when he has a defender in his face. :)

pinhodecarlos 2017-12-28 11:11

Are we there yet?

:popcorn:

pacionet 2017-12-28 11:40

wow ! A new mersenne prime ?
Can we know the range?
How many hours left to finish the check?

axn 2017-12-28 11:41

[QUOTE=pinhodecarlos;475103]Are we there yet?[/QUOTE]

[QUOTE=Madpoo;474957]32 hours to go now [/QUOTE]

That was 31 hours ago...

pacionet 2017-12-28 11:48

Please update the home page with somehow of "Probably found M50!"

Mark Rose 2017-12-28 11:53

[QUOTE=pacionet;475105]wow ! A new mersenne prime ?
Can we know the range?
How many hours left to finish the check?[/QUOTE]

Madpoo should have a result in a couple of hours.

And I think people are going to be surprised by the range.

Yes, I had found a leak that has since been plugged.

ET_ 2017-12-28 13:13

When M50 will be announced, I will appreciate whoever decides to donate a very small part of his CPU time to help sieving for MM50 :smile:

We need about 300 core days in total.

pacionet 2017-12-28 13:30

I try with M77324641

MushNine 2017-12-28 13:37

I thought the residue would be a string of zeroes for a prime. None of Jon's residue are such. So I suppose it's been removed from public record.

pacionet 2017-12-28 13:40

[QUOTE=MushNine;475118]I thought the residue would be a string of zeroes for a prime. None of Jon's residue are such. So I suppose it's been removed from public record.[/QUOTE]

If i remember correctly, the server writes a "masked" residue for a probable prime.

Madpoo 2017-12-28 13:43

Woohoo!
 
Just finished my double-check:
Mxxxxxxxx is prime!

Looks like we have a winner, but my run was using the same Prime95 as the first result and I don't know if George will want to wait for the pending cudalucas check before making the exponent public. Traditionally I think that was the process.

I saved the interim partial residues every 1M iterations (starting at 36M since I forgot at first) so we'll be comparing the cudalucas run along the way.

But...yeah, the "Boxing Day Prime" looks good! :smile:

pacionet 2017-12-28 13:45

[QUOTE=Madpoo;475121]Just finished my double-check:
Mxxxxxxxx is prime!

Looks like we have a winner, but my run was using the same Prime95 as the first result and I don't know if George will want to wait for the pending cudalucas check before making the exponent public. Traditionally I think that was the process.

I saved the interim partial residues every 1M iterations (starting at 36M since I forgot at first) so we'll be comparing the cudalucas run along the way.

But...yeah, the "Boxing Day Prime" looks good! :smile:[/QUOTE]

I think we can say "Probably found M50 : it is ..." .
Come on. :)

Brian-E 2017-12-28 13:49

I didn't believe GIMPS would ever find another one.

We know the exponent is above 36 million now. There I was thinking it might have been the result of a triple check of a mistakenly verified test from years and year ago.:smile:

Mark Rose 2017-12-28 13:54

[QUOTE=pacionet;475122]I think we can say "Probably found M50 : it is ..." .
Come on. :)[/QUOTE]

... higher than the last one ;)

pacionet 2017-12-28 13:56

I hope this forum will be the first to know the newly discovered prime! Come on!

Mark Rose 2017-12-28 14:08

[QUOTE=Brian-E;475124]I didn't believe GIMPS would ever find another one.

We know the exponent is above 36 million now. There I was thinking it might have been the result of a triple check of a mistakenly verified test from years and year ago.:smile:[/QUOTE]

Well we already knew that from earlier posted information. Ernst posted it would take at 72 hours to complete a run at 4 ms/iter, meaning it would be at least around 64 million.

pacionet 2017-12-28 14:13

I found Madpoo test, the prime could be M64764893 :) which seems have 3 LL (1 failed e 2 OK)

MadPoo
Manual testing
64764893
C
2017-12-28 13:29
1.8
163.7113
7B56C023189FB494


rudimeier
vinaka
64764893
C
2017-12-26 14:35
12.6
157.0998
7B56C023189FB494

Is it?

pinhodecarlos 2017-12-28 14:29

[QUOTE=pacionet;475129]I found Madpoo test, the prime could be M64764893 :) which seems have 3 LL (1 failed e 2 OK)

MadPoo
Manual testing
64764893
C
2017-12-28 13:29
1.8
163.7113
7B56C023189FB494


rudimeier
vinaka
64764893
C
2017-12-26 14:35
12.6
157.0998
7B56C023189FB494

Is it?[/QUOTE]

Maybe a decoy?

Madpoo 2017-12-28 14:40

[QUOTE=pacionet;475129]I found Madpoo test, the prime could be M64764893 :) which seems have 3 LL (1 failed e 2 OK)
[/QUOTE]

I try to snag all of the triple-checking assignments and get them finished quickly, otherwise whoever turned in the mismatch is left wondering, and sometimes they run their own triple-check. If they match their own, I'll be doing an independent verification anyway so I'd rather just do the triple-check and get it done for them. :smile:

paulunderwood 2017-12-28 14:43

[QUOTE=Madpoo;475121]Just finished my double-check:
Mxxxxxxxx is prime!
[/QUOTE]

:george:

airsquirrels 2017-12-28 16:04

I have two parallel runs going on Vega 64 FEs. ETA ~ 36 hours. gpuOwl with Jacobi checks.

I’m sure the other verifications that aren’t mprime will also complete before then, but it is nice to be thorough.

Batalov 2017-12-28 16:06

20+ years of team effort continues to amaze!
 
Congratulations to all GIMPS participants!
Everyone who ever tested a composite candidate made this prime possible!

pacionet 2017-12-28 16:22

Please tell us which is the new prime !

ET_ 2017-12-28 16:54

We are waiting (and longing for) the other GPU and mlucas checks.

Go GIMPS Go! :et_:

preda 2017-12-28 17:08

Congrats all!

there was a rather surprisingly large hole between M48 at 57'885'161) and the [ex] M49 at 74'207'281.

R. Gerbicz 2017-12-28 17:15

[QUOTE=preda;475159]Congrats all!

there was a rather surprisingly large hole between M48 at 57'885'161) and the [ex] M49 at 74'207'281.[/QUOTE]

I wouldn't say that: [url]https://primes.utm.edu/notes/faq/NextMersenne.html[/url] .

Luis 2017-12-28 17:22

@preda: LOL!

I tried to predict something similar more than three years ago.
[URL]http://mersenneforum.org/showthread.php?p=388050#post388050[/URL]

Who knows...
We, as humans, always look for linearity and simmetry. Math and prime numbers teach us a different plan.

Madpoo 2017-12-28 17:27

[QUOTE=preda;475159]Congrats all!

there was a rather surprisingly large hole between M48 at 57'885'161) and the [ex] M49 at 74'207'281.[/QUOTE]

My math on this is guaranteed to be wrong, so nobody crunch the numbers... but...

My theory is that we've missed a prime that will later be found by double-checking.

If we reckon that the error rate (so far) on first-time checks is around 3-4%, and we now have 50 Mersenne primes of ever increasing size, I think given that data it's safe to say that eventually a prime will most definitely be missed, sooner or later. My theory is that it's already happened.

If it hasn't already happened, it may still happen but I suppose that depends on the robustness of the new error checking methods that have been introduced. It'd be great if those really prove themselves worthwhile and we didn't have to "waste" 50% of the total compute capacity with double-checks.

Madpoo 2017-12-28 17:30

[QUOTE=Luis;475161]@preda: LOL!

I tried to predict something similar more than three years ago.
[URL]http://mersenneforum.org/showthread.php?p=388050#post388050[/URL]

Who knows...
We, as humans, always look for linearity and simmetry. Math and prime numbers teach us a different plan.[/QUOTE]

Whoa... the crazy part about that old post from just 3 years back is that we were just clearing the 53M first-time check milestone, and here we are coming close to the 77M milestone. We've been rocking it!

Spherical Cow 2017-12-28 17:45

1 Attachment(s)
Wow- Congrats to All! I wonder if this will generate a new button...

Norm

ET_ 2017-12-28 17:59

[QUOTE=Spherical Cow;475169]Wow- Congrats to All! I wonder if this will generate a new button...

Norm[/QUOTE]

If so, I'm asking for one... :razz:

pacionet 2017-12-28 18:08

Some years ago, if I remember correctly, someone sold a poster with the decimal expansion of newly discovered primes (a magnifying glass was necessary to read the digits).
Does it still exist ?

ET_ 2017-12-28 18:21

[QUOTE=pacionet;475174]Some years ago, if I remember correctly, someone sold a poster with the decimal expansion of newly discovered primes (a magnifying glass was necessary to read the digits).
Does it still exist ?[/QUOTE]

I have one of those posters.
Unfortunately, Richard Crandall (owner of Perfectly Scientific Inc.) is no more among us mortals, and AFAIK no one took such heritage.
Consider however that even using a 1-point font, the offset needed for such a huge poster would be enormous.

pacionet 2017-12-28 18:34

[QUOTE=ET_;475175]I have one of those posters.
Unfortunately, Richard Crandall (owner of Perfectly Scientific Inc.) is no more among us mortals, and AFAIK no one took such heritage.
Consider however that even using a 1-point font, the offset needed for such a huge poster would be enormous.[/QUOTE]

I didn't know about Richard's death. I am sorry. I hope he rest in peace.

Uncwilly 2017-12-28 18:40

[QUOTE=pacionet;475174]Some years ago, if I remember correctly, someone sold a poster with the decimal expansion of newly discovered primes (a magnifying glass was necessary to read the digits).
Does it still exist ?[/QUOTE]I talked with a printer that I know about making one as a banner. Let me contact them again. Allowing it to be a banner would give more room than off-set sheets.

GP2 2017-12-28 18:48

[QUOTE=preda;475159]there was a rather surprisingly large hole between M48 at 57'885'161) and the [ex] M49 at 74'207'281.[/QUOTE]

Actually, there are too many primes being discovered in the current range.

There are:

4 between 10[SUP]0[/SUP] and 10[SUP]1[/SUP]
6 between 10[SUP]1[/SUP] and 10[SUP]2[/SUP]
4 between 10[SUP]2[/SUP] and 10[SUP]3[/SUP]
8 between 10[SUP]3[/SUP] and 10[SUP]4[/SUP]
6 between 10[SUP]4[/SUP] and 10[SUP]5[/SUP]
5 between 10[SUP]5[/SUP] and 10[SUP]6[/SUP]
5 between 10[SUP]6[/SUP] and 10[SUP]7[/SUP]
11 between 10[SUP]7[/SUP] and 10[SUP]8[/SUP] (and now probably 12)

vasyannyasha 2017-12-28 18:50

Congratulations!!!

Sweet dreams are made of this
Who am I to disagree?
I travel the forum
And the seven threads

Spherical Cow 2017-12-28 19:02

[QUOTE=GP2;475182]Actually, there are too many primes being discovered in the current range.

There are:

4 between 10[SUP]0[/SUP] and 10[SUP]1[/SUP]
6 between 10[SUP]1[/SUP] and 10[SUP]2[/SUP]
4 between 10[SUP]2[/SUP] and 10[SUP]3[/SUP]
8 between 10[SUP]3[/SUP] and 10[SUP]4[/SUP]
6 between 10[SUP]4[/SUP] and 10[SUP]5[/SUP]
5 between 10[SUP]5[/SUP] and 10[SUP]6[/SUP]
5 between 10[SUP]6[/SUP] and 10[SUP]7[/SUP]
11 between 10[SUP]7[/SUP] and 10[SUP]8[/SUP] (and now probably 12)[/QUOTE]

Yes- When we had found 9 in the current range, I was convinced that was the end for this range. I was more convinced when we got to 10; and 11. Now, I'm only convinced that I'm wrong.

Norm


All times are UTC. The time now is 15:17.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.