User TJAOI - Page 15

Mark Rose · 2015-01-21, 17:21

Given the size and scope of that, having PrimeNet or GPU72 distribute rechecks might be wise.

chalsall · 2015-01-21, 17:27

Quote:

Originally Posted by petrw1

This chart LOTS of attempts (i.e. enough to expect a good average) but LOW success rate and high GhzDays per.
-- Again either large exponents; high bit levels or long bit ranges

Code:

Member Name        GHz-Days	Att	Succ	A/S	GhzDays Per  Att
For Research  	145449.386	25449	143	178.0	5.72
brilong   	210511.103	11283	70	161.2	18.66
Axelsson   	58385.659	10055	64	157.1	5.81

Hmmm... Interesting.

I'm "For Research". And I (hope it can be assumed) never cheat on TF'ing.

Also, I believe "brilong" and "Axelsson" can also be considered to be trusted.

Possibly an issue with the report?

petrw1 · 2015-01-21, 17:30

Quote:

Originally Posted by chalsall

Hmmm... Interesting.

I'm "For Research". And I (hope it can be assumed) never cheat on TF'ing.

Also, I believe "brilong" and "Axelsson" can also be considered to be trusted.

Possibly an issue with the report?

For the record: PLEASE...PLEASE....I am in no way accusing anyone of cheating.
Simply publishing stats (You know: "Lies, Damn Lies, and Statistics")

chalsall · 2015-01-21, 17:34

Further to the above, GPU72 also collects stats as to work assigned by it, comparing to expected vs actually found factors. A few examples:

Chris Halsall
Brian Long
Axelsson
Patrik Johansson

chalsall · 2015-01-21, 17:37

Quote:

Originally Posted by petrw1

For the record: PLEASE...PLEASE....I am in no way accusing anyone of cheating.

Not taken as such!

Quote:

Originally Posted by petrw1

Simply publishing stats (You know: "Lies, Damn Lies, and Statistics")

And SPEs....

flagrantflowers · 2015-01-21, 18:34

Quote:

Originally Posted by petrw1

Code:

Member Name        GHz-Days    Att    Succ    A/S    GhzDays Per  Att
For Research      145449.386    25449    143    178.0    5.72
brilong       210511.103    11283    70    161.2    18.66
Axelsson       58385.659    10055    64    157.1    5.81

There has been an issue in the past of found factors in the high (~74 bit) range being counted as p-1 successes. So a lot of GPU72 users successes get reported as p-1 factors while all failed attempts get TF credit. End up with mostly attempts and few successes.

bloodIce · 2015-01-21, 19:16

Quote:

Originally Posted by petrw1

Code:

Member Name        GHz-Days    Att    Succ    A/S    GhzDays Per  Att
BloodIce       3513.061    2714    415    6.5    1.29

I am doing some double-check in 989M-990M and reporting all factors found. 11% of all factors in the range are new. (second and third). There is no cheating from me and it will be over as of today (in couple of hours).

P.S.: The reason of this double-check is the issue with missed expos. I would like to see if it is a general problem. I was just too lazy to pick the unfactored only and scripted my worktodo.txt. Sorry, soon over, another 100 factors are coming.
P.S.S.: And what I can tell you is that the search of TJAOI is unfortunately non-systematic in respect to finished bit level. If TF55 is clear, then we should not see the factor 23897533126353583, as in http://www.mersenne.ca/exponent/989965729.

TheMawn · 2015-01-21, 19:57

How goes the work on redo.txt? I'd be happy to chip in although even 17000 jobs in the upper 60's is only a couple of days worth, so maybe it's easier to not step on anyone's toes?

TheMawn · 2015-01-21, 20:07

Out of curiosity I looked up the factors found so far, and found the factors of q - 1.

factor 534321209016231454889-1 : 2^3×7×70645457×135061057939 (6 prime factors, 4 distinct)

factor 481980406054172909441-1 : 2^7×5×29×41×8983511×70505249 (12 prime factors, 6 distinct)

factor 70655970542502575393-1 : 2^5×70664707×31246136483 (7 prime factors, 3 distinct)

I found exactly what I was looking for, too. The massive prime factor in the first and third mean that this factor would have been harder to find with P-1 whereas the second factor shouldn't have been all that difficult. Look at the exponent status. It was in fact discovered with P-1.

Mark Rose · 2015-01-21, 20:08

Quote:

Originally Posted by TheMawn

How goes the work on redo.txt? I'd be happy to chip in although even 17000 jobs in the upper 60's is only a couple of days worth, so maybe it's easier to not step on anyone's toes?

I've already queued all the work under 70M. I'll also have the work at 80M and above finished in under 3 hours.

There is about 7 THz-d of work from 70M up to 80M range no one has claimed in this thread, minus that short list of exponents posted a page back that I believe LaurV is working on.

The range 70.2 to 70.3 would be the most useful for the project at the moment.

Edit: Also, I've created a merged list of assignments that cut the list of assignments in half: https://gist.github.com/MarkRose/ebc25e1adc2a1668605b (raw).

TheMawn · 2015-01-21, 20:12

Does anyone know if this page is still being kept up to date?

This could be a useful system to catch the occasional user having trouble with TF.

2015-01-21, 17:34	#158
chalsall If I May "Chris Halsall" Sep 2002 Barbados 11374₁₀ Posts	GPU72's Worker's view... Further to the above, GPU72 also collects stats as to work assigned by it, comparing to expected vs actually found factors. A few examples: Chris Halsall Brian Long Axelsson Patrik Johansson

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Old User	Unregistered	Information & Answers	1	2012-10-18 23:31
The user CP has gone :(	retina	Forum Feedback	5	2006-12-05 16:47
Changing My User ID	endless mike	NFSNET Discussion	1	2004-10-31 19:38
OSX yet? new user here	KevinLee	Hardware	6	2003-12-12 17:06
help for a Mac user	drakkar67	Software	3	2003-02-11 10:55

2015-01-21, 17:21	#155
Mark Rose "/X\(‘-‘)/X\" Jan 2013 https://pedan.tech/ 2⁴·199 Posts	Given the size and scope of that, having PrimeNet or GPU72 distribute rechecks might be wise.

2015-01-21, 19:57	#162
TheMawn May 2013 East. Always East. 1727₁₀ Posts	How goes the work on redo.txt? I'd be happy to chip in although even 17000 jobs in the upper 60's is only a couple of days worth, so maybe it's easier to not step on anyone's toes?

2015-01-21, 20:07	#163
TheMawn May 2013 East. Always East. 11·157 Posts	Out of curiosity I looked up the factors found so far, and found the factors of q - 1. factor 534321209016231454889-1 : 2^3×7×70645457×135061057939 (6 prime factors, 4 distinct) factor 481980406054172909441-1 : 2^7×5×29×41×8983511×70505249 (12 prime factors, 6 distinct) factor 70655970542502575393-1 : 2^5×70664707×31246136483 (7 prime factors, 3 distinct) I found exactly what I was looking for, too. The massive prime factor in the first and third mean that this factor would have been harder to find with P-1 whereas the second factor shouldn't have been all that difficult. Look at the exponent status. It was in fact discovered with P-1.

2015-01-21, 20:12	#165
TheMawn May 2013 East. Always East. 11·157 Posts	Does anyone know if this page is still being kept up to date? This could be a useful system to catch the occasional user having trouble with TF.