Treading LL P-1 for 100M TF - Page 16

Aramis Wyler · 2013-05-06, 12:39

Ok. I queued 332432047 to start after UTC 0000. It's from 72-79 so I guess that will put me in the 415 range if there is no factor. I'll add them all up after it finishes.

Aramis Wyler · 2013-05-08, 01:04

Code:

Number    | Work  | Range | Date Reserved    | GHzD/D | Status
332309833 | LL TF | 73-79 | 2013-04-02 00:30 | 362.67 | No Factor
332310091 | LL TF | 73-79 | 2013-04-05 01:55 | 362.67 | No Factor
332311081 | LL TF | 73-79 | 2013-04-09 19:04 | 362.67 | No Factor
332283779 | LL TF | 73-79 | 2013-04-16 11:58 | 362.70 | No Factor
332283821 | LL TF | 73-79 | 2013-04-16 11:58 | 362.70 | No Factor
332283841 | LL TF | 73-79 | 2013-04-21 19:07 | 362.70 | No Factor
332283859 | LL TF | 73-79 | 2013-04-21 19:07 | 362.70 | No Factor
332229197 | LL TF | 76-79 | 2013-04-25 10:09 | 322.46 | No Factor
332207987 | LL TF | 76-79 | 2013-04-30 09:04 | 322.48 | No Factor
332208131 | LL TF | 76-79 | 2013-05-01 00:52 | 322.48 | No Factor
332229187 | LL TF | 76-79 | 2013-05-02 02:07 | 322.46 | No Factor
332432047 | LL TF | 72-79 | 2013-05-06 12:27 | 365.42 | No Factor

419.41. I'll put another one on tommorow.

Uncwilly · 2013-05-08, 01:20

Quote:

Originally Posted by Aramis Wyler

419.41. I'll put another one on tommorow.

When you fetch them from GPUto72, use 'Lowest Exponent' if you don't mind.

The lower they are, the more likely they are to get sent out for LL sooner.

Aramis Wyler · 2013-05-08, 01:29

Ah, whoops. You know, I pretty much always do. I set it to depth first, change the commitment from 78 to 79, then set it to lowest exponent. I must have skipped a step on that last one.

Just fetched 332216113, 76-79, 322.46 GHD. I was wondering how I had gotten one that wasn't factored past 72, but figured maybe someone had just relinquished it back to the system.

kracker · 2013-05-09, 15:17

I decided to take a few to 78, first exponent on the list and...

no factor for M332426287 from 2^72 to 2^73 [mfakto 0.13pre4-Win cl_barrett15_73_gs_2]

M332426287 has a factor: 12637801665226653007607 [TF:73:74*:mfakto 0.13pre4-Win cl_barrett15_82_gs_2]

found 1 factor for M332426287 from 2^73 to 2^74 (partially tested) [mfakto 0.13pre4-Win cl_barrett15_82_gs_2]

Aramis Wyler · 2013-05-09, 17:00

Lol! Damn, good job.
I don't know where we are, Uncwilly, with regard to how many 332M numbers get requested in a week vs how many we have factored up to whatever level, but I wonder about factoring these up to 79 at a time rather than going with a more breadth first approach. If only 3 of these a week are requested and we've already got 20 to 79, then by all means, taking these up to 79 is fine. But if 50 a week are requested, well, maybe we should work a block of 50 up a bit level at a time until we're ahead of the curve.

That said, this is your project, and I committed to running whatever numbers you want to whatever level you want. I'll run them up to 85 if you want, w/o complaint. I assume you have this in hand and am not actually questioning your methods, and I understand wanting to do the lower bitdepths on cpu and the higher ones on gpu. I simply don't know where the matter stands, and thought it might be time for a reality check since we ran 13 of these up to 79 w/o a factor.

Aramis Wyler · 2013-05-09, 17:39

Update: The first dozen are here. There were no factors. 419.41 GHD/10.

Code:

Number    | Work  | Range | Date Reserved    |  GHzD  | Status
332216113 | LL TF | 76-79 | 2013-05-08 01:26 | 322.47 | No Factor
332205773 | LL TF | 76-79 | 2013-05-09 17:16 | 322.47 | Testing

Uncwilly · 2013-05-10, 04:39

Quote:

Originally Posted by Aramis Wyler

I don't know where we are, Uncwilly, with regard to how many 332M numbers get requested in a week vs how many we have factored up to whatever level,

I don't know exactly. We do get a fair bit of churn, especially of late, since the expirys from the influx after the prime discovery. Since April 20, there has been a drop of assigned LL's in the 332's from 2408 to 2367 (at one point it was about 2600).

Sure, go after some breadth first. I'll take what ever you like.

GPUto72 has a few at 76, PrimeNet may have some too. Run them up one bit. If you run out of candidates, go from 77 to 78.
Or, you could do some work taking available numbers out to 332,999,999 from 72 to 73, there are ~10,500 of them, lots of factors to find.

It is all good.

Aramis Wyler · 2013-05-10, 14:33

Ok. From what I can tell from the mersenne visualization site, it looks like the 332.2M range alone can cover the entire churn plus the completion rate for some time, so I will restrict myself to that range.

Tracking my work gets very difficult the lower I go anyway. Sorting by exponent (lowest to highest) only shows my lowest 1000 assignments (<31M) and so I sort by highest GHD. If I get below 77 they mix in to the LL range numbers I TF'd to 64/65.

kracker · 2013-05-10, 15:50

Quote:

Originally Posted by Aramis Wyler

Tracking my work gets very difficult the lower I go anyway. Sorting by exponent (lowest to highest) only shows my lowest 1000 assignments (<31M) and so I sort by highest GHD. If I get below 77 they mix in to the LL range numbers I TF'd to 64/65.

http://mersenne.org/results/?exp1=1&...B1=Get+results

Aramis Wyler · 2013-05-10, 16:50

Quote:

Originally Posted by kracker

http://mersenne.org/results/?exp1=1&...B1=Get+results

Ah, you know, that's true. I could get the results from primenet itself rather than gpu72, even if I get the work from gpu72. Good call. :)

2013-05-09, 15:17	#170
kracker "Mr. Meeseeks" Jan 2012 California, USA 2³·271 Posts	I decided to take a few to 78, first exponent on the list and... no factor for M332426287 from 2^72 to 2^73 [mfakto 0.13pre4-Win cl_barrett15_73_gs_2] M332426287 has a factor: 12637801665226653007607 [TF:73:74:mfakto 0.13pre4-Win cl_barrett15_82_gs_2] found 1 factor for M332426287 from 2^73 to 2^74 (partially tested) [mfakto 0.13pre4-Win cl_barrett15_82_gs_2] Last fiddled with by kracker on 2013-05-09 at 15:17*

2013-05-09, 17:39	#172
Aramis Wyler "Bill Staffen" Jan 2013 Pittsburgh, PA, USA 2³×53 Posts	Update: The first dozen are here. There were no factors. 419.41 GHD/10. Code: Number \| Work \| Range \| Date Reserved \| GHzD \| Status 332216113 \| LL TF \| 76-79 \| 2013-05-08 01:26 \| 322.47 \| No Factor 332205773 \| LL TF \| 76-79 \| 2013-05-09 17:16 \| 322.47 \| Testing

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
To 100M to 75 Bits	petrw1	GPU to 72	6	2017-04-17 04:03
Is LMH > 100M still alive?	TheMawn	LMH > 100M	25	2016-04-18 16:11
100m p-1 and tf	aurashift	Software	18	2016-04-14 13:48
Anyone working in 79.3M - 100M ??	markr	Lone Mersenne Hunters	21	2008-12-21 16:02
100M madness	stars10250	Hardware	8	2008-10-02 15:21

2013-05-06, 12:39	#166
Aramis Wyler "Bill Staffen" Jan 2013 Pittsburgh, PA, USA 650₈ Posts	Ok. I queued 332432047 to start after UTC 0000. It's from 72-79 so I guess that will put me in the 415 range if there is no factor. I'll add them all up after it finishes.

2013-05-08, 01:29	#169
Aramis Wyler "Bill Staffen" Jan 2013 Pittsburgh, PA, USA 2³×53 Posts	Ah, whoops. You know, I pretty much always do. I set it to depth first, change the commitment from 78 to 79, then set it to lowest exponent. I must have skipped a step on that last one. Just fetched 332216113, 76-79, 322.46 GHD. I was wondering how I had gotten one that wasn't factored past 72, but figured maybe someone had just relinquished it back to the system.

2013-05-09, 17:00	#171
Aramis Wyler "Bill Staffen" Jan 2013 Pittsburgh, PA, USA 2³×53 Posts	Lol! Damn, good job. I don't know where we are, Uncwilly, with regard to how many 332M numbers get requested in a week vs how many we have factored up to whatever level, but I wonder about factoring these up to 79 at a time rather than going with a more breadth first approach. If only 3 of these a week are requested and we've already got 20 to 79, then by all means, taking these up to 79 is fine. But if 50 a week are requested, well, maybe we should work a block of 50 up a bit level at a time until we're ahead of the curve. That said, this is your project, and I committed to running whatever numbers you want to whatever level you want. I'll run them up to 85 if you want, w/o complaint. I assume you have this in hand and am not actually questioning your methods, and I understand wanting to do the lower bitdepths on cpu and the higher ones on gpu. I simply don't know where the matter stands, and thought it might be time for a reality check since we ran 13 of these up to 79 w/o a factor.

2013-05-10, 14:33	#174
Aramis Wyler "Bill Staffen" Jan 2013 Pittsburgh, PA, USA 2³×53 Posts	Ok. From what I can tell from the mersenne visualization site, it looks like the 332.2M range alone can cover the entire churn plus the completion rate for some time, so I will restrict myself to that range. Tracking my work gets very difficult the lower I go anyway. Sorting by exponent (lowest to highest) only shows my lowest 1000 assignments (<31M) and so I sort by highest GHD. If I get below 77 they mix in to the LL range numbers I TF'd to 64/65.