Trial Factor and P-1 assignments squatting in Cat 0 and Cat 1 ranges - Page 3

ATH · 2020-09-30, 06:42

Yeah, it does not sound like they are interested in changing the assignment code, so if we get TF assignments in those categories that are not completed for weeks, we will have to poach them in an organized way.

Ensigm · 2020-09-30, 20:35

Quote:

Originally Posted by Viliam Furik

Yeah, uhm, sorry for that, but I have checked that exponents which were assigned to me for TF are not assigned for DC yet. Plus expected throughput is about 70 expos per day.

Also a tip that for TF of a certain bit level the larger the exponent the shorter it takes. That's because there are less \(k\)s to try. So it's probably wiser to TF the largest exponent available for a certain bit level in itself, in addition to the argument of not blocking DC progress. Though I don't know much about factor possibility so I'm not really sure about the efficiency.

I once reserved several assignments in the 53M range too, but I noticed that it had higher GHz*d credits per exponent (and did took longer) so I thought about why. As a result now I mostly do 60M's and 59M's, because I care about factors not credits.

Uncwilly · 2020-09-30, 21:05

All of the exponents look like they are at 74 bits already (I did a spot check of about a dozen). That is 1 bit above the recommended level as seen in yellow here: https://www.mersenne.ca/status/tf/0/0/4/5300

It looks like everything below 100M is at the goal bit depth or above it. Maybe any TF in the DC should be aimed at those actually at the goal bit depth.

LaurV · 2020-10-01, 02:49

Quote:

Originally Posted by retina

We're all in it for the life-changing prize money. You mean you aren't?

Well... hmm... arrr.. don't tell these guys!

S485122 · 2020-10-01, 09:54

I went to look at the Manual assignment page the lowest exponent I could get for TF wath in the 200M range : no problem there.

However on the "Manual GPU assignment page, if one select TF for double check the "optional" range starts at the lowest unverified exponent. The same applies for TF for fist time tests...

It would be easy to modify the code to use the current CAT 2 or the CAT 3 threshold instead of the lowest unverified exponent.

Jacob

Aramis Wyler · 2020-10-01, 12:12

This still seems like mountains from molehills to me. There are lots of numbers inthat range, and taking 3 days to TF a couple dozen isn't hurting anything or slowing the project.

At worst, set the expiration to 5-7 days for cat 0 numbers and call it good.

It looks like Villiam has pulled a new batch of numbers in that range yesterday (having returned the initial ones when complete) and so it's moving on.

Uncwilly · 2020-10-01, 12:37

But there have been some users holding TF assignments for months in that range. As noted elsewhere, there was a user in the 93M range with assignments many months old, with no progress.

S485122 · 2020-10-01, 13:36

Quote:

Originally Posted by Aramis Wyler

...
At worst, set the expiration to 5-7 days for cat 0 numbers and call it good.
...

Problem is "we" can't do that.
And at the moment those manual TF assignments are not subject to recycling rules, they have no expiry date.
It would be easier not to assign CAT0 and CAT for manual testing as stated by the Assignment rules.
At the moment CAT2 starts at 54571632 in the DC range and at 105751532 for the first time checks.
This means that unverified numbers the ranges 53xxxxxx to 54571632, and untested numbers in the range 91000000 to 105751532 should not be handled out for manual testing, be it ECM or TF.

Jacob

(I'd say it is not as if there were not enough numbers to work on, to absolutely go for those numbers that hold up the milestones, and are more than sufficiently factored as well. Especially it it makes a number of people nervous and worried ;-)

chalsall · 2020-10-01, 15:29

Quote:

Originally Posted by S485122

And at the moment those manual TF assignments are not subject to recycling rules, they have no expiry date.

Technically not true. They are recycled after 180 days.

S485122 · 2020-10-01, 18:01

Quote:

Originally Posted by chalsall

Technically not true. They are recycled after 180 days.

OK.

I thought I had seen exponents more than a year late in being reported about, but a quick search didn't find any in the relevant ranges... It even seems that the "first" rule "Since PrimeNet began, the server has recycled exponents where the client computer is 60 days past due in updating the server." is enforced in those ranges.

Jacob

Uncwilly · 2020-10-09, 17:59

It is happening again.

https://www.mersenneforum.org/showthread.php?p=559340

2020-10-01, 09:54	#27
S485122 Sep 2006 Brussels, Belgium 3157₈ Posts	manual assignment pages I went to look at the Manual assignment page the lowest exponent I could get for TF wath in the 200M range : no problem there. However on the "Manual GPU assignment page, if one select TF for double check the "optional" range starts at the lowest unverified exponent. The same applies for TF for fist time tests... It would be easy to modify the code to use the current CAT 2 or the CAT 3 threshold instead of the lowest unverified exponent. Jacob

2020-10-01, 12:12	#28
Aramis Wyler "Bill Staffen" Jan 2013 Pittsburgh, PA, USA 3·137 Posts	This still seems like mountains from molehills to me. There are lots of numbers inthat range, and taking 3 days to TF a couple dozen isn't hurting anything or slowing the project. At worst, set the expiration to 5-7 days for cat 0 numbers and call it good. It looks like Villiam has pulled a new batch of numbers in that range yesterday (having returned the initial ones when complete) and so it's moving on. Last fiddled with by Aramis Wyler on 2020-10-01 at 12:12

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
possible overlapping Fermat factor ranges	MattcAnderson	FermatSearch	3	2021-01-05 14:34
Trial Factoring - Factor Confirmation?	butera	GPU to 72	3	2021-01-02 23:43
Trial Factor Bit Depth	lavalamp	Operation Billion Digits	8	2010-08-02 18:49
trial division over a factor base	Peter Hackman	Factoring	7	2009-10-26 18:27
Shortest time to complete a 2^67 trial factor (no factor)	dsouza123	Software	12	2003-08-21 18:38

2020-09-30, 06:42	#23
ATH Einyen Dec 2003 Denmark 3²×5×67 Posts	Yeah, it does not sound like they are interested in changing the assignment code, so if we get TF assignments in those categories that are not completed for weeks, we will have to poach them in an organized way.

2020-09-30, 21:05	#25
Uncwilly 6809 > 6502 """"""""""""""""""" Aug 2003 101×103 Posts 2²·7²·47 Posts	All of the exponents look like they are at 74 bits already (I did a spot check of about a dozen). That is 1 bit above the recommended level as seen in yellow here: https://www.mersenne.ca/status/tf/0/0/4/5300 It looks like everything below 100M is at the goal bit depth or above it. Maybe any TF in the DC should be aimed at those actually at the goal bit depth.

2020-10-01, 12:37	#29
Uncwilly 6809 > 6502 """"""""""""""""""" Aug 2003 101×103 Posts 2²·7²·47 Posts	But there have been some users holding TF assignments for months in that range. As noted elsewhere, there was a user in the 93M range with assignments many months old, with no progress.

2020-10-09, 17:59	#33
Uncwilly 6809 > 6502 """"""""""""""""""" Aug 2003 101×103 Posts 2²·7²·47 Posts	It is happening again. https://www.mersenneforum.org/showthread.php?p=559340