A couple of largish numbers

fivemack · 2011-03-24, 15:21

I have been working on various sequences below 10^4 for a couple of years now.

3270.632 is a C170 on which I've done 16000 curves at 3e7
3678.3076 is a C162 on which I've done 20000 curves at 3e7

I have the resources to do these GNFS jobs, but not really the interest; if anyone else wants to pick up the sequences, they're welcome to them

debrouxl · 2011-03-24, 18:36

A C162 is within the reach of RSALS.
But someone else would need to perform the polynomial selection: my 8 cores, on 3 computers, are currently busy with preparatory ECM work @ B1=43e6 and polynomial selection for RSALS, and some work for another BOINC-based project

Syd · 2011-03-24, 21:33

I could do the poly selection for 3678.3076, would take me a few days to get a good one.

edit:
I'll just start it. 5 cores without cuda, one with GTX260. I'll post results here later.

Syd · 2011-03-25, 09:36

so far: found an outstanding poly,

Code:

# norm 1.253170e-15 alpha -8.532833 e 1.271e-12 rroots 5
skew: 133435193.68
c0:  9278864245525985137550344687245446069980225
c1:  180109282635950772896812489720834585
c2: -8995871353090924335725511063
c3: -17784552123636433957
c4:  552333312806
c5:  540
Y0: -45655070065487497189556645317254
Y1:  180378937982427677

second best is 1.063e-12

debrouxl · 2011-03-28, 04:34

I'll queue this polynomial onto RSALS in the next few hours.

debrouxl · 2011-03-28, 17:24

I have queued to RSALS the C162 from sequence 3678, with Syd's poly, over the 15M-55M range for now. There's room for a team sieving here

debrouxl · 2011-03-30, 06:20

Raising the upper limit to 65M, one quarter of the 15M-65M range returned so far.
Forecast number of raw relations (over-estimated, due to simple linear projection): 81M.
Could someone give an estimation of how many raw relations are needed ?

debrouxl · 2011-03-31, 17:24

I have raised the upper limit to 80M, and forced the generation of all WUs. ~54% of the 15-80M range has been returned so far, yielding a projection of ~113M raw relations.

debrouxl · 2011-04-02, 10:31

~83% of the results returned so far, ~115M raw relations expected. WUs near q=15M give lower yields than WUs at e.g. q=40M, which is infrequent.

We're currently seeing a low point in sieving activity on the grid, so I won't be participating in further team sievings until the activity raises.

Syd · 2011-04-02, 11:05

Quote:

Originally Posted by debrouxl

~83% of the results returned so far, ~115M raw relations expected. WUs near q=15M give lower yields than WUs at e.g. q=40M, which is infrequent.

We're currently seeing a low point in sieving activity on the grid, so I won't be participating in further team sievings until the activity raises.

Maybe you can increase that activity by granting more credits to your users. Currently you grant less than average per CPU-hour, not taking the memory usage into account.

http://allprojectstats.com/cpuinfo.php?id=47029

xilman · 2011-04-02, 17:55

Quote:

Originally Posted by Syd

Maybe you can increase that activity by granting more credits to your users. Currently you grant less than average per CPU-hour, not taking the memory usage into account.

http://allprojectstats.com/cpuinfo.php?id=47029

Sounds like an invitation to an inflationary arms race to me...

Paul

2011-03-24, 15:21	#1
fivemack (loop (#_fork)) Feb 2006 Cambridge, England 1936₁₆ Posts	A couple of largish numbers I have been working on various sequences below 10^4 for a couple of years now. 3270.632 is a C170 on which I've done 16000 curves at 3e7 3678.3076 is a C162 on which I've done 20000 curves at 3e7 I have the resources to do these GNFS jobs, but not really the interest; if anyone else wants to pick up the sequences, they're welcome to them

2011-03-24, 21:33	#3
Syd Sep 2008 Krefeld, Germany 2·5·23 Posts	I could do the poly selection for 3678.3076, would take me a few days to get a good one. edit: I'll just start it. 5 cores without cuda, one with GTX260. I'll post results here later. Last fiddled with by Syd on 2011-03-24 at 21:49

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
A couple of 15e candidates	fivemack	NFS@Home	1	2014-11-30 07:52
new here with a couple questions	theshark	Information & Answers	21	2014-08-30 17:36
A couple questions from a new guy	Optics	Information & Answers	8	2009-04-25 18:23
Odds a largish number has N divisors?	grandpascorpion	Math	65	2006-02-16 15:20
A couple things	PHinker	Software	3	2004-12-18 17:08

2011-03-24, 18:36	#2
debrouxl Sep 2009 2²×5×7² Posts	A C162 is within the reach of RSALS. But someone else would need to perform the polynomial selection: my 8 cores, on 3 computers, are currently busy with preparatory ECM work @ B1=43e6 and polynomial selection for RSALS, and some work for another BOINC-based project

2011-03-28, 04:34	#5
debrouxl Sep 2009 2²·5·7² Posts	I'll queue this polynomial onto RSALS in the next few hours.

2011-03-28, 17:24	#6
debrouxl Sep 2009 1724₈ Posts	I have queued to RSALS the C162 from sequence 3678, with Syd's poly, over the 15M-55M range for now. There's room for a team sieving here

2011-03-30, 06:20	#7
debrouxl Sep 2009 1111010100₂ Posts	Raising the upper limit to 65M, one quarter of the 15M-65M range returned so far. Forecast number of raw relations (over-estimated, due to simple linear projection): 81M. Could someone give an estimation of how many raw relations are needed ?

2011-03-31, 17:24	#8
debrouxl Sep 2009 2²·5·7² Posts	I have raised the upper limit to 80M, and forced the generation of all WUs. ~54% of the 15-80M range has been returned so far, yielding a projection of ~113M raw relations.

2011-04-02, 10:31	#9
debrouxl Sep 2009 2²·5·7² Posts	~83% of the results returned so far, ~115M raw relations expected. WUs near q=15M give lower yields than WUs at e.g. q=40M, which is infrequent. We're currently seeing a low point in sieving activity on the grid, so I won't be participating in further team sievings until the activity raises.