mersenneforum.org  

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

Reply
 
Thread Tools
Old 2020-04-22, 09:45   #309
lycorn
 
lycorn's Avatar
 
Sep 2002
Oeiras, Portugal

101011111102 Posts
Default

Something´s fishy with the numbers:

The total number of prime exponents below 1G doesn´t match the known number of 50847534.

If you take data from 1 day prior, both the number of factored and unfactored exponents appear in green.

Most of the progress done yesterday on the 17M range is not reflected in the table.
lycorn is offline   Reply With Quote
Old 2020-04-22, 15:39   #310
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

312510 Posts
Default

Quote:
Originally Posted by lycorn View Post
Something's fishy with the numbers
Something is. And I'm not yet sure what or how or why. Looks like this will eat up most of my day trying to hunt down what's going on.
James Heinrich is offline   Reply With Quote
Old 2020-04-22, 16:14   #311
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

55 Posts
Default

Quote:
Originally Posted by James Heinrich View Post
I'm not yet sure what or how or why.
OK, I think I finally grasp where the problem is coming from. In short it's related to the recent (unrelated) problems with data imports and such that caused the visualization stats for a given date (between Apr 16-21) to be rebuilt several times, and due to my SPE could result in exponents being double-counted if a bit range was completed between the first and subsequent data snapshots.

Assuming that is the problem, I have corrected the source of the problem, and rebuilt data snapshot for Apr 21. If this is indeed the problem then data from Apr 22 onwards should be sane.

Thanks for everyone who keeps an eye on these things for me and points out when I break things.
James Heinrich is offline   Reply With Quote
Old 2020-04-22, 17:00   #312
lycorn
 
lycorn's Avatar
 
Sep 2002
Oeiras, Portugal

2·19·37 Posts
Default

Quote:
Originally Posted by James Heinrich View Post

I have corrected the source of the problem, and rebuilt data snapshot for Apr 21. If this is indeed the problem then data from Apr 22 onwards should be sane.
So far, it appears OK . At least, all the discrepancies I reported are corrected.


Quote:
Originally Posted by James Heinrich View Post
Thanks for everyone who keeps an eye on these things for me and points out when I break things.
We thank you for your time and efforts to provide us with an informative and interesting tool!
lycorn is offline   Reply With Quote
Old 2020-05-11, 07:56   #313
LaurV
Romulan Interpreter
 
LaurV's Avatar
 
Jun 2011
Thailand

3·2,957 Posts
Default

Hey James, the graphic called "Known factor bit depths (0-999M)", is indeed to 999M? Or it is to 1G and it is just misnomed?
LaurV is online now   Reply With Quote
Old 2020-05-11, 14:36   #314
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

55 Posts
Default

Quote:
Originally Posted by LaurV View Post
Hey James, the graphic called "Known factor bit depths (0-999M)", is indeed to 999M? Or it is to 1G and it is just misnomed?
The main graphs page lists the exact exponent ranges covered for each graph. The one you're referencing is there described as "PrimeNet range (up to M999,999,999)".
I have adjusted the links on the site menu to be a little more descriptive (and precise, for the pedants ).
James Heinrich is offline   Reply With Quote
Old 2020-05-28, 07:14   #315
snme2pm1
 
"Graham uses ISO 8601"
Mar 2014
AU, Sydney

241 Posts
Default TF Success Stats Latency

Hi James
I have been aware of some latency in the propagation of TF success results, often by a day, but before now have not bothered to mention it, since by the next day the statistics seem to be caught up.

I'm even more puzzled by statistics today after yesterday lodging not one, but two TF success results. At first I worried that I had an unassigned competitor in the space, but then remembered the prior experience of latency.

In particular from https://www.mersenne.ca/status/tf/0/0/4/11800#
I see one result to 75 bits from 118000000 to 118099999, but mersenne.org is vacant for such.

It's no big deal, but there does seem to be an unwarranted delay in the sequence of processing.
My relevant transactions were:
M118074599 has a factor: 59851712641103678686393 [TF:75:76*:mfaktc 0.21 barrett76_mul32_gs]
found 1 factor for M118074599 from 2^75 to 2^76 (partially tested) [mfaktc 0.21 barrett76_mul32_gs]
M118074637 has a factor: 29327686458922061280743 [TF:74:75*:mfaktc 0.21 barrett76_mul32_gs]
found 1 factor for M118074637 from 2^74 to 2^75 (partially tested) [mfaktc 0.21 barrett76_mul32_gs]

P.S. From my distance, it seems to me that results have been accumulated to TF slots despite the imperative that the slot needs to be resigned as TF Factor found.

Last fiddled with by snme2pm1 on 2020-05-28 at 08:07 Reason: Qualification
snme2pm1 is offline   Reply With Quote
Old 2020-05-28, 13:41   #316
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

55 Posts
Default

There are multiple levels of latency, depending exactly what you're looking at. A factor reported to mersenne.org will be bundled up (at midnight UTC) and imported to mersenne.ca (approximately 3 hours later). And then various reports and such need to be rebuilt after that. So it can potentially be anywhere from maybe 4 to 29 hours between reporting a factor and seeing it reflected on mersenne.ca

Looking at your two examples, I don't see anything unusual:
https://mersenne.ca/M118074599 vs 118074599
https://mersenne.ca/M118074637 vs 118074637

You reported both factors right at the last possible minute (literally -- 2020-05-27T23:59:12) so they were included in the 2020-05-27 data bundle produced a couple minutes later, and imported into mersenne.ca a few hours later. However, your post (if I read the forum dates/times correctly) was about 2 hours before the data would have appeared on mersenne.ca

The TF status report is also rebuilt once per day. Looking at the database right now there one exponent at 75 in 118.0M range is no longer there (the other 52 in the 118M range are). Possibly it's an exponent that you were working on and since your submissions appear to have been right around midnight UTC it's possible that exponent's data hadn't been processed at the time of the report cutoff.

With the latency it's inevitable that something will always be "off", and fix itself by the next update (by which time, of course, something else will be "off").
James Heinrich is offline   Reply With Quote
Old 2020-05-29, 06:08   #317
snme2pm1
 
"Graham uses ISO 8601"
Mar 2014
AU, Sydney

241 Posts
Default

Quote:
Originally Posted by James Heinrich View Post
You reported both factors right at the last possible minute (literally -- 2020-05-27T23:59:12) so they were included in the 2020-05-27 data bundle produced a couple minutes later, and imported into mersenne.ca a few hours later. However, your post (if I read the forum dates/times correctly) was about 2 hours before the data would have appeared on mersenne.ca
That doesn't sound right, there were 9 results lodged at the same time and it appears that all were represented in some way. My observations of .ca were plenty of hours after your update processing, estimated to be at about 5:00 UTC, plus or minus an hour due to daylight savings affects.

Quote:
Originally Posted by James Heinrich View Post
The TF status report is also rebuilt once per day. Looking at the database right now there one exponent at 75 in 118.0M range is no longer there (the other 52 in the 118M range are). Possibly it's an exponent that you were working on and since your submissions appear to have been right around midnight UTC it's possible that exponent's data hadn't been processed at the time of the report cutoff.
Now with the turn of another day of .ca processing, one 75 bit and one 76 bit result have been retired to the Factored column. Nothing else has moved.
Maybe there is an argument that sustains some merit in retaining a factored result outside the Factored column for an extra day, as if to expose it?
Yet if that were not a deliberate intention, that is the behaviour that I'm referring to.
snme2pm1 is offline   Reply With Quote
Old 2020-05-29, 13:21   #318
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

55 Posts
Default

Quote:
Originally Posted by snme2pm1 View Post
Maybe there is an argument that sustains some merit in retaining a factored result outside the Factored column for an extra day, as if to expose it?
Nothing is deliberately "delayed". What you see is the data as it existed on PrimeNet as of the date shown at the top of the report. If you want to see how the most recent data compares with the previous day's data, the comparison report is available.
James Heinrich is offline   Reply With Quote
Old 2020-05-30, 03:18   #319
snme2pm1
 
"Graham uses ISO 8601"
Mar 2014
AU, Sydney

111100012 Posts
Default

If I read you properly, there would appear to be a day delay in the transfer of material particularly factor found from mersenne.org to .ca.
Definitely there is a day of delay from a factor found result being issued compared to being retired to the factor found column, by one means or other.
snme2pm1 is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Small inconsistencies between mersenne.org and mersenne.ca factor databases GP2 mersenne.ca 44 2016-06-19 19:29
mersenne.ca (ex mersenne-aries.sili.net) LaurV mersenne.ca 8 2013-11-25 21:01
Gaussian-Mersenne & Eisenstein-Mersenne primes siegert81 Math 2 2011-09-19 17:36
Mersenne Wiki: Improving the mersenne primes web site by FOSS methods optim PrimeNet 13 2004-07-09 13:51

All times are UTC. The time now is 03:49.

Tue Oct 27 03:49:12 UTC 2020 up 47 days, 1 hr, 0 users, load averages: 1.74, 1.82, 1.83

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.