mersenneforum.org > Data processed dc and tc posts
 Register FAQ Search Today's Posts Mark Forums Read

 2021-06-19, 18:09 #474 drkirkby   "David Kirkby" Jan 2021 Althorne, Essex, UK 3×149 Posts I may have put this in the wrong thread by mistake. (Yes, I am human, and I do make mistakes). There was a thread by Uncwilly requesting double-checks of some dubious exponents. Despite I would rather have done a first time PRP test, I decided to help out and check one of the exponents on the suspect list. I did comment that I had started the P-1 by mistake. I realised my error, but decided to let P-1 complete, as I was keen to know how much RAM would be used on stage 2. It was only 190 GB or so, which was far less than the 340 GB or so for a similar sized exponent a couple of weeks ago. I think the other one had erroneously be issued by the server indicting it would safe 2 primality tests, so probably B1 and B2 were larger, requiring more RAM. I was keen to experiment, so let it run. Personally I think the Mersenne numbers look better with the M, so I prefer M102164971 to 102164971, So that was a conscious decision. My comment about the certificate was just to highlight the fact that the prime/composite nature of the Mersenne number had not been fully settled at that point. I do know the server issues the certificates, so the comment about reading more was not called for. Whilst one can't get a manual assignment of verifying the proof, I suspect if one was determined to do so, one could probably get an assignment for a specific exponent n by putting the following in local.txt Code: GetMinExponent=n GetMaxExponent=n
2021-06-19, 22:33   #475
kriesel

"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

5,783 Posts

Quote:
 Originally Posted by drkirkby I completed a PRP test with proof of M102164971. It needs the certificate checked.
That probably was a true statement at the time. But unnecessary and not useful. And imagine if every completed PRP with proof generation was accompanied by such a needless forum post, how very cluttered certain threads in the forum might become. Handling such mundane and ephemeral data is the proper function of the PrimeNet server, not appropriate to the forum threads.
Quote:
 Originally Posted by Uncwilly There is no need to add the M in front of the number when using the [M] tags.
I think the form M102164971 is best (when not a part of a work request or acceptance list in worktodo entry form). Tastes vary.

Quote:
 Your post was not to: requests for DC and TC's and claim of exponents. as in post #1 There is no cause to post this as noted below. Because of the last sentence of the next quote. Not only does it happen automatically, the PrimeNet server must be involved between the person that does the PRP and the one that runs the cert. But, to know that requires reading. Also, there is no manual assignment type for certs.
... and also requires comprehension, memory, and application of good judgment.

Quote:
 Originally Posted by drkirkby I may have put this in the wrong thread by mistake.
There appears to be a consensus, and I concur, that putting a specific request for manual undertaking, of a work type that is only available by automatic assignment, not available for manual assignment, in any forum thread is a mistake.
Quote:
 (Yes, I am human, and I do make mistakes).
As do we all. And you seem sometimes to be particularly diligent about demonstrating it.
Quote:
 There was a thread by Uncwilly requesting double-checks of some dubious exponents. ... I decided to help out and check one of the exponents on the suspect list.
Effective contribution to the joint effort is always welcome.
Quote:
 I did comment that I had started the P-1 by mistake. I realised (Brit spelling?) my error, but decided to let P-1 complete, as I was keen to know how much RAM would be used on stage 2. It was only 190 GB or so, which was far less than the 340 GB or so for a similar sized exponent a couple of weeks ago.
(Yes, you've got a lot of ram in your one expensive system. And mention the outsized ram or corecount peculiarly frequently. Why is that?)

Such extreme RAM quantity may mildly speed factoring but is unnecessary for normal or even gigabit size exponent P-1. Such large ram allocations might be useful for gigadigit or F33 P-1 stage 2 factoring when suitable software becomes available. Wavefront exponents can be satisfactorily P-1 factored in systems with as little as 2GB ram. I've run up to 1G or even M1G+7 P-1 on 16GB of GPU ram in a few GPU days. (Only once >1G in response to an absurdly high M52 guess. Don't do more such, as there's nowhere to report them via scripting. James Heinrich generously manually entered that high P-1 in the mersenne.ca database, on the condition that I not do it again, a very reasonable position. There are better uses for his time, than acting as a clerk for not-needed-next-century results.)
Quote:
 I think the other one had erroneously be been issued by the server indicting indicating it would safe save 2 primality tests, so probably B1 and B2 were larger, requiring more RAM. I was keen to experiment, so let it run. Personally I think the Mersenne numbers look better with the M, so I prefer M102164971 to 102164971, So that was a conscious decision. My comment about the certificate was just to highlight the fact that the prime/composite nature of the Mersenne number had not been fully settled at that point. I do know the server issues the certificates, so the comment about reading more was not called for.
Please be open to coaching. A defensive reaction is understandable but works against hearing and taking to heart the coaching.
Quote:
 Whilst one can't get a manual assignment of verifying the proof, I suspect if one was determined to do so, one could probably get an assignment for a specific exponent n by putting the following in local.txt Code: GetMinExponent=n GetMaxExponent=n
Cert assignments can be quite fleeting. (Ben Delo, curtisc, and a few other top producers hoover up many of them fully automatically.) It's easy to miss the window of opportunity to obtain a handpicked exponent's Cert assignment, and leave an mprime worker idle for hours for lack of work as a result of such an attempt at capturing a specific ~1.6 GHzD Cert assignment. In the interest of productivity for the hardware and the participant, I'd advise against attempting it.

We continue to respond to your posts, with coaching posts, in the declining hope that it will pay off in the long run.

Last fiddled with by kriesel on 2021-06-19 at 22:37

 2021-06-21, 10:56 #476 Jan S   Oct 2018 Slovakia 8310 Posts I took these: PRP=1,2,107793883,-1 PRP=1,2,107957713,-1 (2 suspect LL's) and PRP=1,2,90200717,-1 (2 PRP results with shift 0)
 2021-06-21, 14:20 #477 Uncwilly 6809 > 6502     """"""""""""""""""" Aug 2003 101×103 Posts 10,007 Posts The list has been updated.
 2021-06-25, 06:03 #478 kriesel     "TF79LL86GIMPS96gpu17" Mar 2017 US midwest 5,783 Posts https://mersenneforum.org/showpost.p...&postcount=635 done; 3 eliminated by 4 P-1 factors found, 1 of which could have been found by TF to 81 bits first. 332327371 332328103 332328287 Last fiddled with by kriesel on 2021-06-25 at 06:18
 2021-06-25, 14:06 #479 Uncwilly 6809 > 6502     """"""""""""""""""" Aug 2003 101×103 Posts 234278 Posts Lots of new Cat 1 exponents. All above M48.
 2021-06-26, 17:46 #480 kriesel     "TF79LL86GIMPS96gpu17" Mar 2017 US midwest 5,783 Posts Cat 1 DoubleCheck=57906071,76,1 not available, drkirkby did matching TC Took: DoubleCheck=57934507,76,1 DoubleCheck=57939127,76,1 DoubleCheck=57939169,74,1 DoubleCheck=57973771,74,1 DoubleCheck=57987997,76,1 All are above Mp48*. Last fiddled with by kriesel on 2021-06-26 at 17:47
2021-06-26, 18:44   #481
Uncwilly
6809 > 6502

"""""""""""""""""""
Aug 2003
101×103 Posts

1000710 Posts

Quote:
 Originally Posted by kriesel Cat 1 DoubleCheck=57906071,76,1 not available, drkirkby did matching TC
When I las updated the list there was no assignment indicated on that exponent. Looks like somebody has decide to not read this thread and flout the rules. So much for the pledge to not poach.

2021-06-27, 01:48   #482
drkirkby

"David Kirkby"
Jan 2021
Althorne, Essex, UK

1101111112 Posts

Quote:
 Originally Posted by Uncwilly When I las updated the list there was no assignment indicated on that exponent. Looks like somebody has decide to not read this thread and flout the rules. So much for the pledge to not poach.
I thought poaching was taking work assigned to someone else. I checked, and the exponent was not assigned to anyone.

I did not realise that one needed to state the exponents one was going to test. I see the text "Always register your claims right away. which I never understood the meaning of. I'm now interpreting as meaning you want people to state the exponent they will test. Sorry, but I think that could be expressed a bit more clearly.

I assumed that the server would take care of two people trying to do the same thing, as they would not get an assignment. Maybe it's different with double-checks. Looking at my log, I don't see an assignment ID being issued, so I'm now starting to see what the problem is.

Sorry, but I think it could be clearer. Putting it in red does not help if the meaning is not clear.

 2021-06-27, 02:09 #483 Mark Rose     "/X\(‘-‘)/X\" Jan 2013 37·79 Posts I have a few Cat 2 exponents from this thread that I'm pausing work on for a week or two during the heatwave. The computer they're running on shares a circuit with my portable AC and blew the breaker last night.
 2021-06-27, 03:25 #484 Uncwilly 6809 > 6502     """"""""""""""""""" Aug 2003 101×103 Posts 10,007 Posts You should be good. If you need to, get a manual extension to the assignments. I trust you.

 Similar Threads Thread Thread Starter Forum Replies Last Post jasong Forum Feedback 1050 2019-04-29 00:50 10metreh Forum Feedback 6 2013-01-10 09:50 jasonp Forum Feedback 9 2009-07-19 17:35 edorajh Data 10 2003-11-18 11:26 Xyzzy Lounge 10 2002-11-21 00:04

All times are UTC. The time now is 12:16.

Thu Oct 21 12:16:39 UTC 2021 up 90 days, 6:45, 1 user, load averages: 1.42, 1.53, 1.49