mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Data

Reply
 
Thread Tools
Old 2018-06-02, 15:41   #2850
GP2
 
GP2's Avatar
 
Sep 2003

258210 Posts
Default

Quote:
Originally Posted by retina View Post
The assignment had not expired. The page showed it had more than two weeks remaining till expiry. From what I understand it wasn't assigned as cat-0. So the user was promised some time to finish it. The user was still within the time. And later someone/something decides that the expiry date was not sufficient and it gets finished by someone else. At least that it my understanding. I'm sorry if I got some facts wrong, but it feels wrong that a promise is given and then revoked later.
Even for Category 1, the expectation is that it will be completed within 30 days:


Category 1

Computer must have enough LL and DC GHz-days over the last 120 days to indicate the assignment will be completed in 30 days


That's not phrased as a requirement, and if this particular user was doing their best and making steady daily progress, then probably they would have been given some slack. But "stalled at 0% progress per day for the last couple weeks", as Madpoo noted? They're going to get poached without sympathy.

This happens with every milestone when it drops down to the low single digits. Stuff happens, a bit of drama ensues, and then everybody forgets about it until the next milestone.
GP2 is offline   Reply With Quote
Old 2018-06-02, 15:53   #2851
GP2
 
GP2's Avatar
 
Sep 2003

2×1,291 Posts
Default

PS,

I run LL tests on the cloud. Sometimes spot instances prices spike, and then the tests stall until prices drop back to more normal levels.

For that reason, I keep the "promise to complete assignments quickly" flag turned off. I don't get automatically assigned anything below category 2, although on rare occasions I manually grab something smaller if it's strategic.

Maybe the Assignment Rules phrasing could be changed to say not just "promise to complete assignments quickly", but also add an admonition about needing to make steady progress. Category 1 assignments that stall without progress for a couple of weeks are fair game when the milestone countdown gets low enough, that's the reality and the wording should be changed to reflect it.
GP2 is offline   Reply With Quote
Old 2018-06-03, 00:32   #2852
Madpoo
Serpentine Vermin Jar
 
Madpoo's Avatar
 
Jul 2014

327710 Posts
Default

Quote:
Originally Posted by rudy235 View Post
I do not see it that way. i believe thar Madpoo has been restrained.

He probably has already the residue for 43967279 calculated but he is waiting for the original assignee to conclude his process.

At this point it looks likely that it will conclude and there is no indication that Madpoo is going to go ahead with his results, if that were the case.
Indeed, I've already run my own test on M43967279 but I noticed that the assignee finally got off his behind and got those assignments going again, so I held off checking those in.

All 4 of those exponents were idle for 3 weeks. I did 3 of them, the ones that would expire first, and it's still likely they would have expired before he finished. This last one may finish in time with a couple days to spare. It all depends on whether he stalls out again or not.

For what it's worth, I don't think poached assignments are considered when determining if a computer can get cat 0 or cat 1 exponents; only expirations that occur from actually taking too long (the method of expiration is logged).

Was I too impatient? Well... time will tell. Based on how the progress had stalled out entirely, in my mind it was reasonable to assume they were just stuck. And yeah, it did kind of bug me that we have someone who got cat 1 assignments and then for 3 weeks nothing was happening. That probably made me more impatient than I would have been otherwise.

Anyway, at this point I guess I'll see if the other 3 I poached finish before they would have expired or not. If they do, then I'll have to suck it up and issue a big mea culpa. Plus I've learned that I shouldn't let my annoyance at a particular exponent's progress make me more impatient. I could have at least waited a bit longer to see if progress resumed (which it did, a few days later).
Madpoo is offline   Reply With Quote
Old 2018-06-03, 00:41   #2853
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

249F16 Posts
Default

Quote:
Originally Posted by Madpoo View Post
I could have at least waited a bit longer to see if progress resumed (which it did, a few days later).
Patience, young grasshopper...

At the end of the day, if you hadn't done this, someone else probably would have.

And with regards to the claim that you've "wasted cycles", it could be pointed out that none of the work we do here is really all that important. Other than presenting "driving problems", of course....
chalsall is offline   Reply With Quote
Old 2018-06-04, 16:35   #2854
Mark Rose
 
Mark Rose's Avatar
 
"/X\(‘-‘)/X\"
Jan 2013

2,917 Posts
Default

It would be nice if the "lowest exponents" setting were per-machine and not account-wide. I have some machines that get shut off occasionally that I don't want to get low exponents. I have others that run 24/7, and I'd like these to work on low exponents.
Mark Rose is offline   Reply With Quote
Old 2018-06-05, 05:16   #2855
S485122
 
S485122's Avatar
 
Sep 2006
Brussels, Belgium

110010111012 Posts
Default

Quote:
Originally Posted by Mark Rose View Post
It would be nice if the "lowest exponents" setting were per-machine and not account-wide. I have some machines that get shut off occasionally that I don't want to get low exponents. I have others that run 24/7, and I'd like these to work on low exponents.
If you set the "days of work to queue up" higher than the threshold you will get a higher category. If you set it to 11 days you should get CAT 3 work units. This of a setting that you can apply to selected machines.

Jacob
S485122 is offline   Reply With Quote
Old 2018-06-06, 23:05   #2856
Mark Rose
 
Mark Rose's Avatar
 
"/X\(‘-‘)/X\"
Jan 2013

2,917 Posts
Default

Quote:
Originally Posted by S485122 View Post
If you set the "days of work to queue up" higher than the threshold you will get a higher category. If you set it to 11 days you should get CAT 3 work units. This of a setting that you can apply to selected machines.

Jacob
True!
Mark Rose is offline   Reply With Quote
Old 2018-06-08, 04:38   #2857
Madpoo
Serpentine Vermin Jar
 
Madpoo's Avatar
 
Jul 2014

29×113 Posts
Default

Quote:
Originally Posted by chalsall View Post
Patience, young grasshopper...

At the end of the day, if you hadn't done this, someone else probably would have.
...
Probably.

I checked his 4 exponents (3 of which I poached). None of them are done yet, and curiously enough, the last check in on all 4 are showing zero progress from the day before. That's after a decent 6.6% and 7.3% progress the previous 2 days (after the 3 week slumber they were in). 2 of those 4 decided to stall out at 98% and 97% which, if I hadn't already poached them, would have been even more aggravating if they were stuck there for a while with everyone watching and waiting. LOL

That last remaining exponent that I haven't poached (well, not officially... my result is done and waiting for a better time) is now stuck again at 83.2%. I just hope it picks up again soon. It expires in 8 days so it should really only take about 3 days of it's *actual* rate of progress that I know it's capable of, so if I see it stall out for 5-6 more days I'll really think about checking mine in when I feel like it couldn't finish in time even if it did start up again.
Madpoo is offline   Reply With Quote
Old 2018-06-09, 05:11   #2858
Madpoo
Serpentine Vermin Jar
 
Madpoo's Avatar
 
Jul 2014

29·113 Posts
Default

Well, it's still stalled out, second day in a row of that.

Of his last 4 (3 I already poached):
43920223 and 43921463 - would have expired on it's own in 2 more days
43962551 - would have expired on it's own in 7 more days
43967279 - will expire in 7 more days

The 2 that would have expired in a couple days are *soooo* close. 98% and 97%. Just one more little oomph will finish those. 43962551 at 78.1% and 43967279 at 83.2% would normally be expected to finish in 7 days, but these periods of inactivity throw a wrench into it.

Now, when his system is really cooking along, it can do an average 7% per day, give or take. That's only 2-3 days for those exponents that are 78.1% and 83.2%, and only a few hours for the ones at 98% / 97%. So... c'mon!
Madpoo is offline   Reply With Quote
Old 2018-06-09, 13:51   #2859
rudy235
 
rudy235's Avatar
 
Jun 2015
Vallejo, CA/.

23×112 Posts
Default

I was soo confident that 43967279 was going to finish in time.

Now I'm beginning to doubt it.


In the meantime the under 79 million first time verifications have gone down from 18 to 6. https://www.mersenne.org/assignments...xtf=1&exdchk=1
rudy235 is offline   Reply With Quote
Old 2018-06-10, 14:28   #2860
patrik
 
patrik's Avatar
 
"Patrik Johansson"
Aug 2002
Uppsala, Sweden

1101010002 Posts
Default

Quote:
Originally Posted by rudy235 View Post
In the meantime the under 79 million first time verifications have gone down from 18 to 6. https://www.mersenne.org/assignments...xtf=1&exdchk=1
What does it mean that one of the exponents is done to 100%, but is still listed as assigned?
Attached Thumbnails
Click image for larger version

Name:	under79Mexp.jpg
Views:	55
Size:	209.7 KB
ID:	18492  
patrik is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Newer X64 build needed Googulator Msieve 73 2020-08-30 07:47
Performance of cuda-ecm on newer hardware? fivemack GMP-ECM 14 2015-02-12 20:10
Cause this don't belong in the milestone thread bcp19 Data 30 2012-09-08 15:09
Newer msieves are slow on Core i7 mklasson Msieve 9 2009-02-18 12:58
Use of large memory pages possible with newer linux kernels Dresdenboy Software 3 2003-12-08 14:47

All times are UTC. The time now is 02:55.

Fri Dec 4 02:55:36 UTC 2020 up 23:06, 0 users, load averages: 1.99, 2.01, 1.71

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

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.